Abstract:

 This paper develops a computational procedure named MiRfilter to predict the virus miRNAs. MiRfilter selects some important biological features of miRNAs as parameters to predict the potential precursors and miRNAs. When using two samples to test MiRfilter’s accuracy, it successfully retrieves 30 EBVencoded miRNAs from 39 known ones, and predicts 9 false positive miRNAs from 116 negative ones. Then it retrieves 15 RLCV-encoded miRNAs from 20 known ones and predicts 2 false miRNAs from 108 negative ones. Further, when using another species to test MiRfilter’s specificity, it predicts none false positive miRNA from 186 negative ones in D.melanogaster and C.elegans. 

Key words:  virus miRNA; MiRfilter; biology genome; hairpin structure; biological features