A New Ontology-Based Method for Functional Composed Comparison of MicroRNAs

Abstract

MicroRNAs (miRNAs) are small non-coding RNA molecules that negatively regulate gene expression, playing critical roles in many relevant biological processes. Since there are no terms of miRNAs annotation in Gene Ontology (GO) nor a database with miRNA functional annotation, the direct computation of functional similarity between miRNAs cannot be done under an established standardized approach. However, a miRNA can be annotated with a set of information, such as if it acts as oncogene or as tumour suppressor, the organism that it belongs, its association with diseases, target genes, proteins and pathological events. This way, the similarity between two miRNAs can be inferred based, for example, in the relative position of their respective target genes in GO. In this study, we propose and evaluate CFSim, a method that uses GO and the disease ontology MeSH to compute miRNAs composed similarity by combining different information related to them. We validated CFSim by examining functional similarity values inferred intra and inter miRNA families, and the results showed that our method is efficient in sense that the functional similarity between miRNAs in the same family was higher compared to other miRNAs from distinct families. Furthermore, in comparison with existing methods for functional similarity, CFSim is more effective in distinguishing miRNA families.