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

We next aligned the collection of sequences; a non-trivial task given the vast amount of sequence data. The alignment of protease in the Stanford drug resistance database is well maintained, so we assumed that this portion of the alignment is accurate. Note that this is why we remove the Los Alamos database copy where there is overlap between the Stanford drug resistance database and Los Alamos sequence databases. In the case of the sequences taken from the Los Alamos HIV sequence database, we find that attempting to generate an alignment when downloading ∼70,000 sequences yields nonsensensical results. However, there is a lack of indels in the protease region. We therefore simply considered sequences of exactly 297 nucleotides for the alignment (after removal of gaps from the alignment proposed using the Los Alamos database default alignment software). Performing this step led to a loss of <500 sequences out of a total of ∼70,000. We then checked the resulting distribution of nucleotides across the region. Finally, the UCLA positive selection database was assumed to be aligned correctly.