Luiz Carlos Vieira

The paper evaluates protein language models' performance on mutational effect prediction across datasets and finds that dataset features, not model architecture, primarily drive predictive success.

This study identifies a fold switch in ONNV nsP4 that regulates its conformation, RNA binding, and polymerase activity through proteolytic cleavage and fold switching.

Medium-sized protein language models (e.g. ESM C 600M) with mean embeddings nearly match ESM-2 15B and ESM C 6B in transfer learning on mutational scanning and PISCES data.

The paper evaluates a language-of-viral-escape model using multiple machine learning frameworks to identify immune escape mutations in viruses.