Property guidance for protein sequence generative models with ProteinGuide

AI Summary

ProteinGuide provides a principled, on-the-fly framework to condition protein sequence generative models on auxiliary data without retraining.
Method is compatible with masked language models, any-order autoregressive models, and discrete diffusion or flow-matching models such as ESM3, ProteinMPNN, MultiFlow.
Demonstrated designs for enhanced stability or activity, simultaneous optimisation of competing properties, and improved adenine base editor in vivo efficiency surpassing seven directed evolution rounds.

TL;DR (AI Generated)

ProteinGuide is amenable to a broad class of protein generative models including masked language models such as ESM3, any-order autoregressive models such as ProteinMPNN and diffusion and flow-matching models on discrete state-spaces such as MultiFlow.

Lastly, we apply ProteinGuide jointly with wet-lab data generation to increase the editing activity of an adenine base editor in vivo, resulting in a base editor with higher editing efficiency than was previously achieved using seven rounds of directed evolution.

Basic summary

Nat Biotechnol. 2026 Jul 29. doi: 10.1038/s41587-026-03207-z. Online ahead of print.

ABSTRACT

No principled framework exists for conditioning sequence generative models for protein engineering on auxiliary information, such as experimental data, without additional training of a generative model. Here we present ProteinGuide, a method for such ‘on-the-fly’ conditioning. ProteinGuide is amenable to a broad class of protein generative models including masked language models such as ESM3, any-order autoregressive models such as ProteinMPNN and diffusion and flow-matching models on discrete state-spaces such as MultiFlow. ProteinGuide stems from a unifying statistical framework for these model classes. As proof of principle, pretrained generative models are used to design proteins with user-specified properties, such as higher stability or activity. Proteins are additionally designed to optimize for two desired properties that are in tension with each other. Lastly, we apply ProteinGuide jointly with wet-lab data generation to increase the editing activity of an adenine base editor in vivo, resulting in a base editor with higher editing efficiency than was previously achieved using seven rounds of directed evolution.

PMID:42527525 | DOI:10.1038/s41587-026-03207-z

Document this CPD