Abductive Network Action Inference for Targeted Therapy Discovery
Résumé
Complex diseases such as cancer result from the combined actions of genetic perturbations whose characterization is crucial to determine the healing treatment. The challenge of therapy discovery focuses on the identification of the causal mechanisms underlying the genotype-phenotype relationships. In this undertaking, networks provide suitable representations to model molecular interactions and enable the analysis of the effect of multiple molecular perturbations on the cell system behaviour. Although network-based analysis was announced as a key milestone for drug discovery, the challenge remains daunting. A main issue is to properly qualify the actions of diseases on networks and their dynamical effects to discover the appropriate targets for drugs. In this article, we propose a new computational method for network action inference using Boolean networks to model the dynamics of biological networks and where disease/drug actions are represented as arc additions and deletions. Based on abductive reasoning, the method finds the actions that provide the best parsimonious explanation for shifting the cell from a diseased state to a healed state. The method was applied to retrieve the necessary drug actions in the case of synthetic lethality for Breast Cancer.