A gene essentiality signature enables predicting the mechanism of action of drugs

Abstract

Cancer drugs often kill cells independent of their putative targets, suggesting the limitation of existing knowledge on the mechanisms of action. In this study, we explored whether the integration of loss-of-function genetic and drug sensitivity screening data can define a gene essentiality signature to better understand the drug target interactions. We showed that our gene essentiality signature can predict drug targets more accurately than chemical fingerprints and drug-perturbated gene expression signatures. We further showed how gene essentiality signature can help identify mechanisms of action of drugs de novo, including the EGFR inhibitor lapatinib, and drugs associated with DNA mismatch repair. Finally, we established gene essentiality signatures for noncancer drugs and used them to predict their anticancer targets. We have successfully validated the target predictions for multiple noncancer drugs, using cell-based drug target deconvolution by the proteome integral solubility alteration assay. Our study provides a novel signature of drugs that may facilitate the rational design of drug repurposing.