REPOSITORY_FILE_CLASSIFICATION.csv

"current_path","classification","original_path","destination_applied","in_repo_now","active_consumer","referenced_by","requires_local_data","manuscript_relevant","safe_to_move","notes"
".gitignore","production",".gitignore","KEEP (not moved)",TRUE,"git itself, on every status/add in this working tree.","n/a (consumed by git)","no","indirect","no","12 lines, 2 active rules. EXACT LINES THAT MUST BE UPDATED IF THE SCHEMA FOLDER MOVES: line 12 'manuscript/BehavioralDynamics_schema/rendered/' (the active ignore rule - becomes dead and the moved rendered/ files start showing as untracked), plus the explanatory comments at line 3 '# manuscript/BehavioralDynamics_schema/R/), so the builders are versioned but' and line 10 '# manuscript/BehavioralDynamics_schema/R/ from version control entirely.'. Lines 2, 8 and 11 reference manuscript/Fig1_behavior_candidates/ and need NO change under the proposed plan since that folder stays. The header comment records that these two narrow rules replaced a blanket /manuscript rule which had been hiding tracked source - so do NOT re-broaden this file during restructuring."
"Analysis/_archive/03b_multiscale_metric_validation_notes.md","legacy","Analysis/_archive/03b_multiscale_metric_validation_notes.md","KEEP (not moved)",TRUE,"none","Analysis/_archive/README_archive.md","no","no","no","Development note for metric validation; not an executable stage. Contains no S: data paths. Not churned in this pass."
"Analysis/_archive/04_gamm_movement_proximity_phase_and_early_window.R","legacy","Analysis/_archive/04_gamm_movement_proximity_phase_and_early_window.R","KEEP (not moved)",TRUE,"none - superseded by Analysis/07_gamm_trajectory_features.R","Analysis/14_systems_neuroscience_summary_dashboard.R (comment/provenance mention); Analysis/README_pipeline.md (old-to-new filename map); Analysis/_archive/README_archive.md","yes","no","no","Large exploratory/superseded GAMM/LME layer. Still holds 4 hard-coded S:/Lab_Member data paths, so it cannot be run reproducibly off-site. Provenance only; not churned in this pass."
"Analysis/_archive/08_early_prediction_models.R","legacy","Analysis/_archive/08_early_prediction_models.R","KEEP (not moved)",TRUE,"none - superseded by Analysis/09_early_prediction_model_ladder.R","Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/15_behavior_proteomics_integration.R; Analysis/README_pipeline.md; Analysis/_supporting/13_nonlinear_systems_dynamics.R; Analysis/_archive/README_archive.md; Testing/audit_first_night_domain_scores_v2.R; Testing/audit_first_night_time_anchor.R; Testing/audit_stage09_stale_artifacts.R; Testing/check_behavioral_dynamics_structure.R; Testing/run_behavioral_dynamics_pipeline.R; Testing/run_full_systems_behavior_pipeline.R","yes","no","no","The most-referenced archive file, and it carries a pre-existing STALE-PATH problem worth flagging to the Testing/ audit (not for fixing here): Testing/run_behavioral_dynamics_pipeline.R line 23 and Testing/run_full_systems_behavior_pipeline.R line 5 still list the pre-archive path ""Analysis/08_early_prediction_models.R"", which no longer exists - the file is now at Analysis/_archive/08_early_prediction_models.R. Testing/check_behavioral_dynamics_structure.R line 9 has the same stale path in a comment. It is also the only archive script that sources _pipeline_setup.R. Holds 3 hard-coded S: paths. Provenance only; not churned in this pass."
"Analysis/_supporting/13_nonlinear_systems_dynamics.R","production","Analysis/_supporting/13_nonlinear_systems_dynamics.R","Analysis/_supporting/ (moved out of _archive)",TRUE,"none","Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/README_pipeline.md; Analysis/_archive/README_archive.md","yes","no","no","TYPE B artifact dependency: not source()d anywhere, but active Stages 10 and 14 read its output tree and run_all_analysis.R does not regenerate it. Moved out of _archive so live-dependency code is not described as archived. See docs/ARCHIVE_DEPENDENCY_AUDIT.csv. || Exploratory nonlinear dynamics layer, never part of the staged active pipeline. README_pipeline.md states nonlinear outputs remain exploratory and are not promoted to the primary registry. 1 hard-coded S: path. Not churned in this pass."
"Analysis/_supporting/14_nextgen_behavioral_phenotyping.R","production","Analysis/_supporting/14_nextgen_behavioral_phenotyping.R","Analysis/_supporting/ (moved out of _archive)",TRUE,"none","Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/README_pipeline.md; Analysis/_archive/README_archive.md","yes","no","no","TYPE B artifact dependency: not source()d anywhere, but active Stages 10 and 14 read its output tree and run_all_analysis.R does not regenerate it. Moved out of _archive so live-dependency code is not described as archived. See docs/ARCHIVE_DEPENDENCY_AUDIT.csv. || [VERIFIED-CORRECTION] REFUTES ""active_consumer: none"" (data-artifact + declared-dependency consumers, both in ACTIVE scripts):

1) SOLE PRODUCER. Analysis/_supporting/14_nextgen_behavioral_phenotyping.R:42 `output_dir <- file.path(project_root, ""analysis_ready/14_nextgen_behavioral_phenotyping"", bin_level)` — `git grep -n 'analysis_ready/14_nextgen' -- '*.R'` returns exactly one writer: this archived file. No active script regenerates these outputs.

2) ACTIVE CONSUMER #1 — dashboard reads 5 of its tables and binds them into the feature matrix. Analysis/14_systems_neuroscience_summary_dashboard.R:1290 defines `load_nextgen_selective_features <- function(scale_label = primary_bin_level)`; :1291 `ng_dir <- file.path(project_root, ""analysis_ready/14_nextgen_behavioral_phenotyping"", scale_label, ""tables"")`; :1300-1306 reads multiscale_complexity_features.csv, early_warning_summary_by_animal.csv, behavioral_energy_landscape_summary_by_animal.csv, animal_social_coupling_summary.csv, nextgen_behavioral_phenotype_matrix.csv. It is NOT dead code — it is invoked at :1387 `map_dfr(optional_import_bin_levels, load_nextgen_selective_features)` inside the `optional_long` bind_rows that feeds `optional_wide` (:1389-1394).

3) ACTIVE CONSUMER #2 — prediction ladder harvests its outputs as candidate predictors. Analysis/10_systems_feature_prediction_ladder.R:101 `file.path(base_dir, ""analysis_ready/14_nextgen_behavioral_phenotyping"")` inside `feature_search_dirs` (:91), consumed at :534-536 `feature_search_dirs[dir.exists(feature_search_dirs)] %>% map(~list.files(.x, pattern = ""\\.(csv|tsv|xlsx|xls)$"", recursive = TRUE, ...))` -> :542 `candidate_feature_tables` -> :544 `systems_extra`. So this ""legacy"" script's features can enter an active model.

4) DECLARED REQUIRED SCRIPT. Analysis/14_systems_neuroscience_summary_dashboard.R:3641 lists `""14_nextgen_behavioral_phenotyping.R""` in `sis_dependency_audit$RequiredScript` (tibble opens :3635), tagged BiologicalDomain ""Nonlinear systems organization"" (:3648), with ExpectedPath at :3659. Also enumerated as a required stage at :1935 with expected_output :1957.

5) DEPENDENCY IS LIVE, NOT DORMANT. Artifacts exist on the data root: S:/Lab_Member/Tobi/Experiments/Exp9_Social-Stress/Analysis/Behavior/RFID/analysis_ready/14_nextgen_behavioral_phenotyping/5min_based/tables/ contains nextgen_behavioral_phenotype_matrix.csv, multiscale_complexity_features.csv, early_warning_summary_by_animal.csv, behavioral_energy_landscape_summary_by_animal.csv, animal_social_coupling_summary.csv (all 5 files load_nextgen_selective_features requests).

DESTINATION UNCHANGED (KEEP) but for a STRONGER reason than ""not churned in this pass"": the bare filename is hard-coded as a manuscript-facing provenance/dependency string in the active dashboard at :1645 `str_detect(source, ""nextgen|nonlinear"") ~ ""14_nextgen_behavioral_phenotyping.R""`, :1935, and :3641. A rename would desync the dashboard's provenance labels and its own dependency audit. Note also the audit's notes field understates the hard-coding: there are TWO hard-coded absolute paths, not one — :37 `project_root <- ""S:/...""` AND :38 `repo_root <- ""C:/Users/topohl/Documents/GitHub/MMMSociability""`, the latter used at :67-68 to source Functions/behavioral_dynamics_helpers.R and Functions/behavioral_dynamics_stats_helpers.R.

NOT refuted: no active script `source()`s this file (checked `git grep -nE 'source *\(' -- Analysis/14_systems_neuroscience_summary_dashboard.R Analysis/10_systems_feature_prediction_ladder.R` -> only _pipeline_setup.R at :58/:56). The coupling is via data artifacts on S: plus declared filename strings, which is why KEEP remains correct. || Exploratory next-generation phenotyping layer, never part of the staged active pipeline. 1 hard-coded S: path. Not churned in this pass."
"Analysis/_archive/18_raw_movement_publication_trajectory.R","legacy","Analysis/_archive/18_raw_movement_publication_trajectory.R","KEEP (not moved)",TRUE,"none - superseded by Analysis/03_primary_raw_movement_phase_stats.R","Analysis/README_pipeline.md; Analysis/_archive/README_archive.md","yes","no","no","Superseded raw-movement trajectory script; README_pipeline.md explicitly states ""The older 18_raw_movement_publication_trajectory.R and 18b_raw_movement_broad_phase_stats.R are archived."" 3 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/18b_raw_movement_broad_phase_stats.R","legacy","Analysis/_archive/18b_raw_movement_broad_phase_stats.R","KEEP (not moved)",TRUE,"none - superseded by Analysis/03_primary_raw_movement_phase_stats.R","Analysis/README_pipeline.md; Analysis/_archive/README_archive.md","yes","no","no","Superseded by the corrected raw-movement statistics in Stage 03 (renamed from 18c_raw_movement_broad_phase_stats_corrected.R per the old-to-new map). 1 hard-coded S: path. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.0.r","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.0.r","KEEP (not moved)",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md; Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.1.r","yes","no","no","Versioned deprecated script. Filename contains a SPACE (""...shannon v.1.0.0.r"") and a lowercase .r extension - any scripted move or reference-rewrite must quote it. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.1.r","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.1.r","KEEP (not moved)",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md","yes","no","no","Versioned deprecated script; space in filename, lowercase .r extension. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon.R","KEEP (not moved)",TRUE,"none","none","yes","no","no","Deprecated Shannon-entropy variant of the AnimalPos analysis. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing.R","KEEP (not moved)",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md","yes","no","no","Deprecated pre-pipeline AnimalPos analysis script. 2 hard-coded S: paths. Covered by the README_archive.md wildcard row ""legacy_deprecated/E9_SIS_AnimalPos*"". Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-comparing.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-comparing.R","KEEP (not moved)",TRUE,"none","none","yes","no","no","Deprecated AnimalPos comparison script. 1 hard-coded S: path. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-comparing2.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-comparing2.R","KEEP (not moved)",TRUE,"none","none","yes","no","no","Deprecated AnimalPos comparison script, second variant. 1 hard-coded S: path. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-lme v.2.0.0.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-lme v.2.0.0.R","KEEP (not moved)",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md; Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R","yes","no","no","Versioned deprecated LME script; space in filename. 1 hard-coded S: path. Note a near-duplicate legacy file also exists outside Analysis/ at Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R - that one is in the Testing/ audit's scope, not this pass. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-lme.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-lme.R","KEEP (not moved)",TRUE,"none","none","yes","no","no","Deprecated LME script. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-plots_and_statistics.R","legacy","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-plots_and_statistics.R","KEEP (not moved)",TRUE,"none","none","yes","no","no","Deprecated plotting/statistics script. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/lme v.3.1.1.r","legacy","Analysis/_archive/legacy_deprecated/lme v.3.1.1.r","KEEP (not moved)",TRUE,"none","none","yes","no","no","Versioned deprecated LME script; space in filename, lowercase .r. Successor copies live outside Analysis/ (Testing/lme v.3.1.2.r) and are out of scope for this pass. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/sociability_lme v.1.0.0.r","legacy","Analysis/_archive/legacy_deprecated/sociability_lme v.1.0.0.r","KEEP (not moved)",TRUE,"none","Analysis/_archive/README_archive.md","yes","no","no","Versioned deprecated sociability LME script; space in filename, lowercase .r. 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/legacy_deprecated/sociability_lme v.1.0.1.r","legacy","Analysis/_archive/legacy_deprecated/sociability_lme v.1.0.1.r","KEEP (not moved)",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md; Analysis/_archive/README_archive.md","yes","no","no","Versioned deprecated sociability LME script; space in filename, lowercase .r. A later copy exists at Testing/sociability_lme v.1.0.2.r (out of scope here). 2 hard-coded S: paths. Not churned in this pass."
"Analysis/_archive/README_archive.md","legacy","Analysis/_archive/README_archive.md","KEEP (not moved)",TRUE,"none","none","no","no","no","Index of the archive with a per-file reason column, plus the explicit statement ""Archived scripts are not part of run_all_analysis.R. They may contain older hard-coded paths or output conventions and should be used only for provenance checks."" Confirms the archive is provenance-only. Not churned in this pass."
"Analysis/_pipeline_setup.R","production","Analysis/_pipeline_setup.R","KEEP (not moved)",TRUE,"Sourced by every active stage 00-16 (Analysis/00_qc_tracking_integrity.R ... Analysis/16_manuscript_behavior_report.R), Analysis/run_all_analysis.R, Analysis/run_cookiehab_preprocessing_and_metrics.R, Formatting/01_preprocess_cookiehab_animalpos.R, Functions/duration_normalization_helpers.R, Functions/hmm_stage14_helpers.R, ~50 Testing/ scripts, and manuscript/Fig1_behavior_candidates/build_fig1_candidates.R","Analysis/00_qc_tracking_integrity.R; Analysis/01_build_multiscale_behavior_metrics.R; Analysis/02_build_dyadic_rfid_contacts.R; Analysis/03_primary_raw_movement_phase_stats.R; Analysis/04_temporal_instability.R; Analysis/05_behavioral_state_space.R; Analysis/06_dynamic_social_networks.R; Analysis/07_gamm_trajectory_features.R; Analysis/08_hmm_behavioral_states_optional.R; Analysis/09_early_prediction_model_ladder.R; Analysis/10_systems_feature_prediction_ladder.R; Analysis/11_behavioral_adaptation_kinetics.R; Analysis/12_sleep_like_quiescence_metrics.R; Analysis/13_ethological_phase_organization.R; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/15_behavior_proteomics_integration.R; Analysis/16_manuscript_behavior_report.R; Analysis/run_all_analysis.R; Analysis/run_cookiehab_preprocessing_and_metrics.R; Analysis/_archive/08_early_prediction_models.R; Formatting/01_preprocess_cookiehab_animalpos.R; Functions/duration_normalization_helpers.R; Functions/hmm_stage14_helpers.R; README.md; ~50 Testing/*.R audits and tests; manuscript/Fig1_behavior_candidates/build_fig1_candidates.R","no","indirect","no","Highest-fanout file in the repo (75 in-repo path references to the literal string ""Analysis/_pipeline_setup.R""). Defines find_mmm_repo_root() (walks up for Functions/behavioral_dynamics_helpers.R + Analysis/), MMM_REPO_ROOT, MMM_ANALYSIS_DIR, source_mmm_helper(), the behavior_stage_dir()/behavior_stage_tables()/behavior_stage_figures()/behavior_stage_audit()/behavior_manuscript_dir() canonical-path family, resolve_behavior_artifact(), and the Stage 14 upstream registry (.resolve_stage_artifact_across_resolutions, resolve_stage04_temporal_instability_artifact, resolve_stage09_early_prediction_artifact). Reads no data itself. Must not move: dozens of consumers hard-code the literal relative path ""Analysis/_pipeline_setup.R""."
"Analysis/00_qc_tracking_integrity.R","production","Analysis/00_qc_tracking_integrity.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""00"", unconditional (optional_flag = NA)","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","Non-destructive RFID/tracking integrity QC. Reads the S: root: line 46 ""S:/Lab_Member/Tobi/Experiments/Exp9_Social-Stress/Analysis/Behavior/RFID"". Resolves setup via the standard three-candidate vector (getwd()/Analysis/_pipeline_setup.R, getwd()/_pipeline_setup.R, dirname(ofile)/_pipeline_setup.R) at lines 103-105, so it is depth-robust when invoked from the repo root. Stage 16 consumes its QC tables."
"Analysis/01_build_multiscale_behavior_metrics.R","production","Analysis/01_build_multiscale_behavior_metrics.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""01"", unconditional; also sourced directly by Analysis/run_cookiehab_preprocessing_and_metrics.R","Analysis/run_all_analysis.R; Analysis/run_cookiehab_preprocessing_and_metrics.R; Analysis/README_pipeline.md; Analysis/04_temporal_instability.R; Analysis/05_behavioral_state_space.R; Analysis/07_gamm_trajectory_features.R; Analysis/08_hmm_behavioral_states_optional.R; Analysis/09_early_prediction_model_ladder.R; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; Analysis/19_spatial_occupancy_maps.R; Analysis/_archive/03b_multiscale_metric_validation_notes.md; Analysis/_archive/08_early_prediction_models.R; Analysis/_supporting/13_nonlinear_systems_dynamics.R; Analysis/_supporting/14_nextgen_behavioral_phenotyping.R; Functions/animal_identity_invariants_helpers.R; README.md; Testing/audit_first_night_candidate_set_scores.R; Testing/audit_first_night_domain_scores_v2.R; Testing/repair_existing_metrics_identity_utility.R; Testing/test_animalpos_preprocessing_helpers.R; Testing/validate_cross_scale_animal_identity.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R; manuscript/BehavioralDynamics_schema/R/30_write_number_manifest.R; manuscript/BehavioralDynamics_schema/data/figure_number_manifest.csv","yes","yes","yes_with_reference_updates","The canonical upstream of the whole pipeline: writes all_behavior_metrics.csv at multiple bin levels and the canonical Stage 01 animal roster that Stage 08 inherits Group/Sex from. Reads/writes the S: root at lines 68-69 and reads S:/.../Analysis/sus_animals.csv and con_animals.csv (lines 79-80). Note line 68 contains "".../RFID/MMMSociability/preprocessed_data"" - that ""MMMSociability"" is a DATA-path segment, not a repo-identity reference, and must never be renamed. Also carries a dataset_id == ""cookiehab"" branch (add_cookiehab_bin_annotations, export_cookiehab_primary_window_metrics) used by the cookiehab runner. Cited directly in manuscript/BehavioralDynamics_schema provenance and figure_number_manifest.csv."
"Analysis/02_build_dyadic_rfid_contacts.R","production","Analysis/02_build_dyadic_rfid_contacts.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""02"", unconditional; also sourced directly by Analysis/run_cookiehab_preprocessing_and_metrics.R","Analysis/run_all_analysis.R; Analysis/run_cookiehab_preprocessing_and_metrics.R; Analysis/README_pipeline.md; Analysis/06_dynamic_social_networks.R; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; Analysis/_archive/03b_multiscale_metric_validation_notes.md; README.md; Testing/test_downstream_boundary_and_gap_contract.R; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","Builds the dyadic contact tables and dyadic_network_ready.csv consumed by Stage 06. S: paths at lines 57-58 (preprocessed_data in; analysis_ready/06_behavioral_dynamics/dyadic_contacts out). Line 57 contains the data-path ""MMMSociability"" segment - do not rename. Has cookiehab annotation-column handling."
"Analysis/03_primary_raw_movement_phase_stats.R","production","Analysis/03_primary_raw_movement_phase_stats.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""03"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/16_manuscript_behavior_report.R; Analysis/_archive/README_archive.md; Functions/identity_correction_comparison_helpers.R; README.md; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","Active secondary phenotype/group-characterization script for broad raw movement; supersedes the archived 18_ and 18b_ scripts. Reads the S: RFID root (line 33) and the endpoint workbook S:/.../Analysis/SIS_Analysis/E9_Behavior_Data.xlsx (line 39). Its canonical migrated output root is analysis_ready/pipeline/03_movement_phase_stats/10min/. Stage 16 reads selected Stage 03 tables into the manuscript package."
"Analysis/04_temporal_instability.R","production","Analysis/04_temporal_instability.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""04"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","Temporal instability / burstiness features. Reads S:/.../RFID/analysis_ready/03_derived_metrics (line 50) and writes analysis_ready/06_behavioral_dynamics/temporal_instability (line 51). Deliberately NOT migrated to the analysis_ready/pipeline/ layout - _pipeline_setup.R's resolve_stage04_temporal_instability_artifact() documents that this is its one true output location and that there is no legacy fallback. Stage 14 imports it through that registry function."
"Analysis/05_behavioral_state_space.R","production","Analysis/05_behavioral_state_space.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""05"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; docs/analysis_strategy_for_manuscript.md","yes","indirect","yes_with_reference_updates","State diversity and switching tables from Stage 01 metrics. S: paths at lines 44-45 (03_derived_metrics in, 06_behavioral_dynamics/state_space out). Feeds Stage 14 only; not in the primary manuscript registry."
"Analysis/06_dynamic_social_networks.R","production","Analysis/06_dynamic_social_networks.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""06"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/02_build_dyadic_rfid_contacts.R; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/test_stage19_identity_and_stage06_schema.R; docs/analysis_strategy_for_manuscript.md","yes","indirect","yes_with_reference_updates","Dynamic social network features from the Stage 02 dyadic contacts, with a Stage 01 metric fallback. S: paths at lines 46, 47, 52 (reads .../dyadic_contacts/tables/dyadic_network_ready.csv). Its output schema is contract-tested by Testing/test_stage19_identity_and_stage06_schema.R."
