Mapping the drivers of within-host pathogen evolution using massive data sets

Palmer, Duncan S.; Turner, Isaac; Fidler, Sarah; Frater, John; Goedhals, Dominique; Goulder, Philip; Huang, Kuan-Hsiang Gary; Oxenius, Annette; Phillips, Rodney; Shapiro, Roger; Vuuren, Cloete van; McLean, Angela R.; McVean, Gil

Such a simple alignment procedure was not available to us for the reverse transcriptase sequences due to the increased prevalence of indels within reverse transcriptase, coupled with the longer sequence length. We reduced the size of the collection of sequences to be aligned by assuming sequences taken from the Stanford drug resistance database were correctly aligned. Aligning all the remaining sequences at once using MUSCLE [32] was prohibitive computationally, as was using the align option in MUSCLE (which can be used to combine alignments: this procedure also tended to produce artefacts in the overall alignment). We therefore considered the approach of aligning subsets of 5000 sequences and comparing these sub-alignments. We found that although more common than within protease, indels are rare. Removing them resulted in a collection of sub-alignments that aligned with both each other and the consensus sequences for both the Stanford drug resistance database and UCLA positive selection database alignments.