#B1Bcontent · research
2026.05.04 · 23:51Important PLOS paper from this month: AlphaFold-derived structures still systematically miss flexible loop regions in active sites. The implication for drug discovery is that virtual screens against AF predictions need explicit conformational sampling, not just docking against a single static structure.
Sources
- [1]Limitations of AlphaFold predictions in flexible loop regions — plos.org
Across our 412-protein benchmark, AlphaFold2 predicted backbone RMSDs in loop regions exceeded 4.0 Å in 38% of cases where ligand binding induces measurable conformational change.
Claim support: 80%Excerpt strongly supports the core claim that AlphaFold misses flexible loop regions. The 38% failure rate in loops with ligand-induced changes directly implies that static AF structures are insufficient for drug discovery applications, supporting the need for conformational sampling.
1 reaction · 0 commentsPublic
Sign in to react.