# EvoFlows: Evolutionary Edit-Based Flow-Matching for Protein Engineering

DevFeed: [EvoFlows: Evolutionary Edit-Based Flow-Matching for Protein Engineering](<https://devfeed.tech/articles/evoflows-evolutionary-edit-based-flow-matching-for-protein-engineering-132767.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2603.11703>)

Author: Nicolas Deutschmann, Constance Ferragu, Jonathan D. Ziegler, Shayan Aziznejad, Eli Bixby

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Sequences](<https://devfeed.tech/topics/sequences.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>)

Tags: [binder](<https://devfeed.tech/tags/binder.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [functional](<https://devfeed.tech/tags/functional.md>), [mutation](<https://devfeed.tech/tags/mutation.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [proteins](<https://devfeed.tech/tags/proteins.md>), [sequence](<https://devfeed.tech/tags/sequence.md>), [stable](<https://devfeed.tech/tags/stable.md>), [validation](<https://devfeed.tech/tags/validation.md>), [variable](<https://devfeed.tech/tags/variable.md>)

## AI overview

EvoFlows models protein optimization as variable-length edit flows between evolutionarily related proteins, predicting which mutations to make, where to make them, and how many to introduce. This supports insertions and deletions without pre-specified mutation sites, unlike the autoregressive, masked-language, and discrete-diffusion approaches described in the abstract. Across UniRef and OAS protein families, the authors report variants that remain family-consistent while exploring farther from the template than leading baselines. In vitro tests of a VHH binder and an enzyme found that most generated variants remained functional and stable without computational downselection.

## Source excerpt

arXiv:2603.11703v3 Announce Type: replace Abstract: Protein optimization improves an existing protein through targeted mutations, but existing protein language models are poorly suited to it: autoregressive models require full sequence generation, masked language and discrete diffusion models rely on pre-specified mutation locations, and none naturally support insertions and deletions relative to a template. We introduce EvoFlows, a variable-length sequence-to-sequence approach that learns mutational trajectories between evolutionarily related proteins via edit flows, predicting not only _which_ mutation to perform, but also _where_ and _how many_ should occur. Across diverse protein families from UniRef and OAS, EvoFlows generates variants that remain consistent with natural families while exploring farther from the template than leading baselines. _In vitro_ validation of a VHH binder and an enzyme confirms that the large majority of generated variants remain functional and stable without computational downselection.