{"id":"arxiv:2506.05596","kind":"preprint","title":"Zero-shot protein stability prediction by inverse folding models: a free energy interpretation","authors":["Jes Frellsen","Maher M. Kassem","Tone Bengtsen","Lars Olsen","Kresten Lindorff-Larsen","Jesper Ferkinghoff-Borg","Wouter Boomsma"],"author_count":7,"year":2025,"published":"2025-06-05","venue":null,"cited_by":null,"doi":null,"arxiv_id":"2506.05596","nct_id":null,"source_url":"https://arxiv.org/abs/2506.05596","archive_url":"https://space-kitty.com/papers/arxiv:2506.05596","updated":"2025-10-27","pmid":null,"pmcid":null,"openalex_id":null,"pdf_url":"https://arxiv.org/pdf/2506.05596","abstract":"Inverse folding models have proven to be highly effective zero-shot predictors of protein stability. Despite this success, the link between the amino acid preferences of an inverse folding model and the free-energy considerations underlying thermodynamic stability remains incompletely understood. A better understanding would be of interest not only from a theoretical perspective, but also potentially provide the basis for stronger zero-shot stability prediction. In this paper, we take steps to clarify the free-energy foundations of inverse folding models. Our derivation reveals the standard practice of likelihood ratios as a simplistic approximation and suggests several paths towards better estimates of the relative stability. We empirically assess these approaches and demonstrate that considerable gains in zero-shot performance can be achieved with fairly simple means.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["cs.LG","cs.AI","q-bio.BM","stat.ML"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[{"slug":"protein-design","name":"AI Protein Design & Structure Prediction","domain_slug":"biotech"}],"cited_in_dossiers":[]}