Computational Biology | Interpretable Machine Learning | Systems Biology
I'm a computational biologist developing mathematical models, machine-learning methods, and scientific software to understand complex biological systems.
My work focuses on combining mechanistic knowledge with experimental data to build interpretable, predictive models. My background spans systems biology, network inference, optimization, multi-omics analysis, and computational oncology.
Currently: Postdoctoral Researcher at University of Luxembourg, developing methods for topological analysis, data integration, and Bayesian learning on networks. Head of Computational Biology at EndoVia Therapeutics, developing an interpretable modeling platform for personalized oncology and compound prioritization.
A Python framework for mechanistic modeling of biological signaling networks, building on the probabilistic logical modeling approach introduced in FALCON.
It combines prior network knowledge with experimental perturbation data to estimate model parameters and investigate context-dependent signaling mechanisms.
- Network-constrained, interpretable models
- Parameter estimation, regularization, and uncertainty analysis
- Multi-context modeling across cell lines and experimental conditions
- Reproducible benchmarks and biological applications
Repository · Scientific experiments · Reproducibility · Tests
My PhD project: A MATLAB toolbox for calibrating large-scale logical models of signaling pathways against experimental data.
FALCON provides optimization and regularization methods to contextualize prior-knowledge networks and investigate cell-specific signaling mechanisms.
Published in Bioinformatics (2017), with subsequent methodological developments published in Frontiers in Physiology (2018).
Source code · Original publication · Regularization methods
I also enjoy building algorithms and software outside my main research field.
An experimental chess variant and AI environment for investigating game-tree search, strategic evaluation, and algorithm design.
Includes multiple competing strategies, automated tournaments, and reproducible benchmarks. One evaluation compares strategies across 39,200 games.
A browser-based educational game exploring rapid visual recognition, working memory, and spelling.
Built with React and TypeScript, with automated tests, continuous integration, and a validated French-language stimulus corpus.
An interactive astronomy visualization prototype exploring how large streams of telescope alerts could be inspected and prioritized.
Combines sky maps, light curves, anomaly scores, and object exploration in a browser interface. The current demonstration uses synthetic data rather than live observatory alerts.
- Interpretable machine learning: structured neural networks, biological priors, and mechanistically constrained models
- Systems biology: signaling networks, network inference, and perturbation modeling
- Computational oncology: multi-omics integration, drug-response modeling, and compound prioritization
- Scientific computing: numerical optimization, simulation, algorithms, and reproducible research
Scientific computing: Python, PyTorch, NumPy, SciPy, scikit-learn, MATLAB, R
Methods: Machine learning, optimization, graph algorithms, statistical modeling, network-based modeling
Software development: Git, automated testing, scientific workflows, TypeScript, React
- De Landtsheer et al. (2017). FALCON: A Toolbox for the Fast Contextualization of Logical Networks. Bioinformatics.
- De Landtsheer et al. (2018). Using Regularization to Infer Cell Line Specificity in Logical Network Models of Signaling Pathways. Frontiers in Physiology.
Email: seb@delandtsheer.com



