Poor Generalization by Current Deep Learning Models for Predicting Binding Affinities of Kinase Inhibitors
Wern Juin Gabriel Ong, Palani Kirubakaran, John Karanicolas
This paper shows that current deep learning models for predicting kinase inhibitor binding affinities do not generalize well to unseen data, suggesting they rely on data leakage rather than learning molecular interactions.
bioRxiv · Sep 4, 2023 · PreprintBioinformatics