Mining Molecular Structure Data for the Patterns of Interactions Between Protein and RNA

Mining useful information from a large amount of biological data is becoming important, but most data mining research in bioinformatics is limited to molecular sequence data. We have developed a set of algorithms for analyzing hydrogen bond and van der Waals interactions between protein and RNA. Analysis of the most representative set of protein-RNA complexes revealed several interesting observations: (1) in both hydrogen bond and van der Waals interactions, arginine has the highest interaction propensity, whereas cytosine has the lowest interaction propensity; (2) side chain contacts are more frequent than main chain contacts in amino acids, whereas backbone contacts are more frequent than base contacts in nucleotides; (3) amino acids, in which side chain contacts are dominant, reveal more diverse interaction propensities than nucleotides; and (4) valine rarely binds to any nucleotide. The interaction patterns found in this study should prove useful for determining binding sites in protein-RNA complexes.