MicroRNA binding site scoring model

MicroRNAs(miRNAs) are small, non-coding, endogenous RNA molecules that play an important role in genetics by regulating the expression of target genes. Gene expression regulation is due to its binding to specific location of messenger RNA(mRNA). In case of plants, effective prediction of miRNA interaction on mRNA is possible, as targets needs strict sequence complementarity with that of miRNA, but in animal systems it is difficult due to lack of complementarity. We introduce a scoring based miRNA target prediction system which gives better accuracy than the existing systems. Our system model consists of two stages. The first stage is to build a classifier using a database of miRNA and experimentally validated targets and non targets. Each mRNA target sequence is aligned with miRNA sequence using Smith-Waterman algorithm and 23 features are extracted. A Random Forest based classifier gives an accuracy of 95.56% which is better than the existing models. For a given mRNA and miRNA sequence pair, the system predicts the possible binding location and corresponding score of the binding using the trained classifier. This is cross validated with the validated site and we obtained a correct result.