MiRNA-Disease Association Prediction Using Game Optimization and Multi-view Representation Learning

MiRNAs influence cellular functions by regulating gene expression and interacting with diverse biomolecules within the cell. Accurate prediction of miRNA disease associations (MDA) plays a crucial role in dis ease diagnosis, treatment, and drug development. However, most computational methods focus on graph information and ignore other information when extracting miRNA and disease features. Therefore, we propose a computational method based on game optimization and multi view representation learning (GMIMDA) for predicting MDA. Specifically, GMIMDA first integrates multi-source similarity views using the similarity network fusion (SNF) technique and constructs multi-view complementary representations by extracting linear, nonlinear, and graph structure representations of miRNA and disease via singular value decom position with weighted scaling (SVD-WS), nonnegative matrix factorization (NMF), and graph convolutional networks (GCNs), respectively. Second, to enhance the interactivity between nodes, GMIMDA uses graph features as initial strategies of participants in game theory modeling and selects optimal strategies based on the Nash equilibrium. Finally, GMIMDA fuses the outputs of multi-view representation learning and inputs them into the Kolmogorov-Arnold Networks (KANs) for making predictions, using symbolic activation functions to reveal the mapping relationship between input and output. Experimental results demon strate that GMIMDA outperforms existing approaches in three different databases. Case studies provide additional evidence supporting the effectiveness of GMIMDA in identifying unknown disease-associated miRNAs in practical applications.