Aligning protein-generative models to experimental fitness with ProteinDPO

Summary

Nature Methods, Published online: 14 August 2026; doi:10.1038/s41592-026-03137-3 This Article demonstrates that direct preference optimization (DPO) can be used to effectively align an unsupervised structure-conditioned language model with biophysical information. The aligned model, ProteinDPO, achieves stability prediction competitive with that of task-specific models and consistently outperforms unsupervised and fine-tuned versions of the model.