Fig. 3
From: Grouping miRNAs of similar functions via weighted information content of gene ontology
Our method for clustering miRNAs. The raw input to the method consists of data of miRNAs, the target genes of the miRNAs, and the GO graphs of the miRNAs’ target genes. Firstly, the similarity of two GO graphs is computed according to the weighted information content. Secondly, the similarity of two target genes of two miRNAs is calculated by using the similarity of their GO graphs. Thirdly, the similarity between two miRNAs is computed by their similarity of their target genes. Then the miRNA distance/similarity matrix is constructed by these similarity scores. Finally, the spectral clustering method is applied to cluster the miRNAs