Prioritization And Network Analysis Of MicroRNA Target Genes

microRNAs(miRNAs) are a class of small(19-23nt) endogenous non-coding RNAs known to play important roles in human cell proliferation, metabolism and diverse diseases. mi RNAs don’t encode any proteins. They down-regulate gene expression through sequence-specific binding to the 3’untranslated regions(3’UTR) of target mRNAs. Information about miRNA targets can be used for the study of complex RNA regulatory networks, disease diagnosis and pharmacogenomics.Generally,there are two ways to recognize the targets of miRNA. Molecular biology experiment is the basic way to identify targets of mi RNAs, but it is limited by factors including the complex and time-consuming operation process, so better methods for the identification of miRNA targets are urgently needed. Several computational target prediction approaches, such as Targetscan, microT, miRanda, miRTarget, Pictar and PITA, have been developed to predict target genes. These methods are largely based on characteristic of miRNA seed region such as sequence matches, G-U wobble and thermodynamic duplex stability. However, these feathers can not be used to predict miRNA targets accurately, and the false positive rate is usually high. In recent years, with the accumulation of verified targets, new feathers of genes targeted by the same miRNA have been found, making it possible to develop better methods to prioritize the targets of miRNA. In this study, we developed a method to prioritize the targets of miRNA called TarpriGO. The method included three parts: the first part was based on the hypothesis that genes targeted by the same miRNA tended to have similar or related function, and we ranked the candidate target genes base on their functional scores measured by GO; the second part was based on the fact that different targets prediction methods had their own advantages but had small overlaps, we used a mul-source-based weight approach to calculate the combine score, and ranked the targets; in the end, we combined the two scores and got the final score of TarpriGO. Leave-one-out cross validation had proved that the new method to be successful in identifying verified targets, achieving an AUC score up to 0.93. Validation in high-throughput data proved that TarpriGO had ability to identify potential targets. In comparison with other methods,TarpriGO had an outstanding performance in verified targets. In comparison with traditional ranking from traditional prediction methods, verify targets ranking improved significantly. Through the analysis of best weight of traditional target prediction methods, we found that Pictar and Targetscan were better than other methods, because more than 65% weight greater than the average weight 1/6.One mi RNA may regulate multiple genes, and one gene may be regulated by a variety of miRNAs, such correlation between miRNAs and their targets help to construct a complex but precise regulation network. Analyzing the miRNA-targets network can help us to understand the mechanism of miRNA in diseases. In this paper, we constructed miRNA-disease network and miRNA-target networks with miRNAs and targets associated to drug addiction. For the further analysis, we found miRNAs playing a central role in complex disease, in a way that one miRNA can regulate multiple genes associated withthe disease. At the same time, multiple miRNAs involved in the addiction of the same drug may work in a synergetic way. On the basis of regulatory networks, we selected four possible miRNAs related to nicotine addiction, i.e., hsa-miR-124-3p, hsa-miR-16-5p, hsa-miR-155-5p and hsa-mi R-222-3p. Each of them had several verified targets related to nicotine addiction, and all of them were verified be related to other drug addiction excepted nicotine. We further analyzed the function of target genes with Gene Ontology and KEGG, and found that most of them were related to nicotine mechanism.In conclusion,based on the analysis of mi RNAs and target genes, we developed a method to prioritize targets of miRNAs; then we constructed and analyzed the miRNA-target network related to nicotine addiction.