Data-driven target discovery and validation for cancer therapy

Abstract

Cancer is one of the leading causes of human death, requiring novel therapeutics for the treatment. Despite continuous efforts to conquer cancer, many clinical trials still fail due to efficacy and side effect issues. To reduce the time and costs of the development of anti-cancer drugs, we developed two approaches to the data-driven identification and validation of therapeutic targets for cancer. We introduced intercellular CRISPR screen, which combines two genome-wide CRISPR screen datasets, each in cancer and immune cells, to identify regulators of immune cell function. Our results showed that intercellular CRISPR screens can identify well-known modulators of cytotoxic T cells and suggested seven novel intercellular interactions as the potential targets of immunotherapy. To validate targets for targeted therapy, we developed GESIGAN, a generative adversarial network model for predicting context-specific gene expression signatures upon gene knockouts. We used the predicted gene expression signatures for the validation of cancer therapeutic targets via reversal of cancer gene expression signatures. We envision that data-driven approaches to drug discovery will help to reduce the time and cost required for the development of novel cancer therapeutics.