Consensus RNA Secondary Structure Prediction Based on SVMs

Although many endeavors have been done in the field of RNA secondary structure prediction, it is still an open problem in the computational molecular biology. The comparative sequence analysis is the golden standard method when given homologous sequence alignment. The essential of this method can be regarded as classifier problem: to judge whether any two columns of an alignment correspond to a base pair using provided information by alignment. Here, we employ SVMs to resolve this classifier problem, and select the covaration score, the fraction of complementary nucleotides and the consensus probability matrix as the feature vectors. Test on the Rfam shows that average MCC of our method is higher (0.841) than KnetFold (0.831), Pfold (0.741) and RNAalifold(0.623).