Learning the Interaction Prior for Protein-Protein Interaction Prediction: A Model-Agnostic Approach

Protein-protein interactions (PPIs) are fundamental to cellular function, disease mechanisms, and drug discovery. Current learning-based PPI predictors focus on learning powerful protein representations but neglect designing specialized classification heads. They mainly rely on generic aggregating methods like concatenation or dot products, which lack biological insight. Motivated by the biological "L3 rule", where multiple length-3 paths between a pair of proteins indicate their interaction likelihood, our study addresses this gap by designing a biologically informed PPI classifier. In this paper, we provide empirical evidence that popular PPI datasets strongly support the L3 rule. We propose an L3-path-regularized graph prompt learning method called L3-PPI, which can generate a prompt graph with virtual L3 paths based on protein representations and controls the number of paths. L3-PPI reformulates the classification of protein embedding pairs into a graph-level classification task over the generated prompt graph. This lightweight module seamlessly integrates with leading PPI predictors as a plug-and-play component, injecting the interaction prior of complementarity to enhance performance. Extensive experiments show that L3-PPI achieves superior performance enhancements over state-of-the-art competitors.

Proteins work by grabbing onto each other, and predicting which pairs interact is vital for drug discovery — but running lab experiments for every possible pair is too slow and costly. AI models have gotten good at "reading" individual proteins, but struggle with a surprisingly naive step: most just glue two protein descriptions together to decide if they'll bind, ignoring what biology tells us about how proteins actually recognize each other. Biology offers a useful clue: interacting proteins tend to share many mutual "partner" proteins in the interaction network — much like two people with many mutual friends are likely to know each other. Our module, L3-PPI, teaches any existing AI predictor to think this way by generating a small virtual network between candidate protein pairs and checking whether it resembles patterns seen around truly interacting pairs. L3-PPI clips onto existing models without retraining them, adds minimal computational cost, and consistently improves accuracy — especially for proteins never seen during training, which matters most in real drug discovery settings.