"Analysis/07_gamm_trajectory_features.R","production","Analysis/07_gamm_trajectory_features.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""07"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; Analysis/_archive/README_archive.md; README.md; docs/analysis_strategy_for_manuscript.md","yes","indirect","yes_with_reference_updates","Smallest active stage (8 KB). GAMM trajectory feature extraction; the active replacement for the archived _archive/04_gamm_movement_proximity_phase_and_early_window.R. S: paths at lines 44-45."
"Analysis/08_hmm_behavioral_states_optional.R","production","Analysis/08_hmm_behavioral_states_optional.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""08"", GATED behind RUN_OPTIONAL_HMM = getOption(""mmm.run_optional_hmm"", TRUE) - default ON","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/audit_first_night_candidate_set_scores.R; Testing/audit_first_night_domain_scores.R; Testing/audit_first_night_domain_scores_v2.R; Testing/audit_first_night_window_sensitivity.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R","yes","yes","yes_with_reference_updates","Optional HMM state model, but the option defaults TRUE so it runs by default. Reads the S: RFID root (line 48). Configured primary is 10min_based with 5min_based always run as the required sensitivity; canonicalizes AnimalNum via canonical_animal_id() and inherits Group/Sex only from the current canonical Stage 01 roster, failing closed on alias/roster conflicts. Its outputs are the ONLY HMM artifacts Stage 14 is allowed to import, and it is cited in manuscript supplementary provenance. Heavily audited by the Testing/audit_hmm_state_architecture_* family."
"Analysis/09_early_prediction_model_ladder.R","production","Analysis/09_early_prediction_model_ladder.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""09"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/10_systems_feature_prediction_ladder.R; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; Analysis/_archive/README_archive.md; Functions/first_night_window_helpers.R; Functions/identity_correction_comparison_helpers.R; README.md; Testing/audit_first_night_candidate_set_decision.R; Testing/audit_first_night_candidate_set_scores.R; Testing/audit_first_night_domain_scores_v2.R; Testing/audit_first_night_heatmap_v2.R; Testing/audit_first_night_time_anchor.R; Testing/audit_stage09_stale_artifacts.R; Testing/test_downstream_boundary_and_gap_contract.R; Testing/test_first_night_window_parity.R; Testing/test_stage09_endpoint_identity.R; Testing/test_stage09_primary_window.R; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","THE primary manuscript analysis: does behavior in the first active 12 h after the first cage change predict later CombZ. Canonical resolution 10-min bins, 5-min as predefined sensitivity. Reads the S: RFID root (line 53) and the endpoint workbook S:/.../SIS_Analysis/E9_Behavior_Data.xlsx (line 61). Canonical output root analysis_ready/pipeline/09_early_prediction/10min/, resolved for downstream consumers by resolve_stage09_early_prediction_artifact() in _pipeline_setup.R (canonical class beats legacy class globally). Second-largest active stage (120 KB) and the most test-covered."
"Analysis/10_systems_feature_prediction_ladder.R","production","Analysis/10_systems_feature_prediction_ladder.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""10"", GATED behind RUN_SYSTEMS_EXTENSION = getOption(""mmm.run_systems_extension"", TRUE) - default ON","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/16_manuscript_behavior_report.R; README.md; Testing/audit_stage09_stale_artifacts.R; docs/analysis_strategy_for_manuscript.md","yes","indirect","yes_with_reference_updates","Secondary systems-extension prediction ladder built on Stage 09; framed as extension/sensitivity, not a replacement. Reads the S: RFID root (line 70). Canonical output root analysis_ready/pipeline/10_systems_prediction/10min/. README_pipeline.md states Stage 10 predictive claims remain exploratory and are NOT promoted to the primary registry by Stage 16."
"Analysis/11_behavioral_adaptation_kinetics.R","production","Analysis/11_behavioral_adaptation_kinetics.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""11"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/audit_phase_bug_impact.R; Testing/test_phase_classification.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R; manuscript/BehavioralDynamics_schema/README.md","yes","yes","yes_with_reference_updates","Adaptation/recovery kinetics from Stage 01 metrics. Reads the S: RFID root (line 35). Named in the manuscript supplementary-warnings schema (manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R) and covered by the phase-classification tests."
"Analysis/12_sleep_like_quiescence_metrics.R","production","Analysis/12_sleep_like_quiescence_metrics.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""12"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/audit_first_night_domain_scores_v2.R; Testing/audit_phase_bug_impact.R; Testing/test_phase_classification.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/01_verify_rest_domain_collapse.R; manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R; manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R; manuscript/BehavioralDynamics_schema/README.md","yes","yes","yes_with_reference_updates","Inactivity-bout / sleep-like quiescence summaries. Reads the S: RFID root (line 34). The manuscript schema's rest-domain-collapse verification (manuscript/BehavioralDynamics_schema/R/01_verify_rest_domain_collapse.R) is written specifically against this stage's definitions, so its semantics are directly manuscript-load-bearing."
"Analysis/13_ethological_phase_organization.R","production","Analysis/13_ethological_phase_organization.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""13"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/test_phase_classification.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R; manuscript/BehavioralDynamics_schema/README.md","yes","yes","yes_with_reference_updates","Phase contrast, timing, fragmentation and recovery features. Reads the S: RFID root (line 36). Shares the phase-classification contract with Stages 11 and 12 (Testing/test_phase_classification.R covers all three)."
"Analysis/14_systems_neuroscience_summary_dashboard.R","production","Analysis/14_systems_neuroscience_summary_dashboard.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""14"", unconditional","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/16_manuscript_behavior_report.R; Analysis/_supporting/13_nonlinear_systems_dynamics.R; Functions/first_night_domain_driver.R; README.md; Testing/audit_first_night_candidate_set_scores.R; Testing/audit_first_night_domain_scores.R; Testing/audit_first_night_domain_scores_v2.R; Testing/audit_hmm_state_architecture_components.R; Testing/audit_stage09_stale_artifacts.R; Testing/test_hmm_stage14_contract.R; Testing/test_stage14_upstream_registry.R; docs/analysis_strategy_for_manuscript.md; manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R; manuscript/BehavioralDynamics_schema/R/30_write_number_manifest.R; manuscript/BehavioralDynamics_schema/data/figure_number_manifest.csv","yes","yes","yes_with_reference_updates","By far the largest file in the repo (364 KB). Integrated systems dashboard consuming Stages 01 and 04-13 plus optional proteomics. Reads the S: RFID root (line 73) and the endpoint workbook S:/.../SIS_Analysis/E9_Behavior_Data.xlsx (line 119). Its Stage 04 / Stage 09 imports MUST go through resolve_stage04_temporal_instability_artifact() / resolve_stage09_early_prediction_artifact() in _pipeline_setup.R so the audit can never disagree with what was loaded (Testing/test_stage14_upstream_registry.R enforces this). Fig_sis_active_inactive_domain_heatmap and the HMM-resolution sensitivity output are cited in manuscript/BehavioralDynamics_schema/data/figure_number_manifest.csv."
"Analysis/15_behavior_proteomics_integration.R","production","Analysis/15_behavior_proteomics_integration.R","KEEP (not moved)",TRUE,"Analysis/run_all_analysis.R stage ""15"", GATED behind RUN_BEHAVIOR_PROTEOMICS = getOption(""mmm.run_behavior_proteomics"", FALSE) - default OFF, so it does NOT run unless explicitly enabled","Analysis/run_all_analysis.R; Analysis/README_pipeline.md; Analysis/14_systems_neuroscience_summary_dashboard.R; Analysis/16_manuscript_behavior_report.R; README.md; Testing/audit_stage09_stale_artifacts.R; docs/analysis_strategy_for_manuscript.md","yes","indirect","yes_with_reference_updates","The only stage in the runner that is OFF by default. Reads the S: RFID root (line 30), S:/.../RFID/analysis_ready/proteomics (line 65) and .../analysis_ready/06_behavioral_dynamics (line 67); also depends on the separate proteomics working root (S:/Lab_Member/Tobi/Experiments/Exp9_Social-Stress/Analysis/proteomics), so it has the widest external-data footprint of any stage. README_pipeline.md: behavior-proteomics outputs remain exploratory and are not promoted to the primary manuscript registry."
"Analysis/16_manuscript_behavior_report.R","production","Analysis/16_manuscript_behavior_report.R","KEEP (not moved)",TRUE,"NOT in run_all_analysis.R. Run manually as a separate assembly step: source(""Analysis/16_manuscript_behavior_report.R"") after canonical Stage 09 and selected Stage 03 outputs exist (documented in Analysis/README_pipeline.md and README.md)","Analysis/README_pipeline.md; Functions/identity_correction_comparison_helpers.R; Testing/audit_stage09_stale_artifacts.R; Testing/compare_identity_correction_before_after.R; Testing/test_output_path_length.R; docs/analysis_strategy_for_manuscript.md","yes","yes","yes_with_reference_updates","Functionally the manuscript_builder / export-only assembly layer, but classified production here because it is an active numbered pipeline stage and destination is KEEP. It fits no models and recalculates no statistics: it reads existing canonical Stage 09, Stage 03 and Stage 00 QC tables and writes the manuscript package to analysis_ready/manuscript/behavior/ (entry point Behavioral_Source_Data.xlsx plus primary/supplementary results, animal-level and held-out prediction source data, movement-phase source data, provenance, validation, SHA-256 manifest). It carries an internal pipeline registry that names Analysis/19_spatial_occupancy_maps.R at line 1829 as stage ""19"", ""secondary/spatial"", ""legacy_pending_migration"", with the explicit annotation ""Position-level source and separate legacy publication paths; not included in Stage 16."" Reads the S: RFID root at line 31."
"Analysis/19_spatial_occupancy_maps.R","production","Analysis/19_spatial_occupancy_maps.R","KEEP (not moved)",TRUE,"No runner. Standalone manual execution only - it is absent from run_all_analysis.R's tribble AND from the README_pipeline.md run-order table. Its only automated consumers are Testing/test_stage19_group_sex_labels.R, Testing/test_stage19_identity_and_stage06_schema.R and Testing/test_downstream_boundary_and_gap_contract.R, which read/parse it rather than execute the analysis","Analysis/16_manuscript_behavior_report.R (line 1829 registry entry); Testing/test_downstream_boundary_and_gap_contract.R; Testing/test_stage19_group_sex_labels.R; Testing/test_stage19_identity_and_stage06_schema.R","yes","indirect","yes_with_reference_updates","The one active stage that does NOT use _pipeline_setup.R. It resolves the repo with the absolute-fallback pattern at line 39: .mmm_repo <- Sys.getenv(""MMM_REPO_DIR"", unset = ""C:/Users/topohl/Documents/GitHub/MMMSociability""), then sources Functions/animalpos_preprocessing_helpers.R and Functions/behavioral_dynamics_helpers.R from there - absolute, therefore depth-independent and safe to move. Builds reader-occupancy maps from the PRIOR position-level dataset rather than all_behavior_metrics.csv (line 62 RAW_POSITION_DIR = S:/.../RFID/MMMSociability/preprocessed_data - data-path ""MMMSociability"", do not rename), plus sus_animals.csv / con_animals.csv (lines 64-65). Actively maintained (three recent commits: 4df98b6, c16817e, 15bcb67) and identity-corrected to the same canonical_animal_id contract as Stages 01/03/09, but Stage 16 explicitly excludes it from the manuscript package and no manuscript/ file references spatial occupancy. Three Testing scripts hard-code the literal string ""Analysis/19_spatial_occupancy_maps.R"" in readLines()/load_defs() calls, so any move requires updating those."
"Analysis/README_pipeline.md","production","Analysis/README_pipeline.md","KEEP (not moved)",TRUE,"Documentation only; it is the canonical run-order and stage-contract doc pointed to by README.md line 77 and line 340","README.md; docs/analysis_strategy_for_manuscript.md","no","yes","no","Documents stages 00-16 in a run-order table, the primary-vs-secondary framing, the Stage 08 HMM / Stage 14 state-architecture contract, the canonical analysis_ready/pipeline/<stage>/ output layout, the options() block for run_all_analysis.R, and an old-to-new filename map that also records the six _archive/ moves. Doc gap worth noting (documentation only, no scientific impact): the run-order table stops at Stage 16 and never mentions 19_spatial_occupancy_maps.R. Keep in place: README.md and docs/analysis_strategy_for_manuscript.md reference the literal path Analysis/README_pipeline.md."
"Analysis/run_all_analysis.R","production","Analysis/run_all_analysis.R","KEEP (not moved)",TRUE,"Human entry point; documented in README.md (lines 87, 226, 245, 258) and Analysis/README_pipeline.md ""Running Everything""","Analysis/README_pipeline.md; Analysis/_archive/README_archive.md; README.md; docs/analysis_strategy_for_manuscript.md","yes","yes","no","STAGE LIST VERBATIM from the tibble::tribble(~script, ~stage, ~optional_flag, ~role) pipeline, in execution order: ""00_qc_tracking_integrity.R"" (00, no flag), ""01_build_multiscale_behavior_metrics.R"" (01, no flag), ""02_build_dyadic_rfid_contacts.R"" (02, no flag), ""03_primary_raw_movement_phase_stats.R"" (03, no flag), ""04_temporal_instability.R"" (04, no flag), ""05_behavioral_state_space.R"" (05, no flag), ""06_dynamic_social_networks.R"" (06, no flag), ""07_gamm_trajectory_features.R"" (07, no flag), ""08_hmm_behavioral_states_optional.R"" (08, flag ""RUN_OPTIONAL_HMM""), ""09_early_prediction_model_ladder.R"" (09, no flag), ""10_systems_feature_prediction_ladder.R"" (10, flag ""RUN_SYSTEMS_EXTENSION""), ""11_behavioral_adaptation_kinetics.R"" (11, no flag), ""12_sleep_like_quiescence_metrics.R"" (12, no flag), ""13_ethological_phase_organization.R"" (13, no flag), ""14_systems_neuroscience_summary_dashboard.R"" (14, no flag), ""15_behavior_proteomics_integration.R"" (15, flag ""RUN_BEHAVIOR_PROTEOMICS""). ANSWERS: Stage 16 is NOT in the runner (assembled separately via source(""Analysis/16_manuscript_behavior_report.R"")). Stage 19 is NOT in the runner and is not in the README_pipeline.md run-order table either. Stage 15 IS in the runner but is gated and OFF by default. Gated stages and exact option names: 08 -> RUN_OPTIONAL_HMM <- isTRUE(getOption(""mmm.run_optional_hmm"", TRUE)); 10 -> RUN_SYSTEMS_EXTENSION <- isTRUE(getOption(""mmm.run_systems_extension"", TRUE)); 15 -> RUN_BEHAVIOR_PROTEOMICS <- isTRUE(getOption(""mmm.run_behavior_proteomics"", FALSE)). A fourth non-stage option controls error handling: CONTINUE_ON_ERROR <- isTRUE(getOption(""mmm.continue_on_error"", FALSE)). Runner does setwd(MMM_REPO_ROOT) with on.exit restore and sources each stage into new.env(parent = .GlobalEnv). Mechanically it resolves analysis_dir from sys.frame(1)$ofile with a getwd()/""Analysis"" fallback, so moving it out of Analysis/ would still work when sourced from the repo root but would silently depend on CWD; keep it in Analysis/ because README.md and README_pipeline.md hard-code ""Analysis/run_all_analysis.R""."
"Analysis/run_cookiehab_preprocessing_and_metrics.R","preprocessing_active","Analysis/run_cookiehab_preprocessing_and_metrics.R","KEEP (not moved)",TRUE,"None - no file in the repo references it. Human-invoked entry point for the separate cookie-habituation dataset","none","yes","no","yes","Lightweight auxiliary runner for a DIFFERENT dataset (dataset_id = ""cookiehab""). It sets the mmm.dataset_id / mmm.preprocessed_dir / mmm.cookiehab_preprocessed_dir / mmm.derived_metrics_dir / mmm.dyadic_contacts_dir options to the S:/.../RFID/cookiehab/... tree (lines 18-21), then sources Formatting/01_preprocess_cookiehab_animalpos.R followed by Analysis/01_build_multiscale_behavior_metrics.R and Analysis/02_build_dyadic_rfid_contacts.R. It is the live consumer of the cookiehab branches inside Stages 01 and 02, so those branches are not dead code. Zero inbound references and no mention in README.md, Analysis/README_pipeline.md or docs/analysis_strategy_for_manuscript.md - the only Analysis/ file that is entirely undocumented. Uses the same sys.frame/getwd()-fallback analysis_dir resolution as run_all_analysis.R, so it is technically movable; keeping it beside run_all_analysis.R is the low-risk choice, but if a runners/ subfolder is created it is the safest candidate to move since nothing points at it."
"docs/analysis_strategy_for_manuscript.md","manuscript_builder","docs/analysis_strategy_for_manuscript.md","KEEP (not moved)",TRUE,"Documentation; not executed. It is the normative manuscript-language contract the analysis is written against.","README.md:68 (repo structure listing 'docs/  # Manuscript/analysis strategy notes'); manuscript/BehavioralDynamics_schema/README.md:81; manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R:168","no","yes","no","604 lines. The only file in docs/. Defines the central biological hypothesis, primary-vs-secondary analysis split, and the forbidden/preferred language lists. Two places cite it BY PATH AND LINE NUMBER (docs/analysis_strategy_for_manuscript.md:215-238, the phase-organization section that forbids 'circadian disruption' / 'sleep disruption' - I confirmed that line range does contain exactly that section). Both citations are baked into text: one is a literal string drawn into the rendered supplementary figure, so a path change would invalidate a published figure caption. Treat as pinned - do not move or renumber."
"Formatting/_archive/E9_SIS_AnimalPos-preprocessing.R","legacy","Formatting/E9_SIS_AnimalPos-preprocessing.R","Formatting/_archive/E9_SIS_AnimalPos-preprocessing.R",TRUE,"none","Formatting/00a_raw_tracking_qc_rfid_loss.R:7,30,69 (workflow comments only); Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing.R:21 (comment naming the parallel sibling)","yes","no","yes","Historical serial predecessor of E9_SIS_AnimalPos-preprocessing_parallell.r. Last touched 2025-04-11 and only by a rename commit; superseded in substance by the 2026-08-27 synthetic-row refactor. Already broken as written: line 22 does source('C:/Users/topohl/Documents/GitHub/MMMSociability/E9_SIS_AnimalPos-functions.R') - a REPO-ROOT path, but the file lives at Functions/E9_SIS_AnimalPos-functions.R, so this script cannot run today. It also hardcodes setwd('S:/.../RFID/MMMSociability') at line 20 and calls the retired preprocess_file() path that inserted synthetic boundary rows - i.e. running it would REGRESS the dataset to the pre-refactor state with 80,190 fabricated rows. Archive it; that also removes a live footgun. Note the two MMMSociability occurrences here (line 20 = local DATA path, line 22 = local clone path) must NOT be renamed."
"Formatting/_archive/E9_SIS_lme_data_formatting.r","legacy","Formatting/E9_SIS_lme_data_formatting.r","Formatting/_archive/E9_SIS_lme_data_formatting.r",TRUE,"none","none","yes","no","yes","Zero references anywhere in the repo - the only file in Formatting/ with an empty reference graph. Both ends of its data flow are legacy: its INPUT is S:/.../MMMSociability/tables/noHomeCage (line 19), an output directory produced only by the archived E9_SIS_AnimalPos-analyzing* scripts under Analysis/_archive/legacy_deprecated/ and Testing/; its OUTPUT, processed_data/data_lme_format/data_filtered_agg.csv, is consumed only by Analysis/_archive/04_gamm_movement_proximity_phase_and_early_window.R:130, Analysis/_archive/18_raw_movement_publication_trajectory.R:42, Analysis/_archive/legacy_deprecated/lme v.3.1.1.r:61, .../sociability_lme v.1.0.0.r:70, v.1.0.1.r:70 and Testing/lme v.3.1.2.r:61 - every one archived or a legacy non-test_* Testing script. Sources nothing, so archiving is inert. It also reads con_animals.csv/sus_animals.csv from raw_data/ rather than the S:/.../Analysis/ location the active cookiehab path uses, confirming it predates the current layout."
"Formatting/00a_raw_tracking_qc_rfid_loss.R","data_dependent_audit","Formatting/00a_raw_tracking_qc_rfid_loss.R","KEEP (not moved)",TRUE,"None. No script in the repo source()s it - it is run manually. Analysis/14_systems_neuroscience_summary_dashboard.R:1967 names its OUTPUT filename, not the script, and only as the second element of a fallback vector whose primary is Analysis/00_qc_tracking_integrity.R's output.","Analysis/14_systems_neuroscience_summary_dashboard.R:1967 (output-artifact path only, marked expected_status='qc_available_if_upstream_run'); Formatting/00a_raw_tracking_qc_rfid_loss.R:7,30,69 (its own workflow comments referencing E9_SIS_AnimalPos-preprocessing.R). Not referenced by README.md or any doc.","yes","indirect","yes","ANSWERED CAREFULLY - it is NOT referenced by anything active via source(). The only linkage is a data-path string at Analysis/14:1967, and that string does not even resolve to where this script writes: Stage 14 looks under file.path(dirname(project_root),'raw_tracking_qc_rfid_loss',...) where project_root = 'S:/.../Analysis/Behavior/RFID' (so it searches S:/.../Analysis/Behavior/), whereas 00a writes to file.path(WORKING_DIR,'raw_tracking_qc_rfid_loss') = S:/.../RFID/MMMSociability/raw_tracking_qc_rfid_loss (line 64). That fallback is therefore already stale and never resolves; Stage 14's primary path, analysis_ready/00_tracking_qc_rfid_loss/tables/tracking_qc_by_animal.csv, comes from Analysis/00_qc_tracking_integrity.R:494. NOT an E9 historical variant: dated 2026-05-18 (commit 20d463c 'Robustify RFID parsing, mapping, and QC metrics'), and it is complementary rather than superseded - it is the only RAW-level chip-loss QC (reads raw_data/B*/E9_SIS_B*_CC*_AnimalPos.csv before preprocessing), while Analysis/00 does the same diagnosis on derived metrics AFTER Stage 01. Recommend KEEP alongside the active preprocessing entrypoint whose workflow it documents. Archiving is mechanically safe if the parent prefers, but it would misfile a current, runnable QC tool as historical. || MOVE_BREAKS: Nothing mechanically. It sources nothing from the repo (no _pipeline_setup.R, no Functions/ helpers) and resolves its working directory via DEFAULT_WORKING_DIR = 'S:/.../RFID/MMMSociability' with a detect_working_dir() fallback keyed on the presence of a raw_data/ subdirectory - all absolute or data-side, entirely independent of its own location in the repo."
