Carlos Outeiral

This paper introduces ProtEGNN, a model combining protein language models with geometric equivariance to improve protein property and function prediction with fewer parameters than fine-tuned PLMs.

The paper shows that codon-based language models outperform amino acid sequence models in protein engineering tasks like species recognition and melting point estimation.

This study compares protein structure prediction methods to experimental folding data, finding their simulated dynamics do not better capture folding pathways than simple classifiers.