in the protein-protein interaction network. With the rapid advances in biotechnology, largescale PPIN is currently available and is already rich enough to evaluate the relationship between miRNAs based on their targeting propensity in PPIN. Here, based on the above notion, we proposed a novel computational method, called miRFunSim, to quantify the associations between miRNAs in the context of protein interaction network. We evaluated and validated the performance of our miRFunSim method on miRNA family, miRNA cluster data and experimentally verified miRNA-disease associations. Further comparison analysis showed that our method is more effective and reliable as compared to other existing similar 183204-72-0 chemical information methods, and offers a significant advance in measuring the associations between miRNAs. The high throughput protein-protein interaction data were obtained from Wang��s study consisting of 69,331 interactions between 11,305 proteins, which integrated BioGRID, IntAct, MINT, HPRD and by the Co-citation of text mining databases and made further filtering to improve coverage and quality of PPIN and reduce false-positives produced by different prediction algorithms in different databases. The functional similarity score between miRNAs may be generated by chance. In order to take this effect into account and obtain the statistical significance of scores, we GSK-1120212 performed randomization test and repeated 1000 times. For each score, 1000 simulated miRNA pairs were generated and target genes of simulated miRNA pairs were randomly sampled from all human protein-coding genes keeping the same size as given miRNA pairs. Then the functional similarity scores between simulated miRNA pairs were recomputed for each simulated miRNA pair denoted SFSSM. M denoted the number of simulated miRNA pairs having an equal or larger SFSSM value than the true score. The estimate of the empirical statistical significance value, P-value, of true score was obtained as P =M/1001. The empirical P-value based on such randomizations represented the probability of obtaining a score greater than a given score by chance. In this study, we developed a graph theoretic property based method, miRFunSim, to quantify