"Formatting/01_preprocess_cookiehab_animalpos.R","preprocessing_active","Formatting/01_preprocess_cookiehab_animalpos.R","KEEP (not moved)",TRUE,"Analysis/run_cookiehab_preprocessing_and_metrics.R:24 - source(file.path(MMM_REPO_ROOT, ""Formatting"", ""01_preprocess_cookiehab_animalpos.R""), local = new.env(parent = .GlobalEnv))","Analysis/run_cookiehab_preprocessing_and_metrics.R:24 (sole reference in the entire repo)","yes","indirect","no","Active cookiehab (EPMaftercagechange) preprocessing stage. Sources Functions/cookiehab_preprocessing_helpers.R at line 24. All I/O paths are getOption()-overridable (mmm.cookiehab_raw_dir, mmm.cookiehab_preprocessed_dir, mmm.cookiehab_qc_dir, mmm.sus_animals_file, mmm.con_animals_file) with S:/ defaults, and the runner sets those options before sourcing - a good pattern, but it means the script hard-fails at line 49 (stop('Cookiehab raw root does not exist')) without the S: drive. It is the only numbered, modern-style script in Formatting/; it stays where it is because the runner names its directory literally. || MOVE_BREAKS: Analysis/run_cookiehab_preprocessing_and_metrics.R:24 hardcodes the directory 'Formatting' and the exact filename; moving the file requires editing that line. Secondly the script's own bootstrap (lines 15-23) probes only file.path(getwd(),'Analysis','_pipeline_setup.R'), file.path(dirname(getwd()),'Analysis','_pipeline_setup.R') and dirname(ofile)/../Analysis/_pipeline_setup.R - the third candidate is relative to the script's own location, so moving it one level deeper (e.g. Formatting/cookiehab/) breaks that candidate; the getwd()-based candidates still work because the runner does setwd(MMM_REPO_ROOT) at line 12 before sourcing, so a move is survivable but only by accident."
"Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r","preprocessing_active","Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r","KEEP (not moved)",TRUE,"None sources it - it is a top-level entrypoint, run manually (Rscript) to regenerate the primary dataset. Its OUTPUT, S:/.../RFID/MMMSociability/preprocessed_data/*_preprocessed.csv, is the input of Analysis/01_build_multiscale_behavior_metrics.R:68,260-261, Analysis/02_build_dyadic_rfid_contacts.R, Analysis/14, Analysis/19 and Testing/validate_cross_scale_animal_identity.R.","Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing.R:21 (comment); Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R:22 (comment); Formatting/00a_raw_tracking_qc_rfid_loss.R:7,30,69 (workflow comments, naming the non-parallel sibling)","yes","yes","no","DO NOT ARCHIVE despite the legacy 'E9_SIS_' name and the lowercase .r extension. This is the ACTIVE preprocessing entrypoint: last touched 2026-08-27 (commit 81694c7), and its 30-line architecture note documents the refactor that replaced preprocess_file() with preprocess_animalpos_file(), removing 80,190 fabricated rows (11.9% of all rows) from the analysis record, verified against RFID_snapshot_pre_stage01_refactor_20260827 with 0 genuine reads lost and 0 rows invented. It sources BOTH Functions/E9_SIS_AnimalPos-functions.R and Functions/animalpos_preprocessing_helpers.R (lines 52-57), re-sources them inside each foreach worker (line 89), and enforces stopifnot(sum(audit$n_synthetic_rows) == 0L) at the end. The name is misleading relative to its status - consider a rename in a later, non-frozen release, but not now. || MOVE_BREAKS: Nothing mechanically - it resolves the repo via Sys.getenv('MMM_REPO_DIR', unset='C:/Users/topohl/Documents/GitHub/MMMSociability') and setwd(data_dir), both absolute, so it is depth-independent. The reason not to move it is semantic: this is the producer of every preprocessed_data/*_preprocessed.csv the pipeline consumes. Archiving it would falsely signal the primary dataset is generated by dead code, and it is the sole active consumer of Functions/E9_SIS_AnimalPos-functions.R."
"Functions/animal_identity_invariants_helpers.R","production","Functions/animal_identity_invariants_helpers.R","KEEP (not moved)",TRUE,"Testing/validate_cross_scale_animal_identity.R:35 (data-dependent QA driver) and Testing/test_animal_identity_invariants_engine.R:11 (portable unit test). NO Analysis/ stage sources it.","Testing/test_animal_identity_invariants_engine.R:2,11; Testing/validate_cross_scale_animal_identity.R:23,35","no","indirect","no","GROUP=identity; pure validation engine for the post-Stage-01 cross-scale identity invariants. Deliberately side-effect free and I/O free at source() time so it can be exercised against synthetic in-memory/tempdir fixtures; Testing/validate_cross_scale_animal_identity.R supplies the real reading. Not sourced by any production stage - it underwrites the identity-consistency claim rather than producing manuscript numbers, hence manuscript_relevant=indirect. Keep in Functions/: it is the engine half of an intentional engine/driver split, and moving it to Testing/ would collapse that separation. || MOVE_BREAKS: Both consumers use the literal CWD-relative string source(""Functions/animal_identity_invariants_helpers.R"") - not source_mmm_helper() - so moving it into any subdirectory breaks both immediately even when run from the repo root."
"Functions/animalpos_preprocessing_helpers.R","production","Functions/animalpos_preprocessing_helpers.R","KEEP (not moved)",TRUE,"Analysis/01_build_multiscale_behavior_metrics.R:62; Analysis/02_build_dyadic_rfid_contacts.R:51; Analysis/09_early_prediction_model_ladder.R:46; Analysis/14_systems_neuroscience_summary_dashboard.R:63; Analysis/19_spatial_occupancy_maps.R:40; Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r:54 (active preprocessing entrypoint); Functions/first_night_window_helpers.R:45","Analysis/01:62; Analysis/02:51; Analysis/09:46; Analysis/14:63; Analysis/19:40-42; Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r:54; Functions/first_night_window_helpers.R:45,47; Testing/test_animalpos_preprocessing_helpers.R:5,149,150; Testing/test_downstream_boundary_and_gap_contract.R:32; Testing/test_first_night_domain_contract.R:10; Testing/test_first_night_window_parity.R:17; Testing/test_stage09_primary_window.R:21; Testing/test_stage19_identity_and_stage06_schema.R:26; Testing/audit_first_night_* (9 files)","no","yes","no","GROUP=preprocessing; defines the CURRENT preprocessing contract: parse_animalpos_datetime, animalpos_position_id, animalpos_phase_counters, animalpos_half_hours_elapsed, animalpos_split_intervals and preprocess_animalpos_file() (the synthetic-row-free replacement for the legacy preprocess_file()). CRITICAL COUPLING: line 205 declares the default argument `remove_phases_fn = remove_phases`, and remove_phases() is defined ONLY in Functions/E9_SIS_AnimalPos-functions.R:194. R lazy evaluation means Analysis stages that source this helper but never call preprocess_animalpos_file() are unaffected; only Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r actually triggers it, and that script sources both files. No top-level data reads. || MOVE_BREAKS: source_mmm_helper() hardcodes file.path(MMM_REPO_ROOT, 'Functions', <flat filename>) with no subdirectory support (Analysis/_pipeline_setup.R:40-41); Analysis/19_spatial_occupancy_maps.R:40 and four Testing scripts build file.path(repo,'Functions','animalpos_preprocessing_helpers.R') literally; Testing/test_animalpos_preprocessing_helpers.R:149-150 readLines() the literal relative path 'Functions/animalpos_preprocessing_helpers.R' to source-scan it."
"Functions/behavioral_dynamics_helpers.R","production","Functions/behavioral_dynamics_helpers.R","KEEP (not moved)",TRUE,"Analysis/_pipeline_setup.R:52 sources it unconditionally at setup time, so EVERY Analysis/ stage 00-19, every runner (Analysis/run_all_analysis.R, Analysis/run_cookiehab_preprocessing_and_metrics.R), Formatting/01_preprocess_cookiehab_animalpos.R, manuscript/Fig1_behavior_candidates/build_fig1_candidates.R and every Testing script that sources _pipeline_setup.R loads it. Analysis/19_spatial_occupancy_maps.R:51 also resolves it explicitly.","Analysis/_pipeline_setup.R:24,52; Analysis/19_spatial_occupancy_maps.R:51-53; Analysis/01_build_multiscale_behavior_metrics.R:21 (comment); Functions/behavioral_dynamics_stats_helpers.R:177; Functions/hmm_stage14_helpers.R:6; Testing/check_behavioral_dynamics_structure.R:27; Testing/test_canonical_animal_id.R:2; Testing/test_first_night_window_parity.R:19; Testing/test_stage19_identity_and_stage06_schema.R:25; Testing/test_write_registry_rerun_behavior.R:4,64; manuscript/BehavioralDynamics_schema/R/schema_toolkit.R:18; manuscript/BehavioralDynamics_schema/README.md:22; README.md:65; several Analysis/_archive scripts","no","yes","no","GROUP=output/provenance; foundation module, the single most load-bearing file in the repo. Mixed content: output/provenance layer (write_table, write_output_manifest, harmonize_analysis_outputs, canonicalize_behavior_output_path, mmm_output_dir_slug, mmm_assert_output_path_budget), identity (canonical_animal_id at line 38), time-series primitives (calc_rmssd, calc_acf1, calc_dfa_alpha, calc_transition_metrics, calc_circadian_metrics) and plot themes/palette (line 66). No top-level file I/O at source() time; pure definitions plus library() calls. This file must not move, not be renamed, and its line 66 palette must not shift if the schema README pin is to stay accurate. || MOVE_BREAKS: HARD BLOCK: Analysis/_pipeline_setup.R:24 find_mmm_repo_root() uses file.exists(file.path(current, 'Functions', 'behavioral_dynamics_helpers.R')) && dir.exists(file.path(current,'Analysis')) as the repo-root SENTINEL. Moving this one file makes repo-root detection fail for the entire pipeline (every stage, runner, audit and test). Additionally source_mmm_helper() resolves file.path(MMM_REPO_ROOT,'Functions',<flat name>) with no subdirectory support, and manuscript/BehavioralDynamics_schema/README.md pins the house palette to line 66."
"Functions/behavioral_dynamics_stats_helpers.R","production","Functions/behavioral_dynamics_stats_helpers.R","KEEP (not moved)",TRUE,"Analysis/04_temporal_instability.R:42; Analysis/05_behavioral_state_space.R:36; Analysis/06_dynamic_social_networks.R:38; Analysis/09_early_prediction_model_ladder.R:44; Analysis/14_systems_neuroscience_summary_dashboard.R:59","Analysis/04:42; Analysis/05:36; Analysis/06:38; Analysis/09:44; Analysis/14:59; Analysis/_archive/08_early_prediction_models.R:36; Analysis/_supporting/13_nonlinear_systems_dynamics.R:62; Analysis/_supporting/14_nextgen_behavioral_phenotyping.R:68","no","yes","no","GROUP=statistics; CON/RES/SUS group summaries and pairwise contrasts (cohens_d_pooled, safe_welch_test, normalise_group_levels, make_dynamics_group_summary, make_dynamics_group_contrasts, write_dynamics_statistics). Line 177 raises a runtime stop() if write_table() from behavioral_dynamics_helpers.R has not been sourced first, so the load order _pipeline_setup.R -> this file must be preserved. Has no test_*.R of its own - a test-coverage gap worth noting, but not a restructuring blocker. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution used by all five active stages."
"Functions/cookiehab_preprocessing_helpers.R","production","Functions/cookiehab_preprocessing_helpers.R","KEEP (not moved)",TRUE,"Formatting/01_preprocess_cookiehab_animalpos.R:24, which is itself sourced by Analysis/run_cookiehab_preprocessing_and_metrics.R:24","Formatting/01_preprocess_cookiehab_animalpos.R:24 (sole reference in the entire repo)","no","indirect","no","GROUP=preprocessing; narrowest-scope helper in Functions/ - only cookiehab_read_animalpos_csv and companions for the cookie-habituation (EPMaftercagechange) AnimalPos files, a side dataset run through the same Stage 01/02 metric machinery by Analysis/run_cookiehab_preprocessing_and_metrics.R. Single consumer, so it is the one Functions/ file whose reference graph is trivial to update - but there is no benefit to moving it, and it has no test_*.R coverage. requires_local_data=no at source() time; its functions need S:/.../Raw Data/Behavior/RFID/MMMdata at call time. || MOVE_BREAKS: source_mmm_helper('cookiehab_preprocessing_helpers.R') at Formatting/01:24 resolves file.path(MMM_REPO_ROOT,'Functions', <flat name>); a subdirectory would not be found and source_mmm_helper() would stop() since required=TRUE by default."
"Functions/duration_normalization_helpers.R","production","Functions/duration_normalization_helpers.R","KEEP (not moved)",TRUE,"Analysis stages 01, 02, 04, 05, 06, 07, 08, 09, 11, 12, 13, 14 (12 active stages source it via source_mmm_helper)","Analysis/01:61; Analysis/02:50; Analysis/04:43; Analysis/05:37; Analysis/06:39; Analysis/07:37; Analysis/08:39; Analysis/09:45; Analysis/11:32; Analysis/12:31; Analysis/13:33; Analysis/14:60; Testing/audit_first_night_production_parity.R:7; Testing/audit_hmm_state_architecture_identifiability.R:8; Testing/test_phase_classification.R:7,97; Analysis/_archive/04_gamm_movement_proximity_phase_and_early_window.R:94-95","no","yes","no","GROUP=temporal/window; the most widely sourced domain helper (12 active stages). Provides epoch-duration QC, short-epoch/short-cage-change detection without hard-coded CC labels, and rate-normalized versions of count/cumulative metrics. Declares a hard dependency at line 26: source_mmm_helper('phase_classification_helpers.R'), with an explicit stop() at line 29 if unavailable. Pure definitions; no data reads at source() time. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution; Testing/test_phase_classification.R:97 checks the literal relative path 'Functions/duration_normalization_helpers.R'."
"Functions/E9_SIS_AnimalPos-functions.R","preprocessing_active","Functions/E9_SIS_AnimalPos-functions.R","KEEP (not moved)",TRUE,"EXACTLY ONE active consumer: Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r, at lines 52-57 (function_files <- file.path(repo_dir,'Functions', c('E9_SIS_AnimalPos-functions.R','animalpos_preprocessing_helpers.R')); stopifnot(all(file.exists(function_files))); invisible(lapply(function_files, source))) and again inside every parallel worker at line 89. That script is the ACTIVE main-dataset preprocessing entrypoint (last modified 2026-08-27, commit 81694c7 'Separate measurement events from aggregation boundaries') and produces preprocessed_data/*_preprocessed.csv, which is the input of Analysis/01_build_multiscale_behavior_metrics.R:68. NO Analysis/ stage 00-19, NO runner, NOT Formatting/01_preprocess_cookiehab_animalpos.R, NO manuscript builder and NO Testing/test_*.R sources it.","ACTIVE (1): Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r:53. IMPLICIT-ACTIVE (1): Functions/animalpos_preprocessing_helpers.R:205 default arg remove_phases_fn = remove_phases, where remove_phases() is defined ONLY here (line 194). LEGACY-BUT-REAL source() (1): Formatting/E9_SIS_AnimalPos-preprocessing.R:22 (broken path - points at repo-root .../MMMSociability/E9_SIS_AnimalPos-functions.R, file does not exist there). COMMENTED-OUT (4): Formatting/E9_SIS_AnimalPos-preprocessing.R:46; Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.0.r:136; Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing.R:54; Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R:81. ARCHIVED source() (6): Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-analyzing-shannon v.1.0.0.r:137, ...-shannon v.1.0.1.r:464, ...-shannon.R:52, E9_SIS_AnimalPos-analyzing.R:53, E9_SIS_AnimalPos-comparing.R:25, E9_SIS_AnimalPos-comparing2.R:31. ARCHIVED candidate-vector (4 lines): Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-plots_and_statistics.R:66-69. LEGACY Testing (non-test_*) (6): Testing/E9_SIS_AnimalPos-analyzing 2.0.0.R:14; Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R:70-74,78; Testing/E9_SIS_AnimalPos-analyzing-shannon v.1.0.2.r:490. DOC-ONLY (2): Analysis/_archive/03b_multiscale_metric_validation_notes.md:7; Analysis/_archive/legacy_deprecated/E9_SIS_AnimalPos-comparing2.R:13 (comment). SELF (1): Functions/E9_SIS_AnimalPos-functions.R:1 (@file tag). Near-duplicate fork, not a reference: Testing/E9_SIS_AnimalPos-functions v.2.0.0.R.","no","yes","yes_with_reference_updates","GROUP=preprocessing; DEFINITIVE ANSWER: NO active production script in the enumerated set (Analysis/ stage 00-19, runners, Formatting/01_preprocess_cookiehab_animalpos.R, Functions/ helpers via source(), manuscript builders, Testing/test_*.R) sources this file. BUT it is NOT dead code: it is a live runtime dependency of Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r, the current active preprocessing entrypoint that generates the entire analysed dataset. Only 2 of its 30 defined functions are still load-bearing - remove_phases() (line 194) and count_half_hours_elapsed() (line 231), which Functions/animalpos_preprocessing_helpers.R:198-202 documents as 'reused unchanged so epoch inclusion semantics are preserved'. The other ~28 (preprocess_file, compute_phase_transitions, compute_day_transitions, add_half_hour_transitions, find_id, update_proximity/position/movement, generateHeatMap*, testAndPlotVariable, performPosthocAnova, ...) are superseded dead weight, several of them the exact synthetic-row-inserting functions the 2026-08-27 refactor removed from the pipeline. RECOMMENDATION: KEEP at Functions/E9_SIS_AnimalPos-functions.R. Do NOT archive it - archiving silently breaks preprocessing reproducibility with a lazy-evaluation failure that no test in the repo would catch. If cleanup is desired post-release, the correct move is to lift remove_phases() and count_half_hours_elapsed() verbatim into animalpos_preprocessing_helpers.R and only then archive - but that is a code change, out of scope for a no-science-changes restructure. || MOVE_BREAKS: Formatting/E9_SIS_AnimalPos-preprocessing_parallell.r:52-55 builds file.path(repo_dir,'Functions','E9_SIS_AnimalPos-functions.R') and guards it with stopifnot(all(file.exists(function_files))) - moving the file makes the ACTIVE preprocessing entrypoint abort at startup with a stopifnot failure, before any worker starts. A second, subtler break: Functions/animalpos_preprocessing_helpers.R:205 defaults remove_phases_fn = remove_phases and remove_phases() exists nowhere else in the active tree, so if the source() were dropped, preprocess_animalpos_file() would fail at line 241 with 'could not find function remove_phases' - and because the default is lazily evaluated it would fail only at call time, i.e. only during a real preprocessing run, never during any Analysis stage or test. Both references would have to be updated together."
"Functions/first_night_domain_driver.R","production","Functions/first_night_domain_driver.R","KEEP (not moved)",TRUE,"Analysis/14_systems_neuroscience_summary_dashboard.R:66","Analysis/14:66,5602; Testing/audit_first_night_production_parity.R:13; Testing/test_first_night_domain_contract.R:14","no","yes","no","GROUP=first-night; single Stage-14 entry point for BOTH first-night resolutions; pure orchestration - formula, window, scaling, completeness and inference contracts all live in first_night_domain_helpers.R. NOTE: unlike the other Functions/ files this one DOES do file I/O inside its functions (line 37 file.exists / line 40 read_csv of the stage input, plus writing resolution-scoped artifacts), but nothing is read at source() time, so sourcing remains data-independent. Naming caveat: it is a 'driver' living in Functions/, which reads oddly, but it is sourced as a library by Stage 14, not invoked as a script - do not relocate it to Analysis/. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution from Analysis/14:66 and two Testing scripts."
"Functions/first_night_domain_helpers.R","production","Functions/first_night_domain_helpers.R","KEEP (not moved)",TRUE,"Analysis/14_systems_neuroscience_summary_dashboard.R:65; Functions/first_night_domain_driver.R:15","Analysis/14:65,5601; Functions/first_night_domain_driver.R:7,15; Testing/audit_first_night_production_parity.R:12; Testing/test_first_night_domain_contract.R:13","no","yes","no","GROUP=first-night; the ONE canonical implementation of the audited first-night panel; it replaced three separate row-count window constructions and a generic equal-weight inventory scorer that previously lived inline in Stage 14 (see Analysis/14:5601-5602). Header pins the contract to audit branch commits ec1edda / 2075739 / edc49e0. Line 46 loops over hard deps c('phase_classification_helpers.R','first_night_window_helpers.R'). Contract-locked by Testing/test_first_night_domain_contract.R. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution from Analysis/14:65 and from Functions/first_night_domain_driver.R:15 (bare filename)."
