Predicting the Toxicity of Druggable Proteins to Human Tissues

Abstract

Assessing in vivo tissue toxicity of therapeutic targets remains a major challenge in drug development and drug safety research. We developed TissueTox, an algorithm that learns from multi-omic features of a target protein and predicts toxicity in human body systems and tissues. Predicted TissueTox scores accurately differentiate drugs that failed clinical trials from those that succeeded, and, importantly, can be used to identify the tissues where toxic events occurred.