VADR: validation and annotation of virus sequence submissions to GenBank

Alejandro A Schäffer; Eneida Hatcher; Linda Yankie; Lara Shonkwiler; J Rodney Brister; Ilene Karsch-Mizrachi; Eric P Nawrocki

For each sequence S, the model M (S) is used to determine the sequence coverage. This is achieved again using the HMMER3 filter pipeline as implemented in Infernal 1.1.3, but now using the full pipeline that reports local matches in each input sequence, with more precise endpoints than the abbreviated pipeline used in the classification stage. For sequences that include internal short stretches of sequence that are dissimilar from the RefSeq, multiple matches may be returned at this stage, with the dissimilar regions not covered by any of the matches. The coverage of S is determined as the fraction of nucleotides in S that occur in any of the alignments on the top strand (+ strand). An alert (lowcovrg) is reported for any sequence with coverage below 0.9. Additional alerts can be reported at this stage (Table 3 ) based on unexpected characteristics of the set of returned alignments. For example, the indfstrn alert is reported if at least one alignment with a score of 25 bits or more occurs on each strand.