Autonomous agent improves protein complex prediction by customizing sequence data
Auto-Research Agent on MSA Organization to Improve Protein-Protein Complex Structure Prediction
Artificial Intelligence
Summary
Predicting how proteins stick together is important but tricky because it depends on the data used about similar proteins. The authors show that using the same data approach for every protein pair is not the best way. They built an AI agent that tests different ways of organizing this data for each protein pair, improving accuracy and using less data. This method works especially well when the proteins come from different species, where previous methods struggled.
What this means in practice
- •For protein engineers: Increase accuracy of protein-protein interaction models by customizing sequence alignment data during prediction.
- •For drug discovery teams: Improve structural modeling of protein complexes from different species to enhance target identification.
Authors
Zilu Zhang, Jing Bao, Lifeng Qiao, Sheng Xu, Qiantai Feng, Siqi Sun, Yang Yang
Abstract
Protein-complex structure prediction with AlphaFold 3 depends on MSA content, depth, and cross-chain pairing. Current pipelines apply uniform recipes, typically maximizing depth and pairing by species, yet even whether cross-chain pairing helps AF3 remains contested. We show that the value of both depth and pairing is strongly target-dependent, so no single recipe is optimal across interfaces. We introduce an autonomous agent that treats a frozen AF3 as a black-box experimental instrument. For each target, the agent formulates biological hypotheses, analyzes and reorganizes the MSA, and iteratively refines sequence selection and pairing using predictor label-free signals alone. On 1{,}015 protein--protein interfaces from three benchmarks released after AF3's training cutoff, the agent improves overall interface DockQ over AF3's default pipeline and fixed MSA recipes on every benchmark, while using on average 60\% fewer MSA sequences than the default. Gains are largest for inter-species heteromers, where species-based pairing is unreliable: mean DockQ increases from 0.465 to 0.612 on FoldBench and from 0.482 to 0.562 on HD439. The improvements arise from both sequence construction and pairing policy, including deliberate non-pairing.These results recast MSA design as a target-specific problem and show that an autonomous agent can adapt it at test time through biological reasoning and self-directed, confidence-guided experimentation.