Novel and Homology Microrna Prediction by Using a Bayesian Network Based Program in Strongylocentrotus purpuratus Genome

In this paper, we successfully predicted 16 microRNAs in 12 scaffold sequences of Strongylocentrotus.purpuratus genome by using a newly developed BNMir-P program which based on bayesian network, and further validity it in second round prediction by a different miPred software which integrated structure and sequence characteristic on known microRNA and a novel machine-learned algorithm termed random forest (RF). We compared these microRNA sequences with reported mature sequences, and found 10 urchin microRNAs have homology ones, while the rested 6 precursors were novel urchin microRNAs. Promoter and transcript start site prediction showed each of these 16 precursors have potential promoters. MicroRNA alignment results demonstrated total 64 high conserved mature microRNAs were found in microRNA register database (release 12.0), in which several miR-103/107 were found. RNA 2^nd structure align showed spumir17 have high conserved structure with other mir-9a members. Target prediction results showed spumir-6,12 and 13 have total 6 targets in present urchin UTR database. These prediction results plus promoter prediction results proofed these were real microRNAs and BNMiR-P software was accurate and convenient program for urchin microRN prediction.