"Functions/first_night_window_helpers.R","production","Functions/first_night_window_helpers.R","KEEP (not moved)",TRUE,"Analysis/14_systems_neuroscience_summary_dashboard.R:64; Functions/first_night_domain_helpers.R:46","Analysis/14:64,5693; Functions/first_night_domain_helpers.R:46; Functions/first_night_domain_driver.R:82 (embeds the literal provenance string 'Functions/first_night_window_helpers.R :: mmm_select_first_night_window'); Testing/audit_first_night_production_parity.R:11; Testing/test_first_night_domain_contract.R:11; Testing/test_first_night_window_parity.R:18","no","yes","no","GROUP=first-night; the ONE source of truth for the prespecified first-night exposure window (first cage change, first Active phase block, 18:30 inclusive -> 06:30 exclusive, exactly 12 h). Declares hard deps at lines 38 and 45 (phase_classification_helpers.R, animalpos_preprocessing_helpers.R) with explicit stop() guards at 40 and 47. Parity-locked by Testing/test_first_night_window_parity.R. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution. Also Functions/first_night_domain_driver.R:82 writes the literal path string 'Functions/first_night_window_helpers.R :: mmm_select_first_night_window' into the emitted provenance artifact, and Analysis/14:5693 emits it in a message that Testing/test_first_night_window_parity.R asserts on - moving the file makes shipped provenance strings point at a nonexistent path."
"Functions/hmm_stage14_helpers.R","production","Functions/hmm_stage14_helpers.R","KEEP (not moved)",TRUE,"Analysis/08_hmm_behavioral_states_optional.R:40; Analysis/14_systems_neuroscience_summary_dashboard.R:61","Analysis/08:40; Analysis/14:61; Testing/test_hmm_stage14_contract.R:10; Testing/test_first_night_domain_contract.R:12; Testing/test_first_night_window_parity.R:20; and ~28 Testing/audit_hmm_* and audit_first_night_* scripts that call it UNCHANGED as the reference engine (e.g. audit_hmm_state_architecture_components.R:85, audit_hmm_state_architecture_longitudinal.R:78,550, audit_hmm_state_architecture_construct_comparison.R:59,1156)","no","yes","no","GROUP=HMM; the HMM/Stage-14 identity, semantic-state and inference engine (build_hmm_epoch_scores, strict_standardize_within_context, hmm_hedges_g, fit_repeated_measures_domain_contrasts). Header at line 6 states it deliberately reuses canonical_animal_id() from behavioral_dynamics_helpers.R rather than defining a second identity contract - so it must be sourced after _pipeline_setup.R. It is the single most heavily audited helper in the repo (~30 Testing consumers treat it as the immutable reference implementation). Keep path AND line numbers stable. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution from Analysis/08, Analysis/14 and ~30 Testing scripts. Additionally several audit reports embed LINE-NUMBER provenance strings pointing into this file ('Functions/hmm_stage14_helpers.R:280-317 build_hmm_epoch_scores()' in audit_first_night_candidate_set_scores.R:420 and audit_first_night_domain_scores_v2.R:228,771; 'hmm_stage14_helpers.R:268' in audit_hmm_state_architecture_construct_comparison.R:1156). Moving is survivable for those strings but any line-number shift silently falsifies published provenance."
"Functions/identity_correction_comparison_helpers.R","production","Functions/identity_correction_comparison_helpers.R","KEEP (not moved)",TRUE,"Testing/compare_identity_correction_before_after.R:46 (data-dependent driver) and Testing/test_identity_comparison_engine.R:10 (portable unit test). NO Analysis/ stage sources it.","Testing/compare_identity_correction_before_after.R:12,46; Testing/test_identity_comparison_engine.R:2,10","no","indirect","no","GROUP=identity; pure before/after comparison engine plus the Stage 03 / Stage 09 table registry consumed by Testing/compare_identity_correction_before_after.R. Explicitly documented as having no file I/O or side effects at source() time so it can be unit-tested against synthetic tempdir fixtures. Same engine/driver split as animal_identity_invariants_helpers.R - it is methods-provenance for the identity-correction step, not a producer of manuscript figures. || MOVE_BREAKS: Both consumers use the literal CWD-relative string source(""Functions/identity_correction_comparison_helpers.R""), so moving it one directory deeper breaks both even from the repo root."
"Functions/phase_classification_helpers.R","production","Functions/phase_classification_helpers.R","KEEP (not moved)",TRUE,"Analysis/14_systems_neuroscience_summary_dashboard.R:62 directly; and transitively by every stage that sources duration_normalization_helpers.R (12 stages) via Functions/duration_normalization_helpers.R:26, plus Functions/first_night_window_helpers.R:38 and Functions/first_night_domain_helpers.R:46","Analysis/14:62; Functions/duration_normalization_helpers.R:26,29,106; Functions/first_night_window_helpers.R:38,40; Functions/first_night_domain_helpers.R:46; Analysis/11:88, Analysis/12:63, Analysis/13:65 (comments citing the shared classifier); Testing/audit_first_night_production_parity.R:9; Testing/audit_hmm_cross_optimum_gapaware.R:25; Testing/audit_inactive_phase_qc_redesign.R:29; Testing/audit_phase_bug_impact.R:9; Testing/audit_step6_longitudinal_gapaware_robustness.R:8; Testing/test_first_night_domain_contract.R:9; Testing/test_first_night_window_parity.R:16; Testing/test_phase_classification.R:6; manuscript/Fig1_behavior_candidates/build_fig1_candidates.R:272 (comment)","no","yes","no","GROUP=temporal/window; smallest file in Functions/ (3.5 KB) but the highest-leverage scientific invariant: the single exact circadian Active/Inactive classifier that replaced the historical branch-order-dependent pattern (the bug documented in Analysis/11:88, Analysis/12:63, Analysis/13:65). It is a transitive dependency of nearly the whole pipeline. Covered by Testing/test_phase_classification.R. Any change here would alter phase assignment repo-wide - treat as frozen. || MOVE_BREAKS: source_mmm_helper() flat 'Functions/' resolution from Analysis/14 and from three sibling Functions/ helpers that source it by bare filename."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/chain_validation_social_spatial_organization.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/chain_validation_social_spatial_organization.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/chain_validation_social_spatial_organization.csv",TRUE,"Written by R/00_verify_provenance.R:110. Read by nothing.","manuscript/BehavioralDynamics_schema/README.md (lines 42-45)","no","yes","yes","Tracked frozen evidence for the exported-vs-recomputed DomainScore chain check. Preserve byte-identical."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/Fig_sis_active_inactive_domain_heatmap.png","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/Fig_sis_active_inactive_domain_heatmap.png","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/Fig_sis_active_inactive_domain_heatmap.png",TRUE,"none - not written or read by any script in the folder","manuscript/BehavioralDynamics_schema/README.md (line 4, names the panel the schema explains)","no","yes","yes","Tracked static copy of the Stage 14 publication panel the schema documents. Manually placed (no script writes it). Pure asset, no path coupling."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/figure_number_manifest.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/figure_number_manifest.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/figure_number_manifest.csv",TRUE,"Written by R/30_write_number_manifest.R:119. Read by nothing - the figure scripts hardcode their numbers.","manuscript/BehavioralDynamics_schema/README.md (lines 38, 42-45)","no","yes","yes","Tracked. The audit trail mapping every printed number in both schema figures to its source file and derivation. Contains 8 occurrences of the string BehavioralDynamics_schema (figure-name column values, not paths) - those are content, not references, and must not be rewritten on a move."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/rest_domain_collapse_evidence.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/rest_domain_collapse_evidence.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/rest_domain_collapse_evidence.csv",TRUE,"Written by R/01_verify_rest_domain_collapse.R:105. Read by nothing.","manuscript/BehavioralDynamics_schema/README.md (lines 42-45)","no","yes","yes","Tracked frozen evidence for W2 (rest domain collapses to (Movement_acf1_z - Movement_rmssd_z)/3). Preserve byte-identical."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/tile_values_social_spatial_organization.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/tile_values_social_spatial_organization.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/tile_values_social_spatial_organization.csv",TRUE,"Written by R/00_verify_provenance.R:109. Read by nothing.","manuscript/BehavioralDynamics_schema/README.md (lines 42-45)","no","yes","yes","Tracked frozen evidence (heatmap tile values). Preserve byte-identical."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/worked_example_animal1545_CC1_active.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/worked_example_animal1545_CC1_active.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/worked_example_animal1545_CC1_active.csv",TRUE,"Written by R/00_verify_provenance.R:108. Read by nothing.","manuscript/BehavioralDynamics_schema/README.md (lines 42-45)","no","yes","yes","Tracked frozen evidence. This is the only surviving repo-internal record of the worked example, since its generator depends on an ephemeral scratchpad. Preserve byte-identical."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/zero_fill_mechanism_evidence.csv","data_dependent_audit","manuscript/BehavioralDynamics_schema/data/zero_fill_mechanism_evidence.csv","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/data/zero_fill_mechanism_evidence.csv",TRUE,"Written by R/02_verify_zero_fill_mechanism.R:110. Read by nothing.","manuscript/BehavioralDynamics_schema/README.md (lines 42-45)","no","yes","yes","Tracked frozen evidence for the safe_scale zero-fill proof. Preserve byte-identical."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/00_verify_provenance.R","data_dependent_audit","manuscript/BehavioralDynamics_schema/R/00_verify_provenance.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/00_verify_provenance.R",TRUE,"Manual run only; nothing outside the folder sources it.","manuscript/BehavioralDynamics_schema/README.md (line 33); R/21_supp_warnings.R lines 8-9 (cited as the reproduction evidence)","yes","indirect","yes_with_reference_updates","No ROOT variable, but line 18 hardcodes OUT <- ""C:/Users/topohl/Documents/GitHub/MMMSociability/manuscript/BehavioralDynamics_schema/data"" (used at lines 107-111 to write the three provenance CSVs). Line 17 DATA points at a DIFFERENT Claude session's ephemeral scratchpad: C:/Users/topohl/AppData/Local/Temp/claude/c--Users-topohl-Documents-GitHub/77f6ce3c-.../scratchpad/data. That directory still exists today (6 systems_sis_*.csv, dated Sep 2) but is temp storage outside the repo and outside S:, so this script is already non-reproducible from the repo alone. Its scientific claim (recomputed DomainScore == exported, max |diff| = 0 over 882 rows) survives as the frozen CSVs in data/."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/01_verify_rest_domain_collapse.R","data_dependent_audit","manuscript/BehavioralDynamics_schema/R/01_verify_rest_domain_collapse.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/01_verify_rest_domain_collapse.R",TRUE,"Manual run only.","manuscript/BehavioralDynamics_schema/README.md (line 34); R/21_supp_warnings.R lines 8-9","yes","indirect","yes_with_reference_updates","Line 16 hardcodes the same absolute OUT (.../BehavioralDynamics_schema/data), consumed at lines 92 and 105. Line 15 DATA is the same foreign scratchpad path. Documents W1/W2: the Stage-12 substring phase parser relabelled every Inactive epoch Active. Note this finding has since been FIXED upstream (see the comment block at manuscript/Fig1_behavior_candidates/build_fig1_candidates.R:270-278), which is exactly why this whole folder reads as a pre-production audit rather than a live manuscript asset."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/02_verify_zero_fill_mechanism.R","data_dependent_audit","manuscript/BehavioralDynamics_schema/R/02_verify_zero_fill_mechanism.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/02_verify_zero_fill_mechanism.R",TRUE,"Manual run only.","manuscript/BehavioralDynamics_schema/README.md (line 35); R/21_supp_warnings.R lines 8-9","yes","indirect","yes_with_reference_updates","Line 20 hardcodes the same absolute OUT (.../BehavioralDynamics_schema/data), consumed at lines 97 and 110. Line 19 DATA is the same foreign scratchpad path. Proves the safe_scale() zero-fill mechanism (14_..._dashboard.R:349-354) turns total NA into exact 0 rather than dropping the feature."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/10_main_compact_schema.R","manuscript_builder","manuscript/BehavioralDynamics_schema/R/10_main_compact_schema.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/10_main_compact_schema.R",TRUE,"Manual/interactive run only. No runner sources it: no reference from Analysis/run_all_analysis.R, Testing/, or any other script (git grep BehavioralDynamics_schema returns hits only inside the folder plus .gitignore).","manuscript/BehavioralDynamics_schema/README.md (line 36, run-order table); .gitignore lines 3 and 12 (ignores its rendered/ output)","no","yes","yes_with_reference_updates","THE CRITICAL FINDING. Line 16 is a hardcoded absolute ROOT that embeds the folder's own location: ROOT <- ""C:/Users/topohl/Documents/GitHub/MMMSociability/manuscript/BehavioralDynamics_schema"". Not relative to the script, not env-var driven, no fallback. Line 17 source(file.path(ROOT,""R"",""schema_toolkit.R"")) and line 265 save_schema(p, file.path(ROOT,""rendered"",""BehavioralDynamics_schema_main""), W, H) both depend on it. Draws no data: every number in the figure is a hardcoded literal (grep for read.csv/read_csv in 10/20/21/schema_toolkit returns nothing), so it runs offline. Renders the 125x262mm main-paper provenance schema."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/20_supplementary_provenance_schema.R","manuscript_builder","manuscript/BehavioralDynamics_schema/R/20_supplementary_provenance_schema.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/20_supplementary_provenance_schema.R",TRUE,"Manual/interactive run only; nothing outside the folder sources it.","manuscript/BehavioralDynamics_schema/README.md (line 37); .gitignore lines 3 and 12","no","yes","yes_with_reference_updates","Same hardcoded absolute ROOT at line 16: ROOT <- ""C:/Users/topohl/Documents/GitHub/MMMSociability/manuscript/BehavioralDynamics_schema"". Consumed at line 17 (source schema_toolkit.R), line 273 (save rendered/BehavioralDynamics_schema_supplementary), line 343 (source R/21_supp_warnings.R), line 345 (re-save). Renders the A3 supplementary sheet. Line 168 cites docs/analysis_strategy_for_manuscript.md:215-238 as a text string only (no file read) - that is the manuscript language contract forbidding 'circadian/sleep disruption'."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/21_supp_warnings.R","manuscript_builder","manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/21_supp_warnings.R",TRUE,"Sourced by R/20_supplementary_provenance_schema.R:343 via file.path(ROOT, ""R"", ""21_supp_warnings.R"").","manuscript/BehavioralDynamics_schema/README.md (line 37); R/20:343","no","yes","yes","Right-column continuation fragment; operates on the caller's p / yR / XR objects, so it is not standalone-runnable. No ROOT, no path of its own. Line 76 hardcodes the claim that manuscript/Fig1_behavior_candidates/figure_manifest.csv marks active_inactive_systems as blocked_pending_phase_classifier_fix_and_rerun - true of the tracked manifest, but the build script has since moved that panel to optional_after_rerun, so this text is stale. Do not edit the wording during restructuring (that would be a scientific-content change); flag it instead."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/30_write_number_manifest.R","manuscript_builder","manuscript/BehavioralDynamics_schema/R/30_write_number_manifest.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/30_write_number_manifest.R",TRUE,"Manual run only; nothing outside the folder sources it.","manuscript/BehavioralDynamics_schema/README.md (line 38)","no","yes","yes_with_reference_updates","Third copy of the hardcoded absolute ROOT, line 10. Writes data/figure_number_manifest.csv via lines 118-119 (file.path(ROOT,""data"",...)). Self-contained: the manifest body is a hardcoded tibble::tribble, no file is read. Line 11 assigns DATA to a foreign Claude scratchpad path but that variable is NEVER USED (only assignment, no consumer). Line 12 SRC = ""S:/.../analysis_ready/12_systems_neuroscience_summary/5min_based"" is used only at line 113 as a provenance label string, never as a read path - so no S: drive access at runtime."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/schema_toolkit.R","manuscript_builder","manuscript/BehavioralDynamics_schema/R/schema_toolkit.R","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/R/schema_toolkit.R",TRUE,"Sourced by R/10_main_compact_schema.R:17 and R/20_supplementary_provenance_schema.R:17, both via file.path(ROOT, ""R"", ""schema_toolkit.R"").","manuscript/BehavioralDynamics_schema/README.md (line 32); R/10:17; R/20:17","no","yes","yes","Pure function library: no ROOT, no setwd, no Sys.getenv, no file read. Defines SCHEMA_PAL (palette copied from Functions/behavioral_dynamics_helpers.R:66), schema_canvas/sbox/stext/scode primitives, mm text metrics (char_mm/fits/wrp), and save_schema (line 209) which dir.create()s dirname(base) so it writes wherever its caller points it. Moves cleanly with the folder; it only becomes unreachable if 10/20's ROOT strings are not updated in the same commit."
"manuscript/archive/BehavioralDynamics_schema_preproduction_audit/README.md","manuscript_builder","manuscript/BehavioralDynamics_schema/README.md","manuscript/archive/BehavioralDynamics_schema_preproduction_audit/README.md",TRUE,"Documentation; not executed.","none (nothing links to it)","no","yes","yes","Run-order table for R/00 -> R/30, figure spec table, the W1-W9 verified-findings list, and the recommended manuscript inference (DomainScore ~ Group*Sex + CageChange + (1|Animal)). Contains no absolute paths, so it moves without edit. Two internal cross-refs will read as stale after a move: line 81 cites docs/analysis_strategy_for_manuscript.md:215-238 (still valid, that file stays put) and lines 89-91 cite manuscript/Fig1_behavior_candidates/figure_manifest.csv with the old blocked_pending_phase_classifier_fix_and_rerun status. Line 28 says 'Run in order from this directory' but the scripts actually ignore CWD entirely because ROOT is absolute."
"manuscript/Fig1_behavior_candidates/build_fig1_candidates.R","manuscript_builder","manuscript/Fig1_behavior_candidates/build_fig1_candidates.R","KEEP (not moved)",TRUE,"Run manually from the repo root as source(""manuscript/Fig1_behavior_candidates/build_fig1_candidates.R"") per its README line 27. Analysis/14_systems_neuroscience_summary_dashboard.R:6053-6054 pins the Fig_sis_CC1_first_active_domain_heatmap filename specifically so this staging script keeps resolving it.","manuscript/Fig1_behavior_candidates/README.md:27; .gitignore line 2; Analysis/14_systems_neuroscience_summary_dashboard.R:6054 (by folder name)","yes","yes","yes_with_reference_updates","ROOT RESOLUTION (robust, unlike the schema): lines 26-37. script_path from sys.frame(1)$ofile; script_dir falls back to getwd(); repo_candidates = c(normalizePath(file.path(script_dir,"".."","".."")), normalizePath(getwd())); line 35 picks the first candidate where Analysis/_pipeline_setup.R exists; line 37 sources it. The ../.. term is depth-2 dependent but the getwd() term rescues it whenever the CWD is the repo root, so a move would still resolve. HOWEVER line 49 hardcodes out_root <- file.path(repo_root, ""manuscript"", ""Fig1_behavior_candidates"", ""rendered""), so a moved script would silently keep writing to the OLD folder - that line needs updating if it ever moves. READS LOCAL S: DATA: yes. project_root = Sys.getenv(""MMM_BEHAVIOR_PROJECT_ROOT"", unset = ""S:/Lab_Member/Tobi/Experiments/Exp9_Social-Stress/Analysis/Behavior/RFID"") at lines 39-42; then stage09/03/14/16 dirs at lines 44-47 (via behavior_stage_dir / behavior_manuscript_dir from _pipeline_setup.R). Lines 116-124 hard-stop if the four canonical Stage 09/16 inputs are missing, so it cannot run at all off-site. STATUS VOCABULARY emitted by record_status() into rendered/staging_status.csv: generated_from_stage16 (lines 178, 248), copied_svg_pdf (97), copied_partial (97), missing_upstream (101), optional_after_rerun (268, 269, 279, 280), blocked_pending_inactive_qc_audit (281). This is an ACTIVE manuscript builder - keep in place."
"manuscript/Fig1_behavior_candidates/figure_manifest.csv","manuscript_builder","manuscript/Fig1_behavior_candidates/figure_manifest.csv","KEEP (not moved)",TRUE,"none at runtime - build_fig1_candidates.R never reads it; the panel list is duplicated as literals in the script (tribble at lines 266-282 plus the copy_pair calls).","manuscript/Fig1_behavior_candidates/README.md:20; manuscript/BehavioralDynamics_schema/README.md:89-91; manuscript/BehavioralDynamics_schema/R/21_supp_warnings.R:76","no","yes","yes","Tracked declarative 11-row panel roster. STATUS VOCABULARY in its default_status column: core (rows 1-6), optional_after_rerun (rows 7, 10), blocked_pending_phase_classifier_fix_and_rerun (rows 8, 9, 11). NOTE A PRE-EXISTING DIVERGENCE (do not fix during restructuring, just record it): the script now assigns optional_after_rerun to repeated_adaptation and active_inactive_systems and blocked_pending_inactive_qc_audit to rest_state_architecture, because the phase-classifier gate was cleared (script comment lines 270-278). The manifest still carries the old blocked_pending_phase_classifier_fix_and_rerun for all three, and two other docs quote that stale value."
"manuscript/Fig1_behavior_candidates/README.md","manuscript_builder","manuscript/Fig1_behavior_candidates/README.md","KEEP (not moved)",TRUE,"Documentation; not executed.","none","no","yes","yes","Documents the six-panel story, the repo-root invocation contract (line 27), the S: default project root (line 33), the MMM_BEHAVIOR_PROJECT_ROOT override (line 39) and the rendered/core output location (line 45). All paths are repo-relative or the data root - none would need rewriting. Line 20 still describes the Stage 11/12 permissive active|dark|night classifier as unfixed, which the build script's own comment (lines 270-278) says has since been corrected - a stale-doc note, not a move blocker."
