Knowledge Graph-based Recommendation Framework Identifies Novel Drivers of Resistance in EGFR mutant Non-small Cell Lung Cancer

Abstract Resistance to EGFR inhibitors (EGFRi) is complex and only partly understood. It presents a major obstacle in treating non-small cell lung cancer (NSCLC). One of the most exciting new ways to find potential resistance markers involves running functional genetic screens, such as CRISPR, followed by manual triage of significantly enriched genes. This triage stage requires specialized knowledge and can take even experts a number of months to complete. To find key drivers of resistance faster we built a hybrid recommendation system on top of a heterogeneous biomedical knowledge graph. In the absence of either training data or continuous feedback we approached recommendations as a multi-objective optimization problem. Genes were ranked based on trade-offs between diverse types of evidence linking them to potential mechanisms of EGFRi resistance. This unbiased approach narrowed down the number of targets from more than 3,000 to 36 and reduced hit identification time from months to minutes. Similar recommendation system frameworks could be applied to a range of related problems in drug discovery space.