Graph curvature flow improves drug target binding predictions
CurvFlow-DTA: dual-graph discrete Ricci curvature flow for drug--target affinity prediction
Machine Learning
Summary
Predicting how well a drug binds to a target protein helps develop new medicines. This paper presents a method that looks at detailed shapes of both drug molecules and target proteins as graphs, using a math concept called curvature flow. The new approach better captures how the drug and protein interact, leading to more accurate predictions than previous methods. It works even when the drug or protein hasn’t been seen before, which is important for finding new drug candidates.
What this means in practice
- •For drug discovery teams: Generate more accurate predictions for new drugs binding to protein targets without requiring complex structures.
- •For computational chemists: Incorporate adaptive curvature flow features on molecular and protein graphs to improve affinity modeling workflows.
Authors
Jicheng Ma, Yunyan Yang, Juan Zhao, Liang Zhao
Abstract
Graph neural networks are widely used for drug--target affinity (DTA) prediction, and discrete Ricci curvature has recently been used to characterize molecular graph geometry. Existing curvature-aware DTA approaches mainly use static curvature on the drug graph while representing proteins primarily with sequence-derived features. This leaves pair-adaptive use of graph geometry underexplored, which may limit adaptation to unseen entities in cold-start settings relevant to practical screening. We present CurvFlow-DTA, which replaces a single static curvature representation with weighted Forman curvature flow on both molecular and protein residue--residue contact graphs. A label-independent flow trajectory is precomputed for each entity, and a pair-conditioned selector determines the horizons read by a dual-branch Flow-GINE. A frozen ESM-2 supplies residue-level representations and contact scores used to construct the protein graph. Inference requires only SMILES strings and protein sequences, without a bound complex structure. On Davis and KIBA, CurvFlow-DTA improves on the protocol-matched Ricci-GraphDTA baseline in every warm and cold-start setting. Warm-split mean squared error (MSE) decreases by $19.9\%$ on Davis and $18.9\%$ on KIBA. Across the six cold-start comparisons, MSE decreases by $14.3$--$27.4\%$, with higher concordance index (CI) throughout. Within our compiled set of literature baselines, CurvFlow-DTA achieves the lowest MSE on both warm benchmarks and across four out of six cold-start evaluation settings.