The abstract is read at the publisher; the summary is JClub's.
MGCA predicts protein-ligand dissociation kinetics (Koff) with the lowest mean squared error on the Zhao dissociation-kinetics dataset and KinetX benchmark, including warm-start and protein cold-start settings.
- Why it matters: Existing computational approaches largely target binding affinity and do not explicitly model dissociation kinetics, leaving Koff prediction both important for residence-time optimization and difficult to achieve accurately.
- What they did: MGCA uses multi-depth ESM2 protein features and multi-radius Morgan fingerprints for ligands, then combines gated expert fusion with low-rank bidirectional cross-attention and protein-anchored residual fusion.
- The result: MGCA reduced protein cold-start MSE by 42.5% on Zhao and 12.1% on KinetX versus BiCoA-Net, trains substantially faster under matched hardware, and ablations support effective integration for cross-target generalization and compound ranking.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.