Rank-and-Reason: Multi-Agent Collaboration Accelerates Zero-Shot Protein Mutation Prediction
This addresses the inefficiency and subjectivity in manual expert auditing for protein engineering, particularly in low-resource settings, by providing an automated solution that enhances prediction reliability and wet-lab outcomes.
The paper tackled the problem of zero-shot protein mutation prediction by introducing Rank-and-Reason, a multi-agent framework that automates expert auditing and improves prediction accuracy, achieving a Spearman correlation of 0.551 on ProteinGym and up to a 367% increase in Top-5 Hit Rate on ProteinGym-DMS99, with wet-lab validation showing a 46.7% positive rate and novel mutants with 4.23-fold and 5.05-fold activity improvements.
Zero-shot mutation prediction is vital for low-resource protein engineering, yet existing protein language models (PLMs) often yield statistically confident results that ignore fundamental biophysical constraints. Currently, selecting candidates for wet-lab validation relies on manual expert auditing of PLM outputs, a process that is inefficient, subjective, and highly dependent on domain expertise. To address this, we propose Rank-and-Reason (VenusRAR), a two-stage agentic framework to automate this workflow and maximize expected wet-lab fitness. In the Rank-Stage, a Computational Expert and Virtual Biologist aggregate a context-aware multi-modal ensemble, establishing a new Spearman correlation record of 0.551 (vs. 0.518) on ProteinGym. In the Reason-Stage, an agentic Expert Panel employs chain-of-thought reasoning to audit candidates against geometric and structural constraints, improving the Top-5 Hit Rate by up to 367% on ProteinGym-DMS99. The wet-lab validation on Cas12i3 nuclease further confirms the framework's efficacy, achieving a 46.7% positive rate and identifying two novel mutants with 4.23-fold and 5.05-fold activity improvements. Code and datasets are released on GitHub (https://github.com/ai4protein/VenusRAR/).