Dynamics-inspired Structure Hallucination for Protein-protein Interaction Modeling

2026-01-08

Trans. Mach. Learn. Res.

AI Summary

Accurately predicting the effects of mutations on protein–protein interactions (PPIs) remains challenging due to the absence of mutant structures and inadequate modeling of dynamic interaction mechanisms. To address these limitations, this work proposes Refine-PPI, a novel framework that first generates high-quality mutant protein structures using a structure refinement module guided by a Masked Mutation Modeling (MMM)-driven structural hallucination strategy. Subsequently, it introduces a Probability Density Cloud Network (PDC-Net), which leverages geometric deep learning to model the three-dimensional dynamics of PPIs and atomic-level uncertainty. Evaluated on the SKEMPI v2 benchmark, Refine-PPI significantly outperforms existing methods in predicting binding free energy changes, demonstrating the effectiveness and innovation of integrating structural hallucination with dynamic uncertainty modeling.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Protein-protein interaction (PPI) represents a central challenge within the biology field, and accurately predicting the consequences of mutations in this context is crucial for drug design and protein engineering. Deep learning (DL) has shown promise in forecasting the effects of such mutations, but is hindered by two primary constraints. First, the structures of mutant proteins are often elusive to acquire. Secondly, PPI takes place dynamically, which is rarely integrated into the DL architecture design. To address these obstacles, we present a novel framework named Refine-PPI with two key enhancements. First, we introduce a structure refinement module trained by a mask mutation modeling (MMM) task on available wild-type structures, which is then transferred to produce the inaccessible mutant structures. Second, we employ a new kind of geometric network, called the probability density cloud network (PDC-Net), to capture 3D dynamic variations and encode the atomic uncertainty associated with PPI. Comprehensive experiments on SKEMPI.v2 substantiate the superiority of Refine-PPI over all existing tools for predicting free energy change. These findings underscore the effectiveness of our hallucination strategy and the PDC module in addressing the absence of mutant protein structure and modeling geometric uncertainty.

Problem

• Protein-protein interaction
• Mutation effect prediction
• Protein structure
• Dynamic modeling
• Geometric uncertainty

Innovation

• structure hallucination
• mask mutation modeling
• probability density cloud network
• geometric uncertainty
• protein-protein interaction