ProfhEX: Empowering Early Drug Discovery with Machine Learning-Based Target Profiling and Liability Prediction.

The drug discovery process is inherently lengthy, complex, and costly, with high attrition rates driven by safety concerns, limited efficacy, and regulatory barriers. AI-driven computational methods have become crucial in accelerating this process by enabling the prediction of molecular activities, identification of off-target interactions, and prioritization of candidates for drug repurposing. However, existing ligand-based prediction tools often suffer from limited data coverage, narrow target scopes, and usability challenges. Here, we present an enhanced version of ProfhEX, a scalable and user-friendly platform designed for comprehensive drug-target activity profiling. The updated platform features 969 predictive models spanning 693 human targets, trained on over 5 million curated bioactivity data points. ProfhEX demonstrates high predictive accuracy in prospective real-world scenarios and surpasses state-of-the-art tools in primary target prediction benchmarks. ProfhEX represents one of the largest and most accurate platforms for compound-target prediction, supporting early stage drug discovery and enhancing target liability assessment.