HELM-BERT: Peptide Property Prediction with HELM Notation
Published:Dec 29, 2025 03:29
•1 min read
•ArXiv
Analysis
This paper introduces HELM-BERT, a novel language model for predicting the properties of therapeutic peptides. It addresses the limitations of existing models that struggle with the complexity of peptide structures by utilizing HELM notation, which explicitly represents monomer composition and connectivity. The model demonstrates superior performance compared to SMILES-based models in downstream tasks, highlighting the advantages of HELM's representation for peptide modeling and bridging the gap between small-molecule and protein language models.
Key Takeaways
Reference
“HELM-BERT significantly outperforms state-of-the-art SMILES-based language models in downstream tasks, including cyclic peptide membrane permeability prediction and peptide-protein interaction prediction.”