Hedvig Tordai

pLM-SAV predicts pathogenic SAVs using delta-embeddings from protein language models, achieving high performance on labeled data and improving ambiguous predictions over AlphaMissense.

This study uses HS-AFM and MD simulations to show that T7 bacteriophage tail fibers have flexible regions enabling dynamic host recognition and surface exploration.

The paper evaluates AlphaMissense's performance in predicting pathogenicity of missense variants across diverse protein groups and introduces a web resource for access.

The paper collects and predicts structures of ABC transmembrane proteins using AlphaFold2 and presents them via a web application.