Abstract
Achieving precision oncology requires accurate identification of targetable cancer vulnerabilities in patients. Generally, genomic features are regarded as the state-of-the-art method for stratifying patients for targeted therapies. In this work, we conduct the first rigorous comparison of DNA- and expression-based predictive models for viability across five datasets encompassing chemical and genetic perturbations. We find that expression consistently outperforms DNA for predicting vulnerabilities, including many currently stratified by canonical DNA markers. Contrary to their perception in the literature, the most accurate expression-based models depend on few features and are amenable to biological interpretation. This work points to the importance of exploring more comprehensive expression profiling in clinical settings.
Competing Interest Statement
T.R.G. is a paid consultant to GlaxoSmithKiline, is a co-founder of Sherlock Biosciences and FORMA Therapeutics, and receives sponsored research funding from BayerHealthCare, Calico Life Sciences and Novo Ventures. A.T. is a consultant for Tango Therapeutics. W.C.H. is a consultant for ThermoFisher, Solasta, MPM Capital, iTeos, Frontier Medicines and Paraxel and is a Scientific Founder and serves on the Scientific Advisory Board (SAB) for KSQ Therapeutics. F.V. receives funding from Novo Ventures. All other authors declare no competing interests. All authors were partially funded by the Cancer Dependency Map Consortium, but no consortium member was involved in or influenced this study.
Footnotes
New figure on the necessity of key expression features; additional discussion.
