From Big Data to Bedside (BD2B): Precision Oncology in a Big Data Era
Cancer is mainly caused by heterogeneous somatic genome alterations (SGAs). Genome-scale data from individual patients are now readily available, and it is anticipated that precisely targeting specific genomic alterations of individual tumors will bring in more effective therapies. However, there are 3 major gaps that hinder the translation of the genome data of a tumor to personalized therapy. First, a tumor often hosts hundreds to a thousand SGAs, with a few being drivers and the majority being passengers, and the methods for determining the drivers (thereby the targets of therapy) of a given tumor remain to be fully developed. Second, majority of driver genes are “undruggable”. This problem can potentially addressed by mapping undruggable drivers to targetable pathways. However it remains a largely unsolved problem to map driver SGAs to such pathways. Third, cancer results from orchestrated perturbation of multiple pathways, it is a challenge to find combinatorial patterns of pathway perturbations and then design efficient single agent or combination therapy for a specific patient. In this presentation, I will discuss novel computational approaches for bridging the above gaps developed in the Center of Causal Discovery (www.ccd.pitt.edu) at the University of Pittsburgh.