Contrastive learning of PLM embeddings on functional annotation improves variant effect prediction and homolog detection

Summary

Contrastive learning on mean-pooled protein language model embeddings improves variant effect prediction and homolog detection (1). Using the different ESM2 models, authors show that larger models were slightly better, but unclear if statistically significant.

Figures

Figures from (1)

1.

Wu KE, Chang H, Zou J. ProteinCLIP: enhancing protein language models with natural language. openRxiv; 2024. Available from: https://doi.org/10.1101/2024.05.14.594226