Instantiation of Patient-Specific Logical Models With Multi-Omics Data Allows Clinical Stratification of Patients

Logical models of cancer pathways are typically built by mining the literature and they are usually generic as they apply for large cohorts and do not capture the heterogeneity of patient. We present here a novel framework to tailor logical models to a patient's tumor. This methodology enables comparisons between the model simulations and the individual clinical data.
Our approach focuses on integrating mutations, copy number alterations (CNA), and expression data to logical models. These omics data, after appropriate processing, can be incorporated in the model modifying the activity of the node, the initial conditions or the transition rates, as defined in MaBoSS, a tool performing stochastic simulations of logical models. As a first proof of concept, omics data from breast-cancer patients is integrated into several logical models to derive phenotypic outputs that correlate with clinical read-outs such as survival, with better performances combining both mutations and expression data. All in all, we aim to combine the mechanistic insights of logical modeling with multi-omics data integration to provide patient-relevant models to physicians, enabling precision medicine.