Mining graph patterns in the protein-RNA interfaces

Protein-RNA interactions play important roles in the biological systems. The goal of this study is to discover structural patterns in the protein-RNA interfaces that contribute the affinity of the interactions. We represented known protein-RNA interfaces using graphs and then identify common subgraphs enriched in the interfaces. Comparison of the discovered graph patterns with UniProt annotations showed that the graph patterns had a significant overlap with residue sites that had been proven by experimental methods to be crucial for RNA bindings. Using 200 patterns as input features, a Support Vector Machine method was able to classify protein surface patches into RNA-binding sites and non-RNA-biding sites with 84.0% accuracy and 88.9% precision. We built a simple scoring function that calculated the total number of the graph patterns that occurred in a protein-RNA interface. That scoring function was able to discriminate near native protein-RNA complexes from docking decoys with a performance comparable with a state-of-the-art complex scoring function.