"README.md","production","README.md","KEEP (not moved)",TRUE,"Repository entry point; rendered by GitHub. Not executed.","docs/analysis_strategy_for_manuscript.md (lists README.md among the documents to read together)","no","indirect","no","Top-level project documentation. Its Repository Structure block (lines 62-70) lists Analysis/, Functions/, Formatting/, Testing/, docs/, README.md - it does NOT mention manuscript/ or analysis_ready/ at all, so the restructure should add manuscript/ there. It references Analysis/_pipeline_setup.R, Analysis/run_all_analysis.R, Analysis/README_pipeline.md and Analysis/_archive/ by path; if any of those move the README needs the same-commit update. Must stay at the repo root."
"Testing/audits/audit_first_night_candidate_set_algebra_crosscheck.R","data_dependent_audit","Testing/audit_first_night_candidate_set_algebra_crosscheck.R","Testing/audits/audit_first_night_candidate_set_algebra_crosscheck.R",TRUE,"none - one-off orchestrator cross-check, run manually as `Rscript Testing/audit_first_night_candidate_set_algebra_crosscheck.R` from the repo root; no scheduler, no pipeline stage, no test runner sources it","none","yes","indirect","yes","Repo-root resolution: line 4 `setwd(""C:/Users/topohl/Documents/GitHub/MMMSociability"")` (absolute) then line 5 `source(""Analysis/_pipeline_setup.R"")` (CWD-relative, evaluated AFTER the setwd) => fully depth-independent and invocation-CWD-independent. No `--file=`/`dirname()` logic, no `THIS_SCRIPT` literal. READS S: (analysis_ready/03_derived_metrics/5min_based/all_behavior_metrics.csv). WRITES NOTHING - console output only (no write_csv/ggsave/writeLines anywhere). Independent from-scratch re-derivation of the nine first-night z-features and the 8x9 formula-coefficient matrix; it cross-checks the three algebraic identities and the formula-space geometry asserted by Testing/audit_first_night_candidate_set_scores.R. Not superseded (it is the independent check, not the producer), but it is a satellite of the candidate-set chain whose conclusions are now frozen in Functions/first_night_domain_helpers.R."
"Testing/audits/audit_first_night_candidate_set_decision.R","data_dependent_audit","Testing/audit_first_night_candidate_set_decision.R","Testing/audits/audit_first_night_candidate_set_decision.R",TRUE,"none - terminal step of the manual candidate-set chain (scores -> effects -> decision); nothing sources or schedules it","none","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 50) + CWD-relative source of Analysis/_pipeline_setup.R (line 51) => depth-independent, SAFE. READS S: only via the audit dir (OUT/first_night_10domain_scores.csv, _formula_overlap, _redundancy, _locomotion_dominance from candidate_set_scores.R; _effects.csv and first_night_group_sex_interactions.csv from candidate_set_effects.R). WRITES TO S:: first_night_final_domain_decision.csv, first_night_candidate_set_decision_assertions.csv, first_night_candidate_set_readme.md, and two candidate figures Fig_first_night_minimal_heatmap.{svg,pdf} + Fig_first_night_10domain_audit_heatmap.{svg,pdf}, all under <PROJ>/analysis_ready/12_systems_neuroscience_summary/5min_based/audit_hmm_state_architecture/first_night_domain_heatmap/. Hard `stopifnot(dir.exists(OUT))` at line ~48 => it cannot run unless the upstream chain already populated that S: directory. Reference update needed on move: line 60 `THIS_SCRIPT <- ""Testing/audit_first_night_candidate_set_decision.R""` is stamped into the decision table, the assertions table and the README (lines 391, 570, 919); leaving it stale would mislabel provenance, updating it changes only a metadata string, never a number. SUPERSESSION: this is the terminal decision record that supersedes the 5-domain sets of audit_first_night_domain_scores_v2.R / audit_first_night_heatmap_v2.R; it is itself operationally superseded by the productionized Functions/first_night_domain_helpers.R + Functions/first_night_domain_driver.R (driven by Analysis/14) and locked by Testing/test_first_night_domain_contract.R. Keep as the written justification for the displayed row set. || MOVE_BREAKS: nothing functionally; only the embedded provenance literal THIS_SCRIPT (line 60) becomes a stale path inside re-run output tables and the README"
"Testing/audits/audit_first_night_candidate_set_effects.R","data_dependent_audit","Testing/audit_first_night_candidate_set_effects.R","Testing/audits/audit_first_night_candidate_set_effects.R",TRUE,"none - middle step of the manual chain; its S: CSV outputs (first_night_10domain_effects.csv, first_night_group_sex_interactions.csv) are read by Testing/audit_first_night_candidate_set_decision.R, but no script sources this file","none","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 49) + CWD-relative source (line 50) => depth-independent, SAFE. READS S:: OUT/first_night_10domain_scores.csv (from candidate_set_scores.R) plus DERIV/5min_based/all_behavior_metrics.csv for the roster. WRITES TO S: (7 tables into the same first_night_domain_heatmap audit dir): first_night_10domain_effects.csv, first_night_group_sex_interactions.csv, first_night_multiplicity_sensitivity.csv, first_night_volatility_multiplicity_focus.csv, first_night_10domain_resolution_agreement.csv, first_night_10domain_locomotion_dominance.csv, first_night_candidate_set_effects_assertions.csv. `stopifnot(dir.exists(OUT))` => requires the upstream run. Reference update on move: line 60 THIS_SCRIPT and line 61 `UPSTREAM <- ""Testing/audit_first_night_candidate_set_scores.R""` are written into the source_script/script provenance columns of every output table (lines 241-242, 378-379, 503-504, 618, 682, 718); if candidate_set_scores.R also moves, UPSTREAM must move with it. SUPERSESSION: extends and supersedes the contrast tables of audit_first_night_domain_scores_v2.R / audit_first_night_heatmap_v2.R from 5 displayed domains to the 10-candidate set; itself superseded operationally by Functions/first_night_domain_driver.R (Stage 14 contrasts/interactions). || MOVE_BREAKS: nothing functionally; the THIS_SCRIPT/UPSTREAM path literals (lines 60-61) become stale provenance strings in re-run outputs"
"Testing/audits/audit_first_night_candidate_set_scores.R","data_dependent_audit","Testing/audit_first_night_candidate_set_scores.R","Testing/audits/audit_first_night_candidate_set_scores.R",TRUE,"none - first step of the manual chain; its S: output first_night_10domain_scores.csv is consumed by audit_first_night_candidate_set_effects.R and audit_first_night_candidate_set_decision.R, and re-derived independently by audit_first_night_candidate_set_algebra_crosscheck.R","Testing/audit_first_night_candidate_set_effects.R (lines 11, 61)","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 50) + CWD-relative source (line 51) => depth-independent, SAFE. READS S:: 03_derived_metrics/{10min_based,5min_based}/all_behavior_metrics.csv and 06_behavioral_dynamics/hmm_states. WRITES TO S:: first_night_10domain_scores.csv, first_night_10domain_formula_overlap.csv, first_night_10domain_redundancy.csv (dir.create, so it can bootstrap the audit dir). Reference update on move: line 62 THIS_SCRIPT stamped into the scores table (line 428); one referencing file (candidate_set_effects.R lines 11 and 61) hard-codes this path as UPSTREAM and must be updated in lockstep. SUPERSESSION: header states it reuses the window/phase/feature/standardization contract VERBATIM from audit_first_night_domain_scores_v2.R and extends the row set from 5 to 10 candidates, so it supersedes v2 for the candidate-set question; all of it is now productionized in Functions/first_night_domain_helpers.R. || MOVE_BREAKS: nothing functionally; the THIS_SCRIPT literal (line 62) and the UPSTREAM literal in candidate_set_effects.R:61 become stale unless both are updated together"
"Testing/audits/audit_first_night_domain_scores.R","data_dependent_audit","Testing/audit_first_night_domain_scores.R","Testing/audits/audit_first_night_domain_scores.R",TRUE,"none - superseded v1 audit; no script sources or schedules it and its S: outputs are overwritten by the v2 script","Testing/audit_first_night_domain_scores_v2.R (lines 4, 782, 1124)","yes","indirect","yes_with_reference_updates","SUPERSEDED (do not delete - it is the documented defect baseline). audit_first_night_domain_scores_v2.R:3-4 states verbatim 'v2, SUPERSEDES Testing/audit_first_night_domain_scores.R'; v1 built the raw domains on the Stage 14 `local_bin <= 12h/bin` COUNT rule, which matches the canonical clock window for only 50/111 animals at 10 min and 33/111 at 5 min. v2 line 1124 keeps a reproduction of the v1 rule purely to quantify the defect. Repo-root resolution: absolute setwd (line 29) + CWD-relative source (line 30) => depth-independent, SAFE. READS S:: 03_derived_metrics/5min_based/all_behavior_metrics.csv, 16_sleep_like_inactivity_metrics/.../sleep_like_inactivity_features.csv, 06_behavioral_dynamics/hmm_states, plus the CombZ workbook S:/.../SIS_Analysis/E9_Behavior_Data.xlsx. WRITES TO S: (~16 tables) into audit_hmm_state_architecture/first_night_domain_heatmap/, and CRITICALLY it writes the SAME basenames as v2 (first_night_domain_scores.csv, first_night_domain_provenance.csv, first_night_domain_redundancy_audit.csv, first_night_group_sex_n.csv) - so re-running v1 after v2 would silently overwrite the current v2 artifacts with the superseded count-rule values. Flag that collision when placing it. Reference update on move: line 47 THIS_SCRIPT literal, plus the three citing lines in v2. || MOVE_BREAKS: nothing functionally; the THIS_SCRIPT literal (line 47) and v2's three citation strings become stale paths"
"Testing/audits/audit_first_night_domain_scores_v2.R","data_dependent_audit","Testing/audit_first_night_domain_scores_v2.R","Testing/audits/audit_first_night_domain_scores_v2.R",TRUE,"none - manual audit; its S: outputs (first_night_domain_scores.csv, first_night_hmm_component_features_v2.csv, first_night_hmm_state_semantics_v2.csv, first_night_domain_contrasts_v2.csv) are consumed by Testing/audit_first_night_heatmap_v2.R","Testing/audit_first_night_heatmap_v2.R (lines 6, 67, 148); Testing/audit_first_night_candidate_set_scores.R (lines 11, 397)","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 44) + CWD-relative source (line 45) => depth-independent, SAFE. READS S:: 03_derived_metrics/{10min,5min}_based/all_behavior_metrics.csv and 06_behavioral_dynamics/hmm_states. WRITES TO S: (~16 tables) into audit_hmm_state_architecture/first_night_domain_heatmap/. SUPERSESSION: supersedes audit_first_night_domain_scores.R (the local_bin<=12h count rule -> canonical clock window 18:30->06:30); itself extended/superseded by the candidate-set chain (candidate_set_scores.R reuses its code verbatim and grows 5 -> 10 domains) and finally productionized as Functions/first_night_window_helpers.R + Functions/first_night_domain_helpers.R, consumed by Analysis/14_systems_neuroscience_summary_dashboard.R and locked by Testing/test_first_night_window_parity.R / Testing/test_first_night_domain_contract.R. Reference updates on move: line 56 THIS_SCRIPT, line 782 `supersedes` literal, and the two downstream citers above (heatmap_v2.R:67 UPSTREAM is load-bearing for provenance, heatmap_v2.R:148 only regex-matches the basename 'audit_first_night_domain_scores_v2' inside a scores column so it survives a directory change). || MOVE_BREAKS: nothing functionally; THIS_SCRIPT (line 56), the supersedes literal (line 782) and heatmap_v2.R:67 UPSTREAM become stale path strings"
"Testing/audits/audit_first_night_dwell_partition_stability.R","data_dependent_audit","Testing/audit_first_night_dwell_partition_stability.R","Testing/audits/audit_first_night_dwell_partition_stability.R",TRUE,"none - manual prerequisite run; its S: output first_night_dwell_partition_stability_values.csv is read by Testing/audit_first_night_dwell_shipped_vs_refit.R","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 7) + CWD-relative source (line 8) => depth-independent, SAFE. No THIS_SCRIPT/self-path literal. READS S:: 03_derived_metrics/{5min,10min}_based/all_behavior_metrics.csv. WRITES TO S:: first_night_dwell_partition_stability_values.csv and first_night_dwell_partition_stability_contrasts.csv under audit_hmm_state_architecture/first_night_domain_heatmap/. Heavy: refits the 10-min depmixS4 HMM at 5 seeds (7, 1, 101, 23, 313) spanning the 5 distinct optima and defines a custom S4 class HMMRegularizedNORMresponse - long runtime, and it needs depmixS4 installed. Not superseded; it is the construct-stability gate that justified displaying mean_dwell_minutes (and excluding top_proximity_state_fraction). Its stability CSV is a hard input for dwell_shipped_vs_refit.R, so the two should be relocated together."
"Testing/audits/audit_first_night_dwell_shipped_vs_refit.R","data_dependent_audit","Testing/audit_first_night_dwell_shipped_vs_refit.R","Testing/audits/audit_first_night_dwell_shipped_vs_refit.R",TRUE,"none - manual reconciliation run, downstream of audit_first_night_dwell_partition_stability.R via the S: CSV it produces","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 5) + CWD-relative source (line 6) => depth-independent, SAFE. No self-path literal. READS S:: 03_derived_metrics/{5min,10min}_based/all_behavior_metrics.csv, 06_behavioral_dynamics/hmm_states/10min_based/tables/hmm_state_assignments.csv, and OUT/first_night_dwell_partition_stability_values.csv (hard dependency - will error if the partition-stability script has not been run). WRITES TO S: one table (fn_ship joined with the seed-7 refit) into the same first_night_domain_heatmap audit dir. Resolves the Female RES-CON sign discrepancy (+0.059 shipped vs -0.815 refit) and establishes that the SHIPPED Stage 08 Viterbi labels are the production-relevant version - a scientifically load-bearing conclusion for the displayed persistence row. Not superseded."
"Testing/audits/audit_first_night_heatmap_v2.R","data_dependent_audit","Testing/audit_first_night_heatmap_v2.R","Testing/audits/audit_first_night_heatmap_v2.R",TRUE,"none - manual terminal run of the v2 branch; consumes the S: tables written by audit_first_night_domain_scores_v2.R","none","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 52) + CWD-relative source (line 53) => depth-independent, SAFE. READS S:: OUT/first_night_domain_scores.csv, first_night_hmm_component_features_v2.csv, first_night_hmm_state_semantics_v2.csv, first_night_domain_contrasts_v2.csv, DERIV/5min_based/all_behavior_metrics.csv, the production CC1 table, and the CombZ workbook E9_Behavior_Data.xlsx. WRITES TO S: 8 CSVs, a README (first_night_heatmap_readme.md) and TWO figure pairs Fig_first_night_domain_heatmap.{svg,pdf} + Fig_first_night_hmm_components.{svg,pdf} - audit basenames only; the production panel is Fig_sis_CC1_first_active_domain_heatmap.svg / Fig_sis_first_night_domain_heatmap_5min_sensitivity from Analysis/14, so no production figure is touched. Has a hard `stopifnot(all(file.exists(figs)), all(file.size(figs) > 5000))` self-check at line 1200. SUPERSESSION: line 503 declares it supersedes the earlier first_night_hmm_component_results.csv written by Testing/audit_first_night_hmm_components.R (same basename, overwritten); it is in turn superseded for the row set by the candidate-set decision chain and operationally by Analysis/14 + Functions/first_night_domain_driver.R. Reference updates on move: line 66 THIS_SCRIPT, line 67 UPSTREAM, line 503 supersedes literal (all provenance strings, no numbers). || MOVE_BREAKS: nothing functionally; THIS_SCRIPT/UPSTREAM/supersedes literals (lines 66, 67, 503) become stale paths in re-run outputs and the README"
"Testing/audits/audit_first_night_hmm_components.R","data_dependent_audit","Testing/audit_first_night_hmm_components.R","Testing/audits/audit_first_night_hmm_components.R",TRUE,"none - superseded manual scan; nothing sources it","Testing/audit_first_night_heatmap_v2.R (line 503)","yes","indirect","yes","SUPERSEDED (keep, do not delete). audit_first_night_heatmap_v2.R:503-505 states it 'wrote an earlier' version of first_night_hmm_component_results.csv and that the v2 run supersedes it and adds the resolution-agreement columns. Repo-root resolution: absolute setwd (line 38) + CWD-relative source (line 39) => depth-independent, SAFE. No THIS_SCRIPT literal. READS S:: hmm_states assignments, 03_derived_metrics, and the CombZ workbook E9_Behavior_Data.xlsx. WRITES TO S: ~13 CSVs plus Fig_first_night_hmm_components.{svg,pdf} into audit_hmm_state_architecture/first_night_domain_heatmap/ - NOTE the basename collision: first_night_hmm_component_results.csv, first_night_hmm_component_interactions.csv and Fig_first_night_hmm_components.* are all rewritten by audit_first_night_heatmap_v2.R, so re-running this older script after v2 would regress those artifacts. Flag that collision when placing it. It also uses a different window definition (first contiguous block of CC1 Active detected from TimeIndex runs) than the later canonical 18:30->06:30 clock anchor, which is exactly why it was superseded."
"Testing/audits/audit_first_night_production_parity.R","data_dependent_audit","Testing/audit_first_night_production_parity.R","Testing/audits/audit_first_night_production_parity.R",TRUE,"none scheduled - but it is the parity harness for the PRODUCTION code path: it sources Functions/first_night_window_helpers.R, first_night_domain_helpers.R and first_night_domain_driver.R and calls build_first_night_domain_analysis(), the same entry point Analysis/14_systems_neuroscience_summary_dashboard.R:5619-5628 uses","none","yes","yes","yes","Highest manuscript relevance of this family: it runs the SHIPPED production first-night builder standalone and proves numerical parity (item J) against the audit table OUT/first_night_10domain_scores.csv, i.e. it is the bridge that certifies the productionized Functions/first_night_domain_driver.R reproduces the audited numbers. Repo-root resolution: absolute setwd (line 5) + CWD-relative source (line 6) => depth-independent, SAFE. No THIS_SCRIPT literal. READS S:: 03_derived_metrics/5min_based/all_behavior_metrics.csv (roster) plus whatever the driver reads, and the S: audit dir for the parity comparison. DOES NOT WRITE TO S: - line 16 OUT points at a hard-coded scratchpad path 'C:/Users/topohl/AppData/Local/Temp/claude/c--Users-topohl-Documents-GitHub/2603874b-c21a-494f-aefd-10f961b8053d/scratchpad/orch/first_night_prod' carrying a STALE session UUID from a previous session; that is a scratch output dir, not a repo-identity reference, but it is machine- and session-specific and worth normalizing separately (out of scope for a pure move). Complements the portable test Testing/test_first_night_domain_contract.R, which covers the same driver without S: data."
"Testing/audits/audit_first_night_time_anchor.R","data_dependent_audit","Testing/audit_first_night_time_anchor.R","Testing/audits/audit_first_night_time_anchor.R",TRUE,"none - manual forensic run; nothing sources or schedules it","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 9) + CWD-relative source (line 10) => depth-independent, SAFE. No THIS_SCRIPT literal (line 137 only cites Analysis/09 as a descriptive string, it does not read that file). READS S:: 03_derived_metrics/{10min,5min}_based/all_behavior_metrics.csv. WRITES TO S:: first_night_time_anchor_audit.csv, first_night_time_anchor_audit_long.csv, first_night_window_contract.csv under audit_hmm_state_architecture/first_night_domain_heatmap/ (dir.create, self-bootstrapping). This is the script that derived the canonical 18:30->06:30 clock anchor FROM CODE (Analysis/09_early_prediction_model_ladder.R::select_primary_active_window) and documented that the on-disk Stage 09 artifacts are stale. Operationally superseded by the shared selector Functions/first_night_window_helpers.R and its portable test Testing/test_first_night_window_parity.R, but retain it as the derivation record of the window contract."
"Testing/audits/audit_first_night_window_provenance.R","data_dependent_audit","Testing/audit_first_night_window_provenance.R","Testing/audits/audit_first_night_window_provenance.R",TRUE,"none - manual 'Phase B prerequisite' probe; nothing sources it","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 5) + CWD-relative source (line 6) => depth-independent, SAFE. No self-path literal. READS S:: hmm_states/{10min,5min}_based/tables/hmm_state_assignments.csv and 12_systems_neuroscience_summary/5min_based/tables/systems_sis_domain_scores.csv. WRITES NOTHING - console output only. Smallest and earliest member of the family: it exposed (deliberately did not harmonize) the mismatch between the raw 12 h first-night window and the HMM CC1 Active epoch that spans several dark blocks, and the duplicate-row check between 'Early adaptation / prediction' and 'Active-phase adaptation/exploration'. Superseded in substance by audit_first_night_time_anchor.R (which pinned the clock anchor) and by audit_first_night_window_sensitivity.R (which quantified the 12 h vs ~48 h comparison); keep as the diagnostic that motivated both."
