Abstract:

Background: MicroRNAs (miRNAs) are small ncRNAs participating in diverse cellular and physiological processes through the post-transcriptional gene regulatory pathway. Critically associated with the miRNAs biogenesis, the hairpin structure is a necessary feature for the computational classification of novel precursor miRNAs (pre-miRs). Though many of the abundant genomic inverted repeats (pseudo hairpins) can be filtered by comparative genomics techniques, novel specie-specific pre-miRs are likely to remain elusive.

Results: miPred is a newly proposed de novo Support Vector Machine (SVM) classifier for discriminating pre-miRs against pseudo hairpins without relying on phylogenetic conservation. It employs a Gaussian Radial Basis Function kernel as a similarity measure for 29 global and intrinsic hairpin folding measures. They characterize a pre-miR at the dinucleotide sequence, hairpin folding, non-linear statistical thermodynamics, and topological levels. Trained on 200 human pre-miRs and 400 pseudo hairpins, miPred achieves 5-fold cross-validation accuracy of 93.5% and ROC of 0.9833. Tested on the remaining 123 human pre-miRs and 246 pseudo hairpins, it reports 84.55%, 97.97%, and 93.50% for the sensitivity, specificity, and accuracy. Validated onto 1,918 pre-miRs across 40 non-human species and 3,836 pseudo hairpins, it yields 87.65% (92.08%), 97.75% (97.42%), and 94.38% (95.64%) for the mean (global) sensitivity, specificity, and accuracy. Notably, A. mellifera, A. Geoffroyi, C. familiaris, E. Barr virus, H. Simplex virus, H. cytomegalovirus, O. Physcomitrella patens, R. lymphocryptovirus, S. virus, and Z. mays, are unambiguously classified with 100.00% and >93.75% for sensitivity and specificity, respectively.

Conclusions: Our classifier miPred achieves significantly higher sensitivity and specificity with biologically relevant datasets, than existing (quasi) de novo predictors. It will serve as an invaluable tool to classify reliably specie-specific and evolutionary well-conserved pre-miRs without requiring any computationally expensive comparative genomics information.

Keywords: microRNAs; Minimum Free Energy of folding; Support Vector Machine;