Can AI-Predicted Complexes Teach Machine Learning to Compute Drug Binding Affinity?

Journal article

We evaluate the feasibility of using co-folding models for synthetic data augmentation in training machine learning-based scoring functions (MLSFs) for binding affinity prediction. Our results show that performance gains depend critically on the structural quality of augmented data. In light of this, we established simple heuristics for identifying high-quality co-folding predictions without reference structures, enabling them to substitute for experimental structures in MLSF training. Our study informs future data augmentation strategies based on co-folding models.

Authors: Hsu, W; Grevtsev, S; Herz, AM; Douglas, T; Magarkar, A

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: American Chemical Society
• Journal: Journal of Chemical Information and Modeling
• Volume: 65
• Issue: 24
• Pages: 13051-13056
• Publication date: 2025-12-10
• Acceptance date: 2025-12-01
• DOI: 10.1021/acs.jcim.5c01848
• EISSN: 1549-960X
• ISSN: 1549-9596
• Language: English