A Novel Profile-HMM To Predict MicroRNAs And Their Targets Simultaneously

Junwen Wang
University of Hong Kong


Abstract

The miRNAs are small non-coding RNAs of ~22 nucleotides, and are involved in crucial biological processes, including development, differentiation, apoptosis and proliferation. Through interaction with 3í UTR of target genes, the miRNAs can affect the targetís expression by transcriptional or post-transcriptional regulation. In this talk, I will introduce a novel profile-HMM we developed recently to predict miRNAs and their targets simultaneously. We used the hidden states to represent the base frequencies of miRNA and used the emission states to represent the base frequencies of the targets. With its novel HMM structure, our model captures both miRNAs and targets conservations. Trained and tested on experimental validated microRNA-Target pairs, the model showed superior performance over current leading microRNA-target interaction program -RNAhybrid. Furthermore, our model is capable of discovering novel microRNA and targets simultanously. Finally, the model automatically discovered the conservation of miRNA seed region, and provided insights on the conservation of other regions.