Researchers have developed zPocket, a novel method for predicting protein-ligand binding sites, also known as pockets. This approach leverages frozen features from co-folding models and is trained on a new dataset called PocketWatch, designed to evaluate generalization to unseen pockets. zPocket significantly outperforms existing methods like VN-EGNN, UniSite-1D, and P2Rank in predicting binding sites, achieving a higher Area Under the Precision-Recall Curve (AUPRC). The method also shows promise in improving the recall of predicted pocket centroids and has potential applications in fragment screening and large-scale proteome analysis. AI
IMPACT This research could accelerate drug discovery and understanding of protein function by improving the accuracy and efficiency of binding site prediction.
RANK_REASON The cluster contains a research paper detailing a new method and dataset for a specific scientific task. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
