A comprehensive study of a SVM-based miRNA target prediction algorithm

MicroRNAs are single-stranded non-coding RNAs that play important regulatory roles in many biological processes and diseases. Identifying miRNA regulatory targets is paramount in elucidating its function. We carried out a comprehensive study of a new SVM-based target prediction algorithm called SVMicrO in this paper. The training data set is carefully derived from the most up-to-date collection of verified targets and multiple microarray data sets. Several varieties of feature design and selection schemes are investigated. The prediction results are compared with most of the existing algorithms, which show improved sensitivity and specificity of this two-stage SVM algorithm.