SemiGNN-PPI: Self-Ensembling Multi-Graph Neural Network for Efficient and Generalizable Protein–Protein Interaction Prediction

Abstract

Protein-protein interactions (PPIs) are crucial in various biological processes and their study has significant implications for drug development and disease diagnosis. Existing deep learning methods suffer from significant performance degradation under complex real-world scenarios due to various factors, e.g., label scarcity and domain shift. In this paper, we propose a self-ensembling multi-graph neural network (SemiGNN-PPI) that can effectively predict PPIs while being both efficient and generalizable. In SemiGNN-PPI, we not only model the protein correlations but explore the label dependencies by constructing and processing multiple graphs from the perspectives of both features and labels in the graph learning process. We further marry GNN with Mean Teacher to effectively leverage unlabeled graph-structured PPI data for self-ensemble graph learning. We also design multiple graph consistency constraints to align the student and teacher graphs in the feature embedding space, enabling the student model to better learn from the teacher model by incorporating more relationships. Extensive experiments on PPI datasets of different scales with different evaluation settings demonstrate that SemiGNN-PPI outperforms state-of-the-art PPI prediction methods, particularly in challenging scenarios such as training with limited annotations and testing on unseen data.

Effectiveness and efficiency: label-aware hierarchical subgraph learning for protein-protein interaction

“…More recently, graph-based MLL approaches have gotten great attention from scientific communities [27,28]. Especially, the SemiGNN-PPI [17]…”

“…Furthermore, the latest works utilize graph neural networks (GNN) to model the PPI network[12]. Among these works, the SemiGNN-PPI[17] establishes a multi-graph learning framework to improve the model performance, and the HIGH-PPI[18] organizes the structure-based internal graph for each protein to enhance the interpretability of the model. However, all the PPI prediction models face a topological shortcut problem due to annotation imbalance.…”

“…More recently, graph-based MLL approaches have gotten great attention from scientific communities[27, 28]. Especially, the SemiGNN-PPI[17] successfully makes use of Graph Convolutional Network (GCN) by constructing a directed graph over labels to explicitly model the protein-protein interaction dependencies adaptively. In this regard, we add graph-based MLL to hierarchical graph learning for more accurate PPI prediction.…”

“…Compared with classic ML methods, the DL algorithm has an advantage in processing complicated and large-scale data and extracting useful features automatically, achieving huge success in computational biology[13], including PPI prediction[14]. With the development in graph neural networks (GNNs), more recent works focus on modeling the PPIs as a graph, by constructing a PPIs graph with proteins as nodes and interactions as edge types, with various GNNs [15, 16] utilized to explore the PPIs graph to facilitate inference of type of interactions[12, 17, 18]. Among these graph-based methods, the HIGH-PPI reaches state-of-the-art performance with interpretability.…”