VIGOR, an annotation program for small viral genomes

Wang, Shiliang; Sundaram, Jaideep P; Spiro, David

The results were evaluated by comparison between the gene finder predictions and annotations in GenBank. Sensitivity (Sn) and Specificity (Sp) were employed to evaluate the performance of the programs tested. Sn is defined as the percent correct predictions out of all annotated genes in the dataset. Sp is defined as the percent correct predictions out of all predictions. Correct predictions are those which are the same as the Gen-Bank annotations. Any prediction that was not identical to the GenBank annotation was inspected manually by similarity searching against the NCBI non-redundant (NR) protein database. If the new prediction was highly similar (E value <1e-10) to a viral protein spanning 95% of the homologous protein length, it was categorized as a correct prediction. A partially correct prediction was assigned when the prediction overlapped with the Gen-Bank annotation in the same reading frame but with a different start codon. An incorrect prediction was assigned when the prediction overlapped with the Gen-Bank annotation but in a different reading frame, or cannot be validated by sequence similarity.