The abstract is read at the publisher; the summary is JClub's.
OmniCleave, a structure-aware geometric graph learning framework, models protease–substrate specificity across 100+ proteases, trained on 57,278 pairs, and predicts 21 caspase-3 cleavage sites.
- Why it matters: Protease recognition depends on spatial constraints and higher-order relationships that motif- or sequence-only models may miss, limiting accurate, transferable specificity modelling.
- What they did: OmniCleave was trained on 57,278 structure-informed protease–substrate pairs from 9,651 substrates spanning over 100 proteases across six families, using multi-scale geometric graphs with higher-order protease relational topology.
- The result: On large-scale benchmarks it consistently outperforms prior approaches and yields interpretable geometric determinants; experimentally, it confirms three previously unannotated caspase-3 substrates and 21 predicted cleavage sites.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.