CauseHet

Cancers are heterogeneous diseases. Variations in tumor composition, gene expression, and response to environmental changes are key factors in cancer progression, while hampering scientific efforts to understand and treat the disease. Current knowledge of cancer biology and corresponding therapeutic strategies are based on simple molecular classifications, which reflect only the most abundant tumor subtype in the sample examined, and which overlook the fact that cancers are made up of cells with different identities and origins (cellular heterogeneity). This tumor heterogeneity is not only a major source of therapeutic resistance but it also affects all the biological processes underlying tumorigenesis. Deciphering the causes and impacts of tumor heterogeneity has been a scientific bottleneck so far. Bioinformatics and statistical approaches, leveraging high throughput multiomic data, offer a powerful alternative to characterize and understand cancer heterogeneity. The goal of this project is to provide an original framework analyzing the causes and consequences of tumor functional heterogeneity. Rather than focusing on cell types present within a tumor, we will question tumor heterogeneity at the phenotypic level, inferring differentially regulated functional modules between individuals. From there, we will develop integrative models to study the relationship between tumor functional heterogeneity and cancer evolution. Finally, we will perform high-dimension mediation analysis to determine molecular drivers of tumor heterogeneity and investigate the causality link between patient treatment and disease outcomes. This approach is based on statistical analysis and algorithms developments to leverage the recent massive generation of high-throughput multimodal molecular data at the patient level. Such models will allow a better understanding of the biological mechanisms involved to build a functional vision of tumor heterogeneity.