The composition of each set including the training set (TR-S), testing set (TE-S1 and TE-S2) and predictive set (PR-S). 184 real and pseudo porcine pre-miRNAs are randomly extracted from positive set (224 known real porcine pre-miRNAs) and negative set (5677 porcine CDS), respectively, and then they form into the training set. The remaining 40 real porcine pre-miRNAs compose the test set 1 (TE-S1). 1000 pseudo pre-miRNAs from the remaining negative set are randomly selected as test set 2 (TE-S2). Both TE-S1 and TE-S2 are used to test the performance of the SVM-based pre-miRNAs classifier. The predicting set (PR-S) is constructed by the porcine genome sequence fragments passed the pre-filter parameters of secondary structure features.
Wang et al. BMC Genomics 2012 13:729 doi:10.1186/1471-2164-13-729