Summary statistics additional signatures identified with additional searches, using the combined original and PD/DB sets to identify potential signatures and comparing to the holdout set. For legend see table 2.

aOne new acute signature site was identified through restricting the search to just the homogeneous early infection samples