Research On MiRNA Target Prediction Algorithm And MiRNA Function Machnism

MicroRNA(miRNA) is a kind of non-coding RNA which affects the proliferation and death of cells, and the target prediction of miRNA is helpful for understanding the mechanism by which miRNA acts. Based on perfectly matching in seeds, free energy and conservation, many algorithms have been published for the target prediction of miRNA. But because little is known about the way of miRNA binding its target, the sensitivity and specificity of the algorithms for target prediction are not satisfied. In this thesis, the sensitivity is improved by modifying the existed algorithm and analyzing the function of seed, free energy and conservation for the target prediction of miRNA.The main contributions are as following:1. A new miRNA target prediction algorithm fix-microRNA is presented. In this method firstly, the dynamic algorithm is used to find all probable targets for miRNAs just on the basis of the complementarity of nucleotides, and then these probable targets are filtered by perfectly matching of seeds, binding free energy and conservation.2. Fix-microRNA is tested. By using the real targets of miRNAs in the TarBase, the sensitivity is compared between miRanda and fix-microRNA and it is proved that fix-microRNA can predict more real targets than miRanda with the nearly equal quantity of predicted targets. At the same time, it is testified that fix-microRNA can predict the influence of different kinds of miRNA. And by analyzing the signal-noise ratio and the relationship between the length of 3'UTR and its number of targets being predicted, it is showed that fix-microRNA can predict miRNA targets no-randomly.3. By analyzing the signal-noise ratio, it is shown that with requiring conservation, the binding of the first and second nucleotide from miRNA 5' end can make the prediction algorithm get good specificity, and miRNA targets can be predicted no-randomly just by taking the nucleotides 1-8 from miRNA 5' end as seed.