MiRPara: a SVM-based software tool for prediction of most probable microRNA coding regions in genome scale sequences

MiRPara was developed as an effective tool for locating miRNAs coding regions in genome sequences and can be used as a screening step prior to HTS experiments. Goal is to predict most probable mature miRNA coding regions from genome scale sequences in a species specific manner. The authors classified sequences from miRBase into animal, plant and overall categories and used a support vector machine to train three models based on an initial set of 77 parameters related to the physical properties of the pre-miRNA and its mature miRNAs. By applying parameter filtering they found a subset of ~25 parameters which produced higher prediction ability compared to the full set. MiRPara achieves an accuracy of up to 80% against experimentally verified mature miRNAs.

The program has been written in perl so it will run on any operating system. Query sequences can be submitted via their site, or you can download the code and run it yourself on your own computer. Originally released in November 2009, it is currently available as version 2.0. It uses the older miRBase version 13 as positive dataset.

MiRPara is available at http://www.whiov.ac.cn/bioinformatics/mirpara

MiRPara: a SVM-based software tool for prediction of most probable microRNA coding regions in genome scale sequences.
Wu Y, Wei B, Liu H, Li T, Rayner S.
BMC Bioinformatics. 2011 Apr 19;12(1):107.
PMID: 21504621 [PubMed – as supplied by publisher]