Computational biology · Cancer · Somatic evolution
We are a computational biology lab with a small wet-lab component, based at the IRIBHM Jacques E. Dumont (ULB, Brussels). We generate and analyse human ’omics data to understand how normal cells accumulate changes, how cancers arise from them, and how tumours evolve as they progress and respond to treatment.
Our team brings together researchers trained in computer science, mathematics, biology and bioinformatics. We combine new algorithms, statistical modelling and machine learning with close collaborations with clinicians and pathologists, so that our methods stay anchored in real patient material and real clinical questions.
Our work is organised around four complementary axes that share data, methods and people. Many projects sit at the crossroads of two or more of them.
Thyroid cancers span a remarkable range of behaviours, from indolent papillary carcinomas (PTC) to some of the most aggressive human tumours, anaplastic carcinomas (ATC), as well as medullary carcinomas (MTC). We study how these tumours evolve, including radiation-induced PTC arising after Chernobyl, and use patient-derived organoid models to connect genomic changes to cell behaviour.
Gene expression tells us what tumour cells are actually doing. We build and evaluate cancer gene signatures, study RNA editing, and work with spatial and single-nucleus transcriptomics to resolve how cell states are organised within a tumour. Part of this work aims at clinically useful signatures, notably in pancreatic cancer.
Tissue architecture holds information that sequencing alone misses. We use deep learning and image registration to reconstruct tissues in 3D from serial histology sections (building on the CODA and VALIS pipelines), with the goal of a “histome” atlas that maps cell types and structures across entire tissue volumes.
Every tumour genome carries a record of its past. We develop methods to call mutations and copy-number changes, time them relative to one another, and model how tumours grow in 3D. We also study inherited cancer predisposition, including Li-Fraumeni syndrome and xeroderma pigmentosum and its skin lesions in Tanzania.
Our small wet lab lets us generate the data our questions require, rather than relying only on public resources, and keeps analysts close to the biology.
We aim to release reproducible, well-documented code and pipelines, and we run our analyses on our shared high-performance computing infrastructure.
Newcomers are paired with a tutor and onboarded with shared guidelines and HPC documentation. Interested in joining? Get in touch with one of our group leaders.