"Testing/audits/audit_first_night_window_sensitivity.R","data_dependent_audit","Testing/audit_first_night_window_sensitivity.R","Testing/audits/audit_first_night_window_sensitivity.R",TRUE,"none - manual sensitivity run; nothing sources it","none","yes","indirect","yes_with_reference_updates","Repo-root resolution: absolute setwd (line 49) + CWD-relative source (line 50) => depth-independent, SAFE. READS S:: 03_derived_metrics, 06_behavioral_dynamics/hmm_states. WRITES TO S: 4 CSVs (first_night_vs_all_cc1_active_epoch_geometry.csv, the correlation table, first_night_vs_all_cc1_active_sensitivity.csv, first_night_vs_all_cc1_active_assertions.csv) into audit_hmm_state_architecture/first_night_domain_heatmap/; no basename collisions with other family members. Scientifically load-bearing: it compares the canonical 12 h window against the full ~48 h CC1 Active epoch under two temporal-metric variants (bridging_stage08_current vs gap_aware) and explicitly states the manuscript panel keeps the 12 h window REGARDLESS of which produces larger contrasts - i.e. it documents that no window was chosen on p-values. Not superseded. Reference update on move: line 61 THIS_SCRIPT is stamped into the sensitivity table (line 593). || MOVE_BREAKS: nothing functionally; the THIS_SCRIPT literal (line 61) becomes a stale provenance string in re-run outputs"
"Testing/audits/audit_hmm_cross_optimum_gapaware.R","data_dependent_audit","Testing/audit_hmm_cross_optimum_gapaware.R","Testing/audits/audit_hmm_cross_optimum_gapaware.R",TRUE,"none (manual: Rscript Testing/audit_hmm_cross_optimum_gapaware.R from repo root)","none","yes","indirect","yes","Newest file in the set and the subject of HEAD commit 3f6f6a6 'Close the HMM cross-optimum robustness gate under the gap-aware contract' - the most directly manuscript-gating audit here, though still not read by any manuscript/ builder. Repo root via hardcoded absolute setwd at L23 + source('Analysis/_pipeline_setup.R') at L24, then source_mmm_helper('phase_classification_helpers.R') and ('hmm_stage14_helpers.R') at L25-26 - depth-independent. NOT CHAINED: reads only Stage 01 5min/10min all_behavior_metrics.csv under PROJ (L28, L72, L75) and refits depmixS4 at SEEDS c(1,11,101,313,57) spanning all four gap-aware optima (L32) - so it is self-contained but EXPENSIVE (five HMM refits); avoid re-running it merely to validate the move. Writes hmm_cross_optimum_gapaware_state_profiles.csv (L161), _contrasts.csv (L162), _sign_stability.csv (L176), _claim_verdicts.csv (L189)."
"Testing/audits/audit_hmm_state_architecture_component_models.R","data_dependent_audit","Testing/audit_hmm_state_architecture_component_models.R","Testing/audits/audit_hmm_state_architecture_component_models.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_component_models.R from repo root)","none","yes","indirect","yes","CHAINED CONSUMER of components.R. Repo root via repo_candidates at L66-72 with an explicit stop() at L72 if none resolve - depth-independent. HARD chain: L120-124 read hmm_architecture_component_epoch_metrics.csv from audit_out and stop() with the message 'Run Testing/audit_hmm_state_architecture_components.R first' if missing; that L123 string is a stale path once components.R moves (cosmetic, but it is the operator's instruction so it should be retargeted to Testing/audits/...). project_root hardcoded absolute at L76; also reads the Stage 01 canonical roster (L95-113, stop() if missing) and asserts nrow(canonical_roster)==111 at L114. Writes hmm_architecture_component_leaveout_sensitivity.csv (L680-681). L88 prints MMM_REPO_ROOT for provenance only."
"Testing/audits/audit_hmm_state_architecture_components.R","data_dependent_audit","Testing/audit_hmm_state_architecture_components.R","Testing/audits/audit_hmm_state_architecture_components.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_components.R from repo root)","Testing/audit_hmm_state_architecture_component_models.R (L43 comment, L123 stop() message); Testing/audit_hmm_state_architecture_redundancy.R (L31 comment)","yes","indirect","yes_with_reference_updates","ROOT PRODUCER OF THE FOUNDATION TABLE and the single most-depended-on file in this set. Repo root via repo_candidates at L77-83: c('MMMSociability/Analysis/_pipeline_setup.R', 'Analysis/_pipeline_setup.R', 'C:/Users/topohl/.../Analysis/_pipeline_setup.R') picked by file.exists(); all three are CWD-relative or absolute, none derived from the script's own location, so depth-independent. Reads S: Stage 01 roster, Stage 08 hmm_state_summary.csv, chip-loss QC, shipped Stage 14 scores (project_root hardcoded at L87). Writes the foundation table hmm_architecture_component_epoch_metrics.csv (L671-673) plus hmm_architecture_check_j5_active_dropout_epochs.csv (L1228) and hmm_architecture_check_l2_subsampling_length_bias.csv (L1530). Reference updates are COSMETIC ONLY: the two referring files name the old path 'Testing/audit_hmm_state_architecture_components.R' in a comment and a stop() message; nothing resolves that string at runtime, but it should be retargeted to Testing/audits/... so the operator instruction stays accurate. Must be run BEFORE component_models, longitudinal, redundancy, construct_comparison, and the three phaseA audits."
"Testing/audits/audit_hmm_state_architecture_construct_comparison.R","data_dependent_audit","Testing/audit_hmm_state_architecture_construct_comparison.R","Testing/audits/audit_hmm_state_architecture_construct_comparison.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_construct_comparison.R from repo root)","none","yes","indirect","yes","DEEPEST CHAINED AUDIT: depends on two other audits plus a shipped Stage 14 table. Repo root via repo_candidates at L51-57 - depth-independent. HARD dependency at L299-300 on the SHIPPED table systems_hmm_composite_component_audit.csv (stop() if missing) - that is a pipeline artifact, not an audit output. SOFT dependency at L749-751 on hmm_architecture_component_epoch_metrics.csv (from components.R) guarded by file.exists(). SOFT dependency at L848-873 on hmm_architecture_proposed_construct_contrasts.csv (from redundancy.R), which prints an explicit '[4b1] SKIPPED' branch when absent - so a stale run silently degrades rather than erroring; worth running components.R and redundancy.R first. project_root hardcoded absolute at L61, audit_out at L63. Writes hmm_architecture_construct_agreement.csv (L743), hmm_architecture_construct_comparison.csv (L1050), hmm_architecture_construct_comparison_structural.csv (L1077) and a markdown summary (L1376). It carries stale 'Testing/audit_hmm_state_architecture_redundancy.R' path strings at L14/L410/L1027 that should be retargeted when redundancy.R moves."
"Testing/audits/audit_hmm_state_architecture_gap_aware.R","data_dependent_audit","Testing/audit_hmm_state_architecture_gap_aware.R","Testing/audits/audit_hmm_state_architecture_gap_aware.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_gap_aware.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L6 + source('Analysis/_pipeline_setup.R'); source_mmm_helper('hmm_stage14_helpers.R') at L7 - depth-independent. CHAINED CONSUMER: L60 reads hmm_architecture_temporal_epoch_metrics.csv (the 'naive' pre-gap comparison baseline from temporal_components.R), unguarded - this audit's entire purpose is the naive-vs-gap-aware delta, so running it without temporal_components.R having been run first is meaningless as well as fatal. A (audit dir) at L8 and HMM (Stage 06 hmm_states) at L9 are hardcoded absolute; also reads Stage 01 metrics inline at L41 and hmm_state_assignments.csv at L47. Writes hmm_architecture_gap_aware_contrasts.csv (L76)."
"Testing/audits/audit_hmm_state_architecture_identifiability.R","data_dependent_audit","Testing/audit_hmm_state_architecture_identifiability.R","Testing/audits/audit_hmm_state_architecture_identifiability.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_identifiability.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L6 + source('Analysis/_pipeline_setup.R') at L7 - depth-independent. NOT CHAINED: reads only Stage 01 all_behavior_metrics.csv at both bin levels (L70, L77) under PROJ (L12). OUT hardcoded absolute at L11 via getOption('mmm.audit_out', ...). Writes hmm_architecture_identifiability_probe.csv (L130). Self-contained refit audit - safe to move and independently re-runnable."
"Testing/audits/audit_hmm_state_architecture_locomotion_dominance.R","data_dependent_audit","Testing/audit_hmm_state_architecture_locomotion_dominance.R","Testing/audits/audit_hmm_state_architecture_locomotion_dominance.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_locomotion_dominance.R; note it does NOT need CWD=repo root)","none","yes","indirect","yes","THE ONLY FILE IN THIS SET WITH NO REPO-ROOT RESOLUTION AT ALL - it never sources Analysis/_pipeline_setup.R and never calls source_mmm_helper(); the whole 58-line script is library() calls plus hardcoded absolute S: paths (OUT at L4 via getOption('mmm.audit_out', ...), B at L5). It is therefore trivially and unconditionally safe to move to any depth, and is also the only one indifferent to the invocation CWD. CHAINED CONSUMER: L7 reads hmm_architecture_temporal_epoch_metrics.csv (produced by temporal_components.R) with no file.exists() guard, so it fails with a readr error rather than a clean stop() if run out of order. Also reads the shipped systems_sis_domain_scores.csv twice (L9, L13). Writes hmm_architecture_locomotion_dominance_by_metric.csv (L41). Because it skips _pipeline_setup.R it also skips canonical_animal_id()/roster assertions that its siblings apply - a pre-existing property, not affected by the move."
"Testing/audits/audit_hmm_state_architecture_longitudinal.R","data_dependent_audit","Testing/audit_hmm_state_architecture_longitudinal.R","Testing/audits/audit_hmm_state_architecture_longitudinal.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_longitudinal.R from repo root)","none","yes","indirect","yes","CHAINED CONSUMER of components.R. Repo root via repo_candidates at L70-76 - depth-independent. HARD chain: L124 stops with 'Foundation table missing' if hmm_architecture_component_epoch_metrics.csv is absent, then reads it at L125. project_root hardcoded absolute at L80, audit_out at L81; Stage 01 roster checked at L101. Writes hmm_architecture_longitudinal_results.csv (L698-700). Uses the unchanged repo estimator Functions/hmm_stage14_helpers.R::fit_repeated_measures_domain_contrasts() (noted L550), reached via source_mmm_helper() after _pipeline_setup.R, so no path in the model layer depends on script depth."
"Testing/audits/audit_hmm_state_architecture_partition_robustness.R","data_dependent_audit","Testing/audit_hmm_state_architecture_partition_robustness.R","Testing/audits/audit_hmm_state_architecture_partition_robustness.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_partition_robustness.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L8 + source('Analysis/_pipeline_setup.R') at L9 - depth-independent. NOT CHAINED: reads only Stage 01 5min (L71) and 10min (L74) all_behavior_metrics.csv under PROJ (L13). OUT hardcoded absolute at L12 via getOption('mmm.audit_out', ...). Writes hmm_architecture_partition_robustness_contrasts.csv (L112)."
"Testing/audits/audit_hmm_state_architecture_phaseA_formula_units_qc.R","data_dependent_audit","Testing/audit_hmm_state_architecture_phaseA_formula_units_qc.R","Testing/audits/audit_hmm_state_architecture_phaseA_formula_units_qc.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_phaseA_formula_units_qc.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L3 + source('Analysis/_pipeline_setup.R'); source_mmm_helper('hmm_stage14_helpers.R') at L4 - depth-independent. CHAINED CONSUMER of components.R: L9 reads hmm_architecture_component_epoch_metrics.csv, unguarded. A/B/HMM roots hardcoded absolute at L5-L7; also reads shipped qc_chip_loss_flags.csv (L84) and hmm_sequence_quality_audit.csv (L92). Writes phaseA_issue1_formula_preservation.csv (L42), phaseA_issue2_dwell_units.csv (L81) and a QC table at L126."
"Testing/audits/audit_hmm_state_architecture_phaseA_gapaware_contrasts.R","data_dependent_audit","Testing/audit_hmm_state_architecture_phaseA_gapaware_contrasts.R","Testing/audits/audit_hmm_state_architecture_phaseA_gapaware_contrasts.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_phaseA_gapaware_contrasts.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L5 + source('Analysis/_pipeline_setup.R'); source_mmm_helper('hmm_stage14_helpers.R') at L6 - depth-independent. CHAINED CONSUMER of components.R: L8 reads hmm_architecture_component_epoch_metrics.csv, unguarded. OUT hardcoded absolute at L7. Writes phaseA_issue4_gapaware_contrast_comparison.csv (L39). NOTE a pre-existing overlap unrelated to the move: audit_hmm_state_architecture_phaseA_partition_and_gap.R writes the SAME output filename phaseA_issue4_gapaware_contrast_comparison.csv (its L180), so whichever runs last wins; keep both in the same destination directory so the collision stays as visible as it is today."
"Testing/audits/audit_hmm_state_architecture_phaseA_partition_and_gap.R","data_dependent_audit","Testing/audit_hmm_state_architecture_phaseA_partition_and_gap.R","Testing/audits/audit_hmm_state_architecture_phaseA_partition_and_gap.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_phaseA_partition_and_gap.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L5 + source('Analysis/_pipeline_setup.R'); source_mmm_helper('hmm_stage14_helpers.R') at L6 - depth-independent. PARTIALLY CHAINED: the partition half (L42-L102) reads only Stage 01 5min/10min all_behavior_metrics.csv under PROJ (L8) and refits, but the gap half at L136 reads hmm_architecture_component_epoch_metrics.csv from components.R, unguarded. OUT hardcoded absolute at L7. Writes phaseA_issue3_topproximity_state_profiles.csv (L101), phaseA_issue3_topproximity_partition_contrasts.csv (L102) and phaseA_issue4_gapaware_contrast_comparison.csv (L180 - same filename also written by phaseA_gapaware_contrasts.R L39; pre-existing, not caused by the move)."
"Testing/audits/audit_hmm_state_architecture_profile.R","data_dependent_audit","Testing/audit_hmm_state_architecture_profile.R","Testing/audits/audit_hmm_state_architecture_profile.R",TRUE,"none (manually invoked: Rscript Testing/audit_hmm_state_architecture_profile.R from repo root; no orchestrator sources it)","none","yes","indirect","yes_with_reference_updates","THE ONLY FILE IN THIS SET THAT BREAKS ON MOVE. It is also the sole file in the set using the --file= / dirname() mechanism; all 19 siblings use CWD-relative, hardcoded-absolute, or repo_candidates resolution. MINIMAL ROBUST FIX - replace lines 44-48 with a marker-walk that is depth-independent and preserves the existing NA fallback and the unchanged L50 source() call: repo_root <- local({ fallback <- ""C:/Users/topohl/Documents/GitHub/MMMSociability""; if (is.na(script_path)) return(fallback); d <- dirname(script_path); for (i in seq_len(6L)) { if (file.exists(file.path(d, ""Analysis"", ""_pipeline_setup.R""))) return(d); p <- dirname(d); if (identical(p, d)) break; d <- p }; fallback }). This walks up from the script's own directory looking for the same Analysis/_pipeline_setup.R marker that find_mmm_repo_root() uses, so it is correct at ANY nesting depth and stays CWD-independent (a property the CWD-relative siblings do not have). Lines 38-42 and 50 need no change. Scientific behavior is bit-identical: repo_root resolves to the same <repo> value it does today. A one-line alternative (replace L38-50 with source(""Analysis/_pipeline_setup.R"")) also works but would silently downgrade the script to requiring CWD=repo root, so the walk-up is preferred. Data side is unaffected: PROJECT is hardcoded absolute at L53; AUDIT_OUT at L54-57 under S:/.../12_systems_neuroscience_summary/5min_based/audit_hmm_state_architecture. Root producer, not a chained consumer: it reads only Stage 01/08 S: artifacts and writes hmm_state_multidimensional_profile_audit.csv (L638), hmm_state_emission_degeneracy_audit.csv (L639), hmm_state_binlevel_profile_audit.csv (L640); git grep confirms no other audit reads those three tables. || MOVE_BREAKS: BREAKS. Lines 38-48 derive repo_root from the --file= argument: L38-42 build script_path via commandArgs()/normalizePath(); L44-48 set repo_root <- dirname(dirname(script_path)) (L45 is the exact defect). L50 then does source(file.path(repo_root, 'Analysis', '_pipeline_setup.R')). Today script_path=<repo>/Testing/X.R so dirname(dirname()) correctly yields <repo>. After moving to Testing/audits/, script_path=<repo>/Testing/audits/X.R and dirname(dirname()) yields <repo>/Testing, so L50 sources <repo>/Testing/Analysis/_pipeline_setup.R which does not exist -> hard error before any analysis runs. Verified empirically with Rscript: file.exists() == FALSE. The L46-48 fallback does NOT rescue this, because it only fires when script_path is NA; under Rscript script_path is always populated, so the wrong two-level path is used unconditionally."
"Testing/audits/audit_hmm_state_architecture_provenance.R","data_dependent_audit","Testing/audit_hmm_state_architecture_provenance.R","Testing/audits/audit_hmm_state_architecture_provenance.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_provenance.R from repo root)","none","yes","indirect","yes","Repo root via a named absolute constant: repo <- 'C:/Users/topohl/Documents/GitHub/MMMSociability' at L25, setwd(repo) at L26, then source('Analysis/_pipeline_setup.R') at L27 and source_mmm_helper('hmm_stage14_helpers.R') at L28 - depth-independent. Note this variant hardcodes the path rather than reading Sys.getenv('MMM_REPO_DIR'), so it is absolute but not overridable. NOT CHAINED: reads Stage 08 hmm_state_summary.csv / hmm_state_occupancy.csv / hmm_state_assignments.csv (L52, L73, L76) and the Stage 01 roster (L41) under project (L30). Writes six audit_doc_v* tables: v1 (L67), v2 (L90), v3 (L124), v3b (L136), v5 (L155), v5b (L164), v4 (L186), v4b (L200), all under audit_out (L31-32)."
"Testing/audits/audit_hmm_state_architecture_provenance_addendum.R","data_dependent_audit","Testing/audit_hmm_state_architecture_provenance_addendum.R","Testing/audits/audit_hmm_state_architecture_provenance_addendum.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_provenance_addendum.R from repo root)","none","yes","indirect","yes","Companion to provenance.R and uses the identical mechanism: repo <- 'C:/Users/topohl/Documents/GitHub/MMMSociability' at L14, setwd(repo) at L15, source('Analysis/_pipeline_setup.R') at L16, source_mmm_helper('hmm_stage14_helpers.R') at L17 - depth-independent. Despite the name it is NOT chained on provenance.R's output tables: it independently re-reads Stage 08 hmm_state_summary.csv and hmm_state_occupancy.csv (L39, L42) and the Stage 01 roster (L29) under project (L19), so it can be run standalone. Writes audit_doc_v6_sex_and_cc_terms.csv (L81), audit_doc_v7_artifact_consumption_map.csv (L107) and audit_doc_v7b_stage08_files_present.csv (L113). Move it together with provenance.R to keep the v1-v7b table family co-located."
"Testing/audits/audit_hmm_state_architecture_qc_sensitivity.R","data_dependent_audit","Testing/audit_hmm_state_architecture_qc_sensitivity.R","Testing/audits/audit_hmm_state_architecture_qc_sensitivity.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_qc_sensitivity.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L6 + source('Analysis/_pipeline_setup.R') and source_mmm_helper('hmm_stage14_helpers.R') at L7 - depth-independent. CHAINED CONSUMER: L25 reads hmm_architecture_temporal_epoch_metrics.csv (from temporal_components.R), unguarded. Also reads the shipped qc_chip_loss_flags.csv at L11. B hardcoded absolute at L8, A derived from B at L9. Writes hmm_architecture_qc_leaveout_sensitivity.csv (L60)."
"Testing/audits/audit_hmm_state_architecture_redundancy.R","data_dependent_audit","Testing/audit_hmm_state_architecture_redundancy.R","Testing/audits/audit_hmm_state_architecture_redundancy.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_redundancy.R from repo root)","Testing/audit_hmm_state_architecture_construct_comparison.R (L14 comment, L410 report text, L1027 provenance string)","yes","indirect","yes_with_reference_updates","CHAINED CONSUMER AND PRODUCER. Repo root via repo_candidates at L58-64 (same CWD-relative/absolute triple as components.R) - depth-independent. HARD chain dependency: L100-101 read hmm_architecture_component_epoch_metrics.csv from audit_out and stop() if absent, so components.R must run first. SOFT chain: L639-642 optionally read hmm_architecture_check_l2_subsampling_length_bias.csv (also produced by components.R) under file.exists(). Produces hmm_architecture_proposed_construct_definition.csv (L996), hmm_architecture_proposed_construct_contrasts.csv (L1058) and hmm_architecture_proposed_construct_readdensity_sensitivity.csv (L1113); the contrasts table is in turn consumed by construct_comparison.R. project_root hardcoded absolute at L68. Reference updates COSMETIC ONLY: construct_comparison cites the old path in a comment, a rendered report line, and a candidate_C_provenance string - the last one is embedded in a written CSV, so leaving it stale would put an incorrect path into an output artifact; updating it changes only a provenance label, not any number."
"Testing/audits/audit_hmm_state_architecture_semantic_erasure.R","data_dependent_audit","Testing/audit_hmm_state_architecture_semantic_erasure.R","Testing/audits/audit_hmm_state_architecture_semantic_erasure.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_semantic_erasure.R from repo root)","none","yes","indirect","yes","Repo root via hardcoded absolute setwd at L8 + source('Analysis/_pipeline_setup.R') at L9 and source_mmm_helper('hmm_stage14_helpers.R') at L10 - depth-independent. NOT CHAINED on audit tables: reads the Stage 01 roster (L18, with stopifnot(nrow(roster)==111) at L22) and Stage 08 hmm_state_occupancy.csv / hmm_state_summary.csv (L26, L29) under HMM derived from PROJ (L12-13). Output directory is resolved inline inside the write call at L88 via getOption('mmm.audit_out', <absolute S: default>) - unusual style but still fully absolute and depth-independent. Writes hmm_architecture_per_state_occupancy_contrasts.csv (L88)."
"Testing/audits/audit_hmm_state_architecture_temporal_components.R","data_dependent_audit","Testing/audit_hmm_state_architecture_temporal_components.R","Testing/audits/audit_hmm_state_architecture_temporal_components.R",TRUE,"none (manual: Rscript Testing/audit_hmm_state_architecture_temporal_components.R from repo root)","Testing/audit_first_night_domain_scores.R (L90 comment); Testing/audit_first_night_domain_scores_v2.R (L110 comment)","yes","indirect","yes_with_reference_updates","SECOND ROOT PRODUCER: writes hmm_architecture_temporal_epoch_metrics.csv (L94) and hmm_architecture_temporal_component_contrasts.csv (L93), consumed downstream by locomotion_dominance, qc_sensitivity and gap_aware. Repo root via hardcoded absolute setwd('C:/Users/topohl/Documents/GitHub/MMMSociability') at L6 followed by source('Analysis/_pipeline_setup.R') at L7 - the setwd makes the subsequent relative source() absolute in effect, so this is fully depth-independent. Data paths hardcoded absolute: OUT at L10 (getOption('mmm.audit_out', S:/...audit_hmm_state_architecture)), PROJ at L11; reads Stage 01 all_behavior_metrics.csv (L16) and Stage 08 hmm_state_assignments.csv (L54). Reference updates are COSMETIC ONLY - both referring files merely cite it in a comment as the estimator source. Must be run BEFORE locomotion_dominance, qc_sensitivity, gap_aware."
"Testing/audits/audit_inactive_phase_qc_redesign.R","data_dependent_audit","Testing/audit_inactive_phase_qc_redesign.R","Testing/audits/audit_inactive_phase_qc_redesign.R",TRUE,"none - standalone manual audit (newest of the set, commit 54d8757 'Audit the phase-aware inactive RFID QC redesign'); nothing sources it","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 27) + CWD-relative source (line 28) => depth-independent, SAFE. No self-path literal. READS S:: 12_systems_neuroscience_summary/5min_based/tables/qc_chip_loss_flags.csv and 03_derived_metrics/{5min,10min}_based/all_behavior_metrics.csv. WRITES TO S: three CSVs (inactive_qc_threshold_calibration.csv, inactive_qc_epoch_evidence.csv, inactive_qc_proposed_specification.csv) into a DEDICATED sibling dir <ST14>/audit_inactive_phase_qc/ - no collision with the first_night audit dir. Explicitly 'Read-only. Writes audit tables only; changes no production QC': it documents that Stage 14's chip-loss classifier (observed_fraction < 0.10, or < 0.25 with longest_gap_hours >= 4) is confounded with rest in the Inactive phase (median observed_fraction 0.149 Inactive vs 1.000 Active) and proposes a batch-x-system co-housing discriminator. Not superseded and not superseding anything; it is a standing, UNADOPTED proposal, so manuscript relevance is limited to qualifying inactive-phase QC flags. Outside the first-night lineage."
"Testing/audits/audit_phase_bug_impact.R","data_dependent_audit","Testing/audit_phase_bug_impact.R","Testing/audits/audit_phase_bug_impact.R",TRUE,"none - manual counterfactual run; cited only in a comment in the manuscript figure builder","manuscript/Fig1_behavior_candidates/build_fig1_candidates.R (line 278, comment)","yes","yes","yes_with_reference_updates","Directly cited by the manuscript staging script: manuscript/Fig1_behavior_candidates/build_fig1_candidates.R:277-278 ('See Testing/test_phase_classification.R and Testing/audit_phase_bug_impact.R.') justifying why the Stage 11/12-derived panels are marked blocked_pending_phase_classifier_fix_and_rerun in figure_manifest.csv - so this is the evidence record behind a manuscript figure-inclusion decision. That comment is the ONE cross-directory reference in the whole assigned set and must be updated on move (it is a comment, so nothing executes, but the manuscript audit trail would otherwise point at a non-existent path). Repo-root resolution: absolute setwd (line 7) + CWD-relative source (line 8) => depth-independent, SAFE. No THIS_SCRIPT literal. READS S:: 03_derived_metrics/{10min,5min}_based/all_behavior_metrics.csv. WRITES TO S:: phase_bug_impact_counterfactual.csv into 12_systems_neuroscience_summary/5min_based/audit_hmm_state_architecture/ (the PARENT audit dir, not the first_night_domain_heatmap subdir). Uses the shared exact classifier via source_mmm_helper('phase_classification_helpers.R') and re-applies a local copy of the old permissive matcher; not superseded. || MOVE_BREAKS: nothing functionally; the comment reference in manuscript/Fig1_behavior_candidates/build_fig1_candidates.R:278 becomes a stale path in the manuscript audit trail"
"Testing/audits/audit_stage09_stale_artifacts.R","data_dependent_audit","Testing/audit_stage09_stale_artifacts.R","Testing/audits/audit_stage09_stale_artifacts.R",TRUE,"none - manual forensic run; nothing sources it","none","yes","indirect","yes","Repo-root resolution: absolute setwd (line 8) + CWD-relative source (line 9) => depth-independent, SAFE. No self-path literal; the many 'Analysis/...' strings (lines 3, 25-37, 138-151) are descriptive tribble/comment text, NOT file reads, so nothing repo-relative is opened besides Analysis/_pipeline_setup.R after the setwd. READS S: heavily: 7 artifact families under <RFID>/analysis_ready (pipeline/09_early_prediction, three legacy 06_behavioral_dynamics/early_prediction trees, two legacy early_prediction_model_ladder mirrors) plus the quarantined snapshot root S:/.../RFID_snapshot_stage09_erroneous_active_plus_inactive_24h_20260827. WRITES TO S:: stage09_stale_artifact_audit.csv and stage09_downstream_consumer_audit.csv into audit_hmm_state_architecture/first_night_domain_heatmap/. Records one HIGH-risk finding relevant to manuscript reproducibility: Analysis/15_behavior_proteomics_integration.R:28 hard-codes a legacy 5-min early_behavior_features.csv with no resolver and no fallback, while canonical Stage 09 emits 10min only. Not superseded; unrelated to the first-night lineage despite sharing its output directory."
"Testing/audits/audit_step6_longitudinal_gapaware_robustness.R","data_dependent_audit","Testing/audit_step6_longitudinal_gapaware_robustness.R","Testing/audits/audit_step6_longitudinal_gapaware_robustness.R",TRUE,"none (manual: Rscript Testing/audit_step6_longitudinal_gapaware_robustness.R from repo root)","none","yes","indirect","yes","The one non-audit_hmm_* file in the assignment; it belongs with the HMM set by content (gap-aware longitudinal reassessment). Repo root via hardcoded absolute setwd at L6 + source('Analysis/_pipeline_setup.R') at L7, then source_mmm_helper('phase_classification_helpers.R') and ('hmm_stage14_helpers.R') at L8 - depth-independent. NOT CHAINED on any audit table: reads the SHIPPED gap-aware Viterbi labels hmm_state_assignments.csv under H (L10, L51) and the Stage 01 roster (L16), computing metrics within SequenceBlockID. PROJ/H/OUT all hardcoded absolute at L9-L11. Writes step6_longitudinal_gapaware_contrasts.csv (L77). git grep for 'audit_step6' returns zero hits outside the file itself."
"Testing/audits/compare_identity_correction_before_after.R","data_dependent_audit","Testing/compare_identity_correction_before_after.R","Testing/audits/compare_identity_correction_before_after.R",TRUE,"Rscript Testing/compare_identity_correction_before_after.R --baseline-root=<path> (self-executing audit, gated by the --file= main guard at lines 289-293); ALSO source()d as a function library by the portable test Testing/test_compare_identity_correction_driver.R:13","Testing/test_compare_identity_correction_driver.R:2 (comment), Testing/test_compare_identity_correction_driver.R:13 (LIVE source() call); Functions/identity_correction_comparison_helpers.R:6 (comment)","partial","yes","yes_with_reference_updates","DUAL ROLE, resolved definitively. As a LIBRARY it is portable: the guard `.identity_comparison_file_is_main` (lines 289-293) inspects commandArgs --file= and only auto-runs when the file itself is the Rscript entry point, so source()ing it does no work; the driver test exercises run_identity_correction_comparison() entirely against tempdir() fixtures. As an ENTRY POINT it is a data-dependent audit: needs S:/.../RFID plus a separately preserved pre-identity baseline root, and fails closed with no default. Depth safety: it source()s 'Analysis/_pipeline_setup.R' and 'Functions/identity_correction_comparison_helpers.R' CWD-relative, and its main guard matches on BASENAME regex - so moving it deeper breaks nothing internal. THE ONE THING THAT BREAKS: Testing/test_compare_identity_correction_driver.R:13 hardcodes source(""Testing/compare_identity_correction_before_after.R""). Implication for placement: moving the TEST into Testing/tests/ is harmless (that path is CWD-relative from the repo root, which is depth-independent), but moving THIS file requires editing that one line to ""Testing/audits/compare_identity_correction_before_after.R"". Do both moves in the SAME commit and re-run `Rscript Testing/tests/test_compare_identity_correction_driver.R` from the repo root to confirm the suite still passes. Zero-risk alternative if you want to touch nothing: leave this file at Testing/ root (KEEP) and only move the test. Manuscript-relevant: it produces the identity-correction before/after audit tables (Stage 03 + Stage 09) that document the correction for reviewers. Single commit 2026-08-26 - current, not legacy."
"Testing/audits/repair_existing_metrics_identity_utility.R","data_dependent_audit","Testing/repair_existing_metrics_identity_utility.R","Testing/audits/repair_existing_metrics_identity_utility.R",TRUE,"none - invoked by hand only; defines repair_existing_metrics_identity_for_testing() with ZERO callers repo-wide","Analysis/01_build_multiscale_behavior_metrics.R:27 (COMMENT POINTER ONLY, not a dependency)","partial","indirect","yes_with_reference_updates","RESOLVED: Analysis/01:27 is NOT an active dependency. Lines 20-27 of Analysis/01 are prose ('A forensic/testing-only single-file repair helper (never equivalent to a full rebuild) lives in Testing/repair_existing_metrics_identity_utility.R'); Analysis/01's only source() is line 60 source(.pipeline_setup). This file's own header confirms it: 'never invoked automatically by any Analysis/ script' and 'Unlike the removed production code path, this helper never calls quit()'. git grep of the function name returns only its own definition (line 42). It has no auto-run block at all - pure function definition. So the move is code-safe; only the Analysis/01:27 doc pointer needs the new path. Data dependence is partial: source()ing it needs only Analysis/_pipeline_setup.R (CWD-relative, and _pipeline_setup does not touch S: at source time), but calling the function rewrites a real all_behavior_metrics.csv IN PLACE. Belongs in audits/ rather than legacy/: single commit 2026-08-26, current, and Analysis/01's documented recovery narrative depends on the pointer staying accurate. Do NOT put it in tests/ - it is destructive and portable-suite-hostile."
"Testing/audits/test_animal_identity_contract.R","data_dependent_audit","Testing/test_animal_identity_contract.R","Testing/audits/test_animal_identity_contract.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_animal_identity_contract.R` from the repo root, and only succeeds because the S: drive is currently mounted (I confirmed all_behavior_metrics.csv is present).","none","yes","yes","yes","NOT TRULY PORTABLE - this is the clearest exception in the set. Line 14 hardcodes base_dir = S:/Lab_Member/.../Behavior/RFID and line 16 calls read_csv(metrics_path) with NO file.exists guard and no tryCatch; line 41 does a second unguarded read of raw_movement_animal_level_endpoints.csv via behavior_stage_dir(). Without S: it dies at line 16. Only lines 10-12 (the canonical_animal_id vector contract) are portable. It asserts manuscript-facing identity facts directly on the shipped data (animal 3 = Male/RES, animal 4 = Male/SUS, one Group and one Sex per AnimalNum, no duplicate overall_by_phase endpoints), hence manuscript_relevant = yes. Root resolution is `source(""Analysis/_pipeline_setup.R"")` (line 8), CWD-relative + parent-walking - moving it is mechanically harmless. DESTINATION RATIONALE: I deviate from Testing/tests/ because grouping it with the portable suite would make any future 'run the tests' runner fail on a machine without S:. If you prefer name-based grouping, Testing/tests/ works mechanically, but the file must then be excluded from any portable-suite manifest. Note its guarded near-equivalent already exists in test_stage19_identity_and_stage06_schema.R. || MOVE_BREAKS: nothing (the S: dependency is a pre-existing runtime requirement, unaffected by location)"
"Testing/audits/test_reporting_architecture.R","data_dependent_audit","Testing/test_reporting_architecture.R","Testing/audits/test_reporting_architecture.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_reporting_architecture.R` from the repo root, and only succeeds because S: is currently mounted (I confirmed all 9 required manuscript artifacts exist).","none","yes","yes","yes","NOT TRULY PORTABLE, and the most data-bound file in the set. Line 11 hardcodes the S: base_dir; line 25 is a hard `stopifnot(all(file.exists(required_artifacts)))` over nine manuscript artifacts (Behavioral_Source_Data.xlsx, primary_results.csv, supplementary_results.csv, animal_level_source_data.csv, prediction_source_data.csv, movement_phase_source_data.csv, provenance.csv, validation.csv, manifest.csv), so it fail-closes immediately without S:. It then reads every one of them, recomputes the source-data lineage from provenance.csv and asserts all.equal() at 1e-12, checks manifest SHA-256 coverage, validation$status all PASS, nrow(primary) == 10, and unzips the workbook to assert no formulas/error cells/drawings/VBA. Only the middle tempdir() section (lines 34-59, path-grammar and resolve_behavior_artifact precedence) is portable. This is the direct manuscript-artifact contract, so manuscript_relevant = yes. Root resolution `source(""Analysis/_pipeline_setup.R"")` (line 9) is depth-robust; same destination rationale as test_animal_identity_contract.R. || MOVE_BREAKS: nothing (the S: dependency is a pre-existing runtime requirement, unaffected by location)"
"Testing/audits/validate_cross_scale_animal_identity.R","data_dependent_audit","Testing/validate_cross_scale_animal_identity.R","Testing/audits/validate_cross_scale_animal_identity.R",TRUE,"Rscript Testing/validate_cross_scale_animal_identity.R (self-executing via the --file= main guard at lines 152-161)","Functions/animal_identity_invariants_helpers.R:6 (comment)","yes","yes","yes","RESOLVED BY THE FILE ITSELF - it is a data-dependent audit, NOT a portable test. Header line 2: '(S: DRIVE REQUIRED)'. Lines 22-25: 'This script requires the S: drive and real pipeline outputs; it is NOT part of the portable contract-test suite. The pure invariant-checking logic it calls into (Functions/animal_identity_invariants_helpers.R) is covered separately by Testing/test_animal_identity_invariants_engine.R using tempdir()/in-memory fixtures.' Confirmed by the code: defaults read S:/.../RFID/analysis_ready/03_derived_metrics/{10sec,1min,5min,10min,30min,phase}_based/all_behavior_metrics.csv (verified present on the drive), S:/.../Analysis/sus_animals.csv + con_animals.csv, and S:/.../RFID/MMMSociability/preprocessed_data (data-path 'MMMSociability' - never rename), and it stop()s if no derived-metrics outputs are found. Move safety is clean: source()s are CWD-relative ('Analysis/_pipeline_setup.R', 'Functions/animal_identity_invariants_helpers.R') and the guard `.cross_scale_identity_file_is_main` matches on BASENAME regex, so nothing breaks at any depth. The only inbound reference is a prose comment in Functions/animal_identity_invariants_helpers.R:6 - update for tidiness, not correctness. Manuscript-relevant: writes the cross-scale identity validation report backing the post-correction identity claims. Single commit 2026-08-26."
"Testing/legacy/check_behavioral_dynamics_structure.R","legacy","Testing/check_behavioral_dynamics_structure.R","Testing/legacy/check_behavioral_dynamics_structure.R",TRUE,"none (both callers are themselves dead: Testing/run_behavioral_dynamics_pipeline.R:20 and Testing/run_full_systems_behavior_pipeline.R:2)","Testing/run_behavioral_dynamics_pipeline.R:20; Testing/run_full_systems_behavior_pipeline.R:2","yes","no","yes_with_reference_updates","NOT classified legacy from the filename - from evidence. (1) Its only two consumers are the two dead pipeline runners below. (2) Its header names Analysis/06_burstiness_temporal_instability.R, 07_behavioral_state_space.R, 08_early_prediction_models.R, 09_dynamic_social_networks.R, 10_hmm_behavioral_states.R, 11_gamm_trajectory_features.R and 'Run Analysis/03_build_multiscale_behavior_metrics.R first' - ALL EIGHT of those filenames were verified MISSING from Analysis/ (pre-renumbering names). (3) Its input is the CWD-relative 'analysis_ready/03_derived_metrics/5min_based/all_behavior_metrics.csv'; the repo's analysis_ready/ contains only an EMPTY 00_structure_checks/5min_based/ (created 2026-09-02), while the real data lives at S:/.../RFID/analysis_ready/03_derived_metrics/ (verified present). Since it source()s 'Functions/behavioral_dynamics_helpers.R' CWD-relative, CWD must be the repo root, so its hardcoded relative input can NEVER resolve - it is non-functional as written, predating the getOption('mmm.derived_metrics_dir') convention used by Analysis/01. The code itself is intact (ensure_dir/write_table/read_behavior_table/standardize_behavior_columns all exist in Functions/behavioral_dynamics_helpers.R), so reviving it would only need the input root re-pointed - but that is a code change, out of scope for a no-science-change restructure. The QC role is now Analysis/00_qc_tracking_integrity.R. Move it together with its two runners so the 'Testing/check_behavioral_dynamics_structure.R' strings in them stay consistent (they are dead either way). Also note it WRITES into the git working tree (analysis_ready/00_structure_checks/) - separate hygiene issue."
"Testing/legacy/correlate_activity_physiology.R","legacy","Testing/correlate_activity_physiology.R","Testing/legacy/correlate_activity_physiology.R",TRUE,"none","none","yes","no","yes","994 lines, single commit 2026-05-06. ZERO grep hits. Internal header still carries a different filename ('correlate_gamm_auc_CombZ_two_auc_files_nature_FIXED_SCALES_PADDED.R'), a hallmark of an ad-hoc copied one-off. Consumes S:/.../RFID/statistics/gamm_new/analyses/gamm/tables/auc_individual_animals_*.csv (legacy GAMM output, not a current analysis_ready/pipeline artifact) plus S:/.../SIS_Analysis/E9_Behavior_Data.xlsx. The live equivalents are Analysis/07_gamm_trajectory_features.R and Analysis/14/15. No repo-relative source()."
"Testing/legacy/E9_SIS_AnimalPos-analyzing-shannon v.1.0.2.r","legacy","Testing/E9_SIS_AnimalPos-analyzing-shannon v.1.0.2.r","Testing/legacy/E9_SIS_AnimalPos-analyzing-shannon v.1.0.2.r",TRUE,"none","none","yes","indirect","yes","3365 lines, 2 commits, 2026-05-06 -> 2026-06-30. ZERO grep hits for the exact filename. Decisive context: its predecessors 'E9_SIS_AnimalPos-analyzing-shannon v.1.0.0.r' and 'v.1.0.1.r' ALREADY live in Analysis/_archive/legacy_deprecated/, and README_archive.md classes that whole family as legacy. Testing/ is simply holding the newest member of an already-archived family. Marked indirect (not 'no') only because 03b_multiscale_metric_validation_notes.md cites the sibling v.1.0.1 as the Shannon-entropy definition provenance for Analysis/01; v.1.0.2 itself is cited nowhere. All paths absolute on S: (working_directory, raw_data_dir, preprocessed_data_dir, custom_functions_path)."
"Testing/legacy/E9_SIS_AnimalPos-analyzing 2.0.0.R","legacy","Testing/E9_SIS_AnimalPos-analyzing 2.0.0.R","Testing/legacy/E9_SIS_AnimalPos-analyzing 2.0.0.R",TRUE,"none","none","yes","no","yes","git grep of the exact filename returns ZERO hits repo-wide. 293-line stripped variant of the v.2.0.0 analyzer (2 commits, 2025-10-22 -> 2025-11-11). All I/O is absolute on the data drive: working_directory = S:/Lab_Member/Tobi/.../RFID/MMMSociability, source(paste0(working_directory, '/E9_SIS_AnimalPos-functions.R')) (that S: file EXISTS - verified), reads S:/.../MMMSociability/preprocessed_data/*_preprocessed.csv. No repo-relative source() at all, so depth is irrelevant. Superseded by Analysis/01_build_multiscale_behavior_metrics.R. NOTE: the 'MMMSociability' inside these S: paths is the DATA directory, not repo identity - must never be rewritten."
"Testing/legacy/E9_SIS_AnimalPos-analyzing v.2.0.0.R","legacy","Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R","Testing/legacy/E9_SIS_AnimalPos-analyzing v.2.0.0.R",TRUE,"none","Analysis/_archive/03b_multiscale_metric_validation_notes.md:9 (doc-only provenance citation)","yes","indirect","yes_with_reference_updates","1324 lines, 17 commits, 2024-08-28 -> 2026-05-16. Not executed by anything. Its ONE inbound reference is Analysis/_archive/03b_multiscale_metric_validation_notes.md:9, which lists it under 'Checked legacy sources' used to validate that Analysis/01 preserves the legacy Movement/Proximity/Entropy definitions -> that is manuscript-adjacent PROVENANCE, hence indirect, and that .md line must be updated on move. Lines 68-74 build function_file_candidates; candidates 1-2 (S:/.../MMMSociability/Testing/... and /Functions/...) were verified MISSING on the drive, candidate 3 (S:/.../MMMSociability/E9_SIS_AnimalPos-functions.R) EXISTS and wins, so the repo-relative fallback at line 73 ('Testing/E9_SIS_AnimalPos-functions v.2.0.0.R') is never reached while S: is mounted. Still update line 73 to the new path when moving: if S: is ever absent the list would silently fall through to Functions/E9_SIS_AnimalPos-functions.R, which is a DIFFERENT file (2256 vs 1833 lines)."
"Testing/legacy/E9_SIS_AnimalPos-functions v.2.0.0.R","legacy","Testing/E9_SIS_AnimalPos-functions v.2.0.0.R","Testing/legacy/E9_SIS_AnimalPos-functions v.2.0.0.R",TRUE,"none","Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R:70; Testing/E9_SIS_AnimalPos-analyzing v.2.0.0.R:73","no","indirect","yes_with_reference_updates","Pure function-definition file (2256 lines), no I/O at source() time, 16 commits 2024-08-28 -> 2026-05-16. Its ONLY consumer is the v.2.0.0 analyzer above, so the two MUST move together. NOT a duplicate of Functions/E9_SIS_AnimalPos-functions.R - diff confirms they differ (2256 vs 1833 lines), so it cannot simply be deleted in favour of the Functions/ copy without changing behaviour of the legacy analyzer. Move as a pair and update the analyzer's line 73 candidate path."
"Testing/legacy/lme v.3.1.2.r","legacy","Testing/lme v.3.1.2.r","Testing/legacy/lme v.3.1.2.r",TRUE,"none","none","yes","no","yes","2453 lines, 6 commits, 2025-10-23 -> 2026-05-13. ZERO grep hits. Sibling 'lme v.3.1.1.r' is ALREADY in Analysis/_archive/legacy_deprecated/, and README_archive.md documents 'legacy_deprecated/lme v.*: Legacy LME scripts from the deprecated folder.' Reads S:/.../MMMSociability/processed_data/data_lme_format/data_filtered_agg.csv (a legacy Formatting/E9_SIS_lme_data_formatting.r product, not a current pipeline artifact) and writes S:/.../MMMSociability/statistics/lme/. Superseded by Analysis/03_primary_raw_movement_phase_stats.R + Analysis/04/07. No repo-relative source() -> depth-independent."
"Testing/legacy/prediction_analysis.r","legacy","Testing/prediction_analysis.r","Testing/legacy/prediction_analysis.r",TRUE,"none","none","yes","no","yes","121 lines, single commit 2026-05-06. ZERO grep hits. Reads the legacy S:/.../MMMSociability/processed_data/data_lme_format/data_filtered_agg.csv and writes S:/.../MMMSociability/statistics/prediction/. Functionally superseded by Analysis/09_early_prediction_model_ladder.R, which Analysis/_archive/README_archive.md explicitly names as 'the primary, reviewer-safe' replacement for the older prediction layer. Not in run_all_analysis.R. No repo-relative source()."
"Testing/legacy/run_behavioral_dynamics_pipeline.R","legacy","Testing/run_behavioral_dynamics_pipeline.R","Testing/legacy/run_behavioral_dynamics_pipeline.R",TRUE,"none","none","yes","no","yes","DEFINITIVELY SUPERSEDED BY Analysis/run_all_analysis.R - verified by comparing stage lists. Its scripts_to_run (lines 19-24) is: Testing/check_behavioral_dynamics_structure.R, Analysis/06_burstiness_temporal_instability.R, Analysis/07_behavioral_state_space.R, Analysis/08_early_prediction_models.R. I file-tested each: ALL THREE Analysis/ targets are MISSING (they are pre-renumbering names; 08_early_prediction_models.R survives only in Analysis/_archive/, which README_archive.md marks 'Superseded by the primary, reviewer-safe 09_early_prediction_model_ladder.R'). So this runner is already broken - it would report FAILED for every stage. run_all_analysis.R instead declares 17 stages (00-15) with env-gated optionals (RUN_OPTIONAL_HMM, RUN_SYSTEMS_EXTENSION, RUN_BEHAVIOR_PROTEOMICS) and sources _pipeline_setup.R; this runner sources no setup and writes its status CSV into the git working tree (analysis_ready/06_behavioral_dynamics/). Zero inbound references. Single commit 2026-05-10. Paths are CWD-relative so the move itself is depth-independent."
"Testing/legacy/run_full_systems_behavior_pipeline.R","legacy","Testing/run_full_systems_behavior_pipeline.R","Testing/legacy/run_full_systems_behavior_pipeline.R",TRUE,"none","none","yes","no","yes","Same verdict, stronger. 60 lines, no header comment at all, 2 commits 2026-05-10. Its scripts_to_run (lines 1-10) adds Analysis/09_dynamic_social_networks.R, 10_hmm_behavioral_states.R, 11_gamm_trajectory_features.R, 12_behavior_proteomics_integration.R - I file-tested all seven Analysis/ targets and EVERY ONE IS MISSING under those names (current equivalents are 04/05/06/07/08/09/15 after renumbering). Superseded by Analysis/run_all_analysis.R, which covers the same scientific span plus stages 11-15 and gates the optional HMM/proteomics stages. Zero inbound references; also writes its status CSV into the git working tree. CWD-relative paths -> move is depth-independent."
"Testing/legacy/sociability_lme v.1.0.2.r","legacy","Testing/sociability_lme v.1.0.2.r","Testing/legacy/sociability_lme v.1.0.2.r",TRUE,"none","none","yes","no","yes","9031 lines, single commit 2026-05-04. ZERO grep hits. Siblings 'sociability_lme v.1.0.0.r' and 'v.1.0.1.r' are ALREADY in Analysis/_archive/legacy_deprecated/ under the README_archive rule 'legacy_deprecated/sociability_lme v.*'. Reads the same legacy data_filtered_agg.csv plus S:/.../BatchAnalysis/... and S:/.../sus_animals.csv; writes S:/.../RFID/statistics/ - the same tree whose gamm/ tables Testing/correlate_activity_physiology.R consumes, i.e. this pair forms a self-contained legacy GAMM/LME side-branch outside run_all_analysis.R. No repo-relative source()."
"Testing/tests/test_animal_identity_invariants_engine.R","portable_test","Testing/test_animal_identity_invariants_engine.R","Testing/tests/test_animal_identity_invariants_engine.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_animal_identity_invariants_engine.R` from the repo root.","Testing/validate_cross_scale_animal_identity.R:24 (plain header comment, not written to any output)","no","indirect","yes_with_reference_updates","Root resolution: two CWD-relative sources - `source(""Analysis/_pipeline_setup.R"")` (lines 10 AND 118, it re-sources mid-file) plus `source(""Functions/animal_identity_invariants_helpers.R"")` (line 11). All CWD-relative => depth-independent. Pure in-memory make_scale() tibbles; grep found zero S:/read_csv/list.files. Note the companion Testing/validate_cross_scale_animal_identity.R explicitly documents itself as the S:-requiring sibling and names this file as 'the portable contract-test suite' counterpart - that comment is worth keeping accurate; I verified it is a comment only, never written to disk. || MOVE_BREAKS: nothing at runtime; only the stale comment path in validate_cross_scale_animal_identity.R:24"
"Testing/tests/test_animalpos_preprocessing_helpers.R","portable_test","Testing/test_animalpos_preprocessing_helpers.R","Testing/tests/test_animalpos_preprocessing_helpers.R",TRUE,"none - no test runner, no CI, no .Rprofile exists in the repo. Invoked manually as `Rscript Testing/test_animalpos_preprocessing_helpers.R` from the repo root.","none","no","indirect","yes","Root resolution: the ONLY path dependency is `source(""Functions/animalpos_preprocessing_helpers.R"")` at line 5 - CWD-relative, no _pipeline_setup.R, no setwd. Depth-independent: it resolves against the invocation CWD (repo root), not the script location, so moving the file one level deeper is a no-op. Header self-declares 'Pure in-memory; no S: drive access required' and the code confirms it (sub-second timestamp parsing, interval splitting on in-line fixtures). Zero repo-wide references."
"Testing/tests/test_canonical_animal_id.R","portable_test","Testing/test_canonical_animal_id.R","Testing/tests/test_canonical_animal_id.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_canonical_animal_id.R` from the repo root.","none","no","indirect","yes","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 10), CWD-relative; _pipeline_setup then sets MMM_REPO_ROOT via the parent-walking find_mmm_repo_root(), which is depth-robust. Doubly safe. Pure in-memory vectors testing the canonical_animal_id() contract (zero-stripping, whitespace, OR/OQ prefixes, large-RFID-string precision). The only 'RFID' grep hits are in comment/test-label text, not paths. Zero repo-wide references."
"Testing/tests/test_compare_identity_correction_driver.R","portable_test","Testing/test_compare_identity_correction_driver.R","Testing/tests/test_compare_identity_correction_driver.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_compare_identity_correction_driver.R` from the repo root.","none","no","indirect","yes","Root resolution: it does NOT source _pipeline_setup.R. Its one dependency is `source(""Testing/compare_identity_correction_before_after.R"")` (line 13) - a CWD-relative path to a SIBLING file inside Testing/. Because it is CWD-relative (resolved from the repo root), moving THIS file into Testing/tests/ does not affect it. IMPORTANT COUPLING: the path is hardcoded as 'Testing/...', so if Testing/compare_identity_correction_before_after.R is itself relocated in the same restructure, this line must be updated in lockstep. Also note lines 11-12 deliberately clear MMM_IDENTITY_BASELINE_ROOT / options(mmm.identity_baseline_root) to force the fail-closed path. All fixtures under tempdir(); test 4 explicitly asserts outputs land outside getwd(). || MOVE_BREAKS: nothing on its own; but a coordinated move of Testing/compare_identity_correction_before_after.R would break the hardcoded source() at line 13"
"Testing/tests/test_downstream_boundary_and_gap_contract.R","portable_test","Testing/test_downstream_boundary_and_gap_contract.R","Testing/tests/test_downstream_boundary_and_gap_contract.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_downstream_boundary_and_gap_contract.R` from the repo root.","none","no","indirect","yes","Root resolution: `repo <- Sys.getenv(""MMM_REPO_DIR"", unset = ""C:/Users/topohl/Documents/GitHub/MMMSociability""); setwd(repo)` (lines 30-31), then sources Functions/animalpos_preprocessing_helpers.R via file.path(repo, ...) - fully ABSOLUTE, therefore completely independent of the script's own location. The later load_defs(""Analysis/02_build_dyadic_rfid_contacts.R"") and ""Analysis/19_spatial_occupancy_maps.R"" are relative but evaluated AFTER setwd(repo), so they are safe too. Note the default in Sys.getenv is the local CLONE DIRECTORY path (…/GitHub/MMMSociability), a repo-identity reference, distinct from the S: data root. No local data read: the analysis scripts are only PARSED and selected top-level defs evaluated; all fixtures are in-line intervals."
"Testing/tests/test_first_night_domain_contract.R","portable_test","Testing/test_first_night_domain_contract.R","Testing/tests/test_first_night_domain_contract.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_first_night_domain_contract.R` from the repo root.","none","no","indirect","yes","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 8) then six source_mmm_helper() calls (phase_classification, animalpos_preprocessing, first_night_window, hmm_stage14, first_night_domain_helpers, first_night_domain_driver) which resolve off the absolute MMM_REPO_ROOT. Only the first source() is CWD-relative; both mechanisms are depth-robust. No data reads whatsoever - grep's only 'RFID' hit is inside a check() message string ('every declared contributor must be a standardized raw RFID feature'). Locks the audited domain formulas, adjacency-aware RMSSD (mmm_n_adjacent_pairs must exclude the bridged 3->100 jump), within-Sex standardization, sign orientation and FDR family sizes."
"Testing/tests/test_first_night_window_parity.R","portable_test","Testing/test_first_night_window_parity.R","Testing/tests/test_first_night_window_parity.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_first_night_window_parity.R` from the repo root.","Functions/first_night_window_helpers.R:21 (plain comment - safe to edit); Functions/first_night_domain_driver.R:83 - CAUTION: `selector_parity = ""Testing/test_first_night_window_parity.R asserts equality with Stage 09""` is a FIELD of window_contract, written to first_night_window_contract.csv; Analysis/14_systems_neuroscience_summary_dashboard.R:5693 - CAUTION: inside first_night_documentation$text, written to tables/sis_first_night_documentation.csv","partial","indirect","yes_with_reference_updates","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 15) + five source_mmm_helper() calls; parse(""Analysis/09_early_prediction_model_ladder.R"") is CWD-relative. Depth-independent. Section B's live 111-animal check uses getOption(""mmm.project_root"", <S: default>) and is guarded by `if (!file.exists(f)) { message(""SKIP live parity ...""); next }`, so it degrades cleanly; Section A (fixture animals A/B/C, 72 slots, 5 missing leading, elapsed-slot anchoring) always runs. MOST IMPORTANT FINDING OF THIS FILE: 2 of its 3 references are baked into SHIPPED CSV CONTENT (first_night_window_contract.csv and sis_first_night_documentation.csv). Renaming the path there is not a comment edit - it changes artifact bytes and any SHA-256 manifest covering them. Recommend leaving those two strings untouched, or accepting a regeneration+rehash. || MOVE_BREAKS: nothing at runtime; updating the two output-embedded references would change the content of first_night_window_contract.csv and sis_first_night_documentation.csv"
"Testing/tests/test_hmm_stage14_contract.R","portable_test","Testing/test_hmm_stage14_contract.R","Testing/tests/test_hmm_stage14_contract.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_hmm_stage14_contract.R` from the repo root.","Testing/audit_hmm_state_architecture_provenance_addendum.R:105 - CAUTION: that string is a cell in the `consumed` tibble that is written out via write_csv(consumed, .../audit_doc_v7_artifact_consumption_map.csv)","partial","indirect","yes_with_reference_updates","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 9) + source_mmm_helper(""hmm_stage14_helpers.R""). Depth-robust. The S: block is explicitly opt-in and self-documented at line 171: 'The full E9 fixture/data check is enabled automatically when S: is present', guarded by `if (file.exists(full_data_path))` and a nested `if (file.exists(qc_path) && file.exists(occupancy_path))`. Sections A/B/G/H (alias collapsing 0004->4, Welch exclusion, n counting 8 animals not 32 cage-change rows) always run in memory. REFERENCE CAUTION: the addendum's mention embeds the literal 'Testing/test_hmm_stage14_contract.R:196-207' into a written audit CSV, so editing it changes that artifact's bytes - update only if regenerating that audit output is acceptable. || MOVE_BREAKS: nothing at runtime; updating the addendum reference would alter the content of audit_doc_v7_artifact_consumption_map.csv"
"Testing/tests/test_identity_comparison_engine.R","portable_test","Testing/test_identity_comparison_engine.R","Testing/tests/test_identity_comparison_engine.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_identity_comparison_engine.R` from the repo root.","none","no","indirect","yes","Root resolution: single CWD-relative `source(""Functions/identity_correction_comparison_helpers.R"")` at line 10. No _pipeline_setup.R, no setwd, no --file=. Depth-independent. Entirely in-memory tibbles (compare_identity_table on hand-built old/new frames); header states 'no S: drive access required' and grep for S:/read_csv/list.files found nothing. Zero repo-wide references."
"Testing/tests/test_output_path_length.R","portable_test","Testing/test_output_path_length.R","Testing/tests/test_output_path_length.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_output_path_length.R` from the repo root.","Functions/behavioral_dynamics_helpers.R:76 (plain comment, not written to any output)","no","indirect","yes_with_reference_updates","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 16), CWD-relative + parent-walking fallback. Depth-independent. Portable despite the S: string: project_root = getOption(""mmm.project_root"", <S: default>) is used only as ARITHMETIC INPUT - the test builds candidate Stage 15 paths and asserts nchar() budgets; it never reads them. The two live-tree scans are guarded by dir.exists(d) (line 77) and dir.exists(s16) (line 109). LOAD-BEARING CONSTANT: the hardcoded S: root string determines the measured path lengths and hence the >=20-char margin assertion, so it must not be shortened/edited during the restructure or the test's meaning changes. Lines 25-26 also assert the root is NOT a temp dir, so it cannot be neutralized by pointing at tempdir(). No setter for options(mmm.project_root) exists anywhere in the repo, so the default always applies. || MOVE_BREAKS: nothing at runtime; only the stale comment path in Functions/behavioral_dynamics_helpers.R:76"
"Testing/tests/test_phase_classification.R","portable_test","Testing/test_phase_classification.R","Testing/tests/test_phase_classification.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_phase_classification.R` from the repo root.","manuscript/Fig1_behavior_candidates/build_fig1_candidates.R:277 (plain R comment, not written to any output)","no","indirect","yes_with_reference_updates","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 5) then source_mmm_helper() for phase_classification_helpers.R and duration_normalization_helpers.R - those resolve off MMM_REPO_ROOT (absolute), so only the first source() is CWD-relative. Depth-independent. No file reads at all; locks out the Inactive-matches-Active substring bug. The single reference at build_fig1_candidates.R:277 is an ordinary code comment - I verified it is NOT interpolated into any write_table/write_csv call - so updating the path there is a pure documentation edit with no artifact impact. || MOVE_BREAKS: nothing at runtime; only the stale comment path in build_fig1_candidates.R:277"
"Testing/tests/test_stage09_endpoint_identity.R","portable_test","Testing/test_stage09_endpoint_identity.R","Testing/tests/test_stage09_endpoint_identity.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_stage09_endpoint_identity.R` from the repo root.","none","no","indirect","yes","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 16), CWD-relative + parent-walking fallback. Depth-independent. Header states 'Runs entirely in memory / tempdir(); no S: drive access required' and the grep confirms no S: path and no external read. Guards the n=111->88 join-drop regression where Stage 09 canonicalized only one side of the CombZ join. Zero repo-wide references."
"Testing/tests/test_stage09_primary_window.R","portable_test","Testing/test_stage09_primary_window.R","Testing/tests/test_stage09_primary_window.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_stage09_primary_window.R` from the repo root.","none","no","indirect","yes","Root resolution is MIXED but safe: line 20 sources Functions/animalpos_preprocessing_helpers.R through an ABSOLUTE file.path(Sys.getenv(""MMM_REPO_DIR"", unset=""C:/Users/topohl/Documents/GitHub/MMMSociability""), ...), but - unlike its two siblings - it does NOT setwd(), and line 29 then uses the CWD-relative STAGE09 <- ""Analysis/09_early_prediction_model_ladder.R"". So it depends on CWD == repo root (already true for the documented invocation) and is depth-independent either way. Stage 09 is parsed, never executed: only the 9 named top-level defs (select_primary_active_window, is_active_phase, early_window_hours=12, bin_size_min=10, expected_early_bins_per_animal=72, ...) are evaluated, so no data is read. Guards the confirmed results-changing 'inactive contains active' str_detect defect."
"Testing/tests/test_stage14_upstream_registry.R","portable_test","Testing/test_stage14_upstream_registry.R","Testing/tests/test_stage14_upstream_registry.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_stage14_upstream_registry.R` from the repo root.","none","no","indirect","yes","Root resolution: `source(""Analysis/_pipeline_setup.R"")` (line 9), CWD-relative + parent-walking fallback. Depth-independent. Every fixture is built under tempdir() (bases mmm_test1..mmm_testD) with write_stub(); the many analysis_ready/... strings in this file are RELATIVE sub-paths joined onto those tempdirs, not S: reads - I traced each one. Exercises resolve_stage09_early_prediction_artifact() canonical-vs-legacy precedence. Zero repo-wide references."
"Testing/tests/test_stage19_group_sex_labels.R","portable_test","Testing/test_stage19_group_sex_labels.R","Testing/tests/test_stage19_group_sex_labels.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_stage19_group_sex_labels.R` from the repo root.","none","partial","indirect","yes","Root resolution: Sys.getenv(""MMM_REPO_DIR"", unset=""C:/.../GitHub/MMMSociability"") + setwd(repo) (lines 24-25); readLines(""Analysis/19_spatial_occupancy_maps.R"") afterwards. Location-independent. Its optional data block does NOT touch S: at all - it reads OUT <- Sys.getenv(""MMM_STAGE19_OUT"", unset = ""C:/tmp/s19v""), a scratch validation-run directory, and is fully guarded by `if (dir.exists(D) && file.exists(...))` with an explicit SKIPPED else branch. Checks 1-3 (factor-code corruption: chr[i] <- factor_element storing the integer code, so 'RES' becomes '2') are pure in-memory and always run. Checks 4-8 only upgrade when a validated run is present."
"Testing/tests/test_stage19_identity_and_stage06_schema.R","portable_test","Testing/test_stage19_identity_and_stage06_schema.R","Testing/tests/test_stage19_identity_and_stage06_schema.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_stage19_identity_and_stage06_schema.R` from the repo root.","none","partial","indirect","yes","Root resolution: Sys.getenv(""MMM_REPO_DIR"", unset=""C:/.../GitHub/MMMSociability"") + setwd(repo) (lines 23-24), then ABSOLUTE file.path(repo, ""Functions"", ...) sources. Location-independent. Local data is OPTIONAL and correctly guarded: line 135 hardcodes the S: Stage 01 export and line 136 wraps every read in `if (file.exists(s01))`, with an explicit else branch printing 'PASS (structural only; canonical export not mounted)'. So checks 3-4 (roster = 111 animals, 24 CON / 49 RES / 38 SUS, 58 F / 53 M, animal 4 = SUS) upgrade from structural to verified when S: is present, and the script still exits 0 without it. Checks 5-7 readLines(""Analysis/06_dynamic_social_networks.R"") - repo file, always available after setwd."
"Testing/tests/test_write_registry_rerun_behavior.R","portable_test","Testing/test_write_registry_rerun_behavior.R","Testing/tests/test_write_registry_rerun_behavior.R",TRUE,"none - no runner/CI. Invoked manually as `Rscript Testing/test_write_registry_rerun_behavior.R` from the repo root.","none","no","indirect","yes","Root resolution: `source(""Analysis/_pipeline_setup.R"")` TWICE (lines 22 and 32) - deliberate, since the whole point is to simulate a same-session stage rerun re-entering the pipeline entrypoint and re-creating .mmm_output_write_registry. Both are CWD-relative + parent-walking fallback, so depth-independent and the double-source is unaffected by a move. All I/O is tempfile()-based (tmp, tmp2, tmp3). Zero repo-wide references."