RipplePLM: Structural and Property Decoupling for Protein Mutation Effect Generation

2026-10-01

AI Summary

This study addresses the limitation of existing protein mutation effect prediction models in systematically organizing structural perturbation and biochemical property evidence. To this end, we propose RipplePLM, a framework built upon pretrained protein language models that innovatively incorporates a direct-distal cross-attention mechanism and a property latent chain to achieve structured and attribute-aware decoupled modeling of mutation information. Experimental results demonstrate that the proposed method yields significant improvements in the ROUGE-L metric. Furthermore, expert evaluations confirm its superior biological accuracy, effectively validating the advantage of the dual modeling strategy in learning robust mutation representations.

Abstract

Protein mutation effect generation asks a model to describe the functional consequence of a point mutation in natural language. Existing protein-to-text systems typically encode mutation information into undifferentiated representations, overlooking the organization of mutation-induced evidence across structural and biochemical factors. We propose RipplePLM, a mutation-aware generation framework centered on Direct-Distal Cross-Attention (DDCA). By constructing a residue-level Mutation Perturbation Field from pre-trained protein language models, DDCA leverages predicted contact maps to organize mutation representations into two pathways: the mutation site's immediate contact neighborhood and its multi-hop distal context. To complement this structural decomposition, we further introduce the Property Latent Chain (PLChain), which injects expert-guided supervision of biochemical property changes (e.g., thermostability and optimal pH) into the LLM hidden-state pathway through latent property tokens. On MutaDescribe, RipplePLM improves over mutation-specific baselines on temporal and structural splits; under a matched-backbone comparison, average structural-split ROUGE-L increases from {22.23} to {35.65}. Expert evaluation further shows a higher proportion of biologically accurate or relevant descriptions than the mutation-specific baseline. Additional ablations, representation diagnostics, and low-$N$ fitness regression experiments further support the effectiveness of the learned mutation-aware representations. Code: https://github.com/Lyu6PosHao/RipplePLM.

Problem

• Protein mutation effect generation
• Protein-to-text
• Structural and property decoupling
• Point mutation

Innovation

• Protein Mutation Effect Generation
• Direct-Distal Cross-Attention
• Mutation Perturbation Field
• Property Latent Chain
• Structural and Property Decoupling