Prediction of viral microRNA precursors based on human microRNA precursor sequence and structural features

Table 2 Comparison of the number of viral microRNAs predicted by the 3 ab-initio prediction algorithms-BayesmiRNAfind, mir-abela, mirSVM and our method on a dataset of viral microRNA precursors derived from mirBase

Virus expressed microRNAs derived from miRbase	PREDICTIONS
Virus	Bayes miRNAfind	Mir-abela	tripletSVM	MODEL 031
Epstein-Barr Virus (23)	21 -91.3	18 -78.26	21 -91.3	22 -95.65
Cytomegalovirus (11)	10 -90.91	5 -45.45	7 -63.63	10 -90.91
Kaposi Sarcoma Herpesvirus (13)*	9 -69.23	5 -38.46	8 -61.54	11 -84.62
Herpes Simplex Virus (2) **(sequences were derived from literature)	2 100%	0 0%	0 0%	2 100%

The numbers denote the total number of positive predictions and numbers in brackets following the organism name denotes the total number of microRNA hairpins in mirRbase and that following the positive predictions denote the percentage of total predicted. (*)KSHV expresses only 12 microRNA hairpins. The 13^th sequence is one with a possible single nucleotide editing that was cloned. (**) For HSV, the sequences were derived from respective literature as the version of miRbase did not yet include the sequences.

ISSN: 1743-422X