Evaluation Of MicroRNA Target Prediction Programs And Analysis Of The Features Of Target Gene

MicroRNA, which regulates the post-transcriptional expression of target genes, is an important class of small RNA. Identifying the target genes of certain microRNA is fundamental to the understanding of its regulatory functions. Owing to the fact that experimental validation of target gene is highly time-consuming, computational prediction of microRNA target genes becomes one of the most effective methods for target gene identification. Although a number of target-gene prediction programs have been developed, the evaluation criteria for the predicting performance varied significantly. In this regard, the author carried out the research about assessing the genuine predicting performance of several prevalent prediction programs and about analyzing the regulatory features of the target genes.First of all, the author employed the golden-standard criteria (sensitivity, false positive rate and noise ratio) to tackle the inconsistency in the use of different performance-evaluation standards of five most popular programs, and tested their predicting performances on the reliable human microRNA target gene datasets from Tarbase. The assessment results suggest that the prediction accuracy of those programs is still at a low level. The best program yielded the sensitivity of 0.443 and 0.332 respectively when predicting the conserved and nonconserved target genes. Based on the former results and the characteristics of those programs, the author put forward a practical procedure for users with specified purpose to select the proper programs.Then, the transcriptional features of the target genes were further analyzed. With the gene annotation information from database of Ensembl and Vega, the author carried on an analysis on the number of transcripts of target genes and found that the microRNA target genes were highly inclined to generate multiple splice variants. The p-value for this propensity is as low as 1.43E-51. Furthermore, the author analyzed the correlation of target genes to multiple polyA sites. And likewise, the p-value is as low as 9.27E-25. The analysis results reveal that some certain alternative splicing mechanism is relevant to the transcription process of target genes.