MutexaGPT: An Intuition-to-Design Translator for Physics-based Enzyme Engineering.

Authors

Qianzhen Shao, Yinjie Zhong, Sebastian Stull, Xinchun Ran, Ning Ding, Kieran Nehil-Puleo, Ruizhe Yao, Han Xu, Zhongyue Yang

Published in

Research square. Jun 19, 2026. Epub Jun 19, 2026.

Abstract

Physical intuitions about how enzyme structure and dynamics influence its function and property have enabled successful engineering outcomes, yet a systematic approach remains unknown for translating these qualitative and abstract "thoughts" into quantitative, actionable principles that lead to designs. Here we introduce MutexaGPT, an open-access, multi-agent large language model (LLM) platform that translates enzyme engineering intuition to physics-based simulations and thus variant designs. Through a web interface, MutexaGPT takes users' intuition-driven requests (expressed in plain English) as input, and then leverages its LLM agents (i.e., QuestionAnalyzer, WorkPlanningBoard, and ResultExplainer) to comprehend the request and elicit missing information, construct physics-based models, configure and execute high-throughput molecular modeling workflows, and eventually convert the molecular modeling results into actionable design proposals (such as a smart mutation library). An automated evaluation framework was established to systematically benchmark the prompt engineering strategies that allow each individual agent to achieve an optimal performance. We further demonstrate the utility of MutexaGPT in two protein engineering tasks. For the task of engineering halide methyltransferase towards bulkier substrates, MutexaGPT converted a cavity-engineering intuition into a smart library design that shows a 40% hit rate and around 4-fold activity improvement over a baseline strategy. For the task of engineering bidomain amylase for enhanced activity at lower temperature, MutexaGPT translated a statistics-based intuition into cold-adapted amylase variants that show 1.7-fold and 3.7-fold activity enhancement at 0 °C (experimentally validated). These results establish MutexaGPT as an intuition-to-design translator that integrates human creativity with high-throughput molecular modeling to democratize physics-guided, intuition-driven enzyme engineering.

PMID:
42370243
Bibliographic data and abstract were imported from PubMed on 29 Jun 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.