miTarget: microRNA target gene prediction using a support vector machine

Abstract

Background: MicroRNAs (miRNAs) are small noncoding RNAs, which play significant roles as posttranscriptional regulators. The functions of animal miRNAs are generally based on complementarity for their 5' components. Although several computational miRNA target-gene prediction methods have been proposed, they still have limitations in revealing actual target genes.

Results: We implemented miTarget, a support vector machine (SVM) classifier for miRNA target gene prediction. It uses a radial basis function kernel as a similarity measure for SVM features, categorized by structural, thermodynamic, and position-based features. The latter features are introduced in this study for the first time and reflect the mechanism of miRNA binding. The SVM classifier produces high performance with a biologically relevant data set obtained from the literature, compared with previous tools. We predicted significant functions for human miR-1, miR-124a, and miR-373 using Gene Ontology (GO) analysis and revealed the importance of pairing at positions 4, 5, and 6 in the 5' region of a miRNA from a feature selection experiment. We also provide a web interface for the program.

Conclusion: miTarget is a reliable miRNA target gene prediction tool and is a successful application of an SVM classifier. Compared with previous tools, its predictions are meaningful by GO analysis and its performance can be improved given more training examples.

Figure 1

General scheme of miRNA:mRNA interactions.

Figure 2

Three categories of SVM features.

Figure 3

The ROC curves of classifiers created on three combinations of features: an entire set (circles), position-based features only (asterisks), and without position-based features (plus symbols). The red rectangle denotes the performance of TargetScan, the green one shows the performance of RNAhybrid, and the blue one shows the performance of miRanda.

Figure 4

A subgraph of the GO-directed acyclic graph (DAG) to show functional relationships among the statistically significant GO terms for the target genes of miR-124a. The gray vertexes denote statistically significant GO terms based on a hypergeometric distribution. The numbers in brackets denote the numbers of genes annotated to the GO term. The dotted circle shows the best GO term.

Figure 5

Comparisons between a random negative data set and an original negative data set. (a) The plots show the performance of the original (circle) and random (plus) classifiers on the original test data set. (b) The plots show the performance of the original (circle) and random (plus) classifiers on the random test data sets.

Figure 6

Changes in performance according to the numbers of features selected. The rectangle shows the ROC curve of the classifier created with the top feature, the asterisk (*) line is for the top five features, the plus symbol (+) line is for the top 10, the 'x' line is top 15, and the circle line is for the complete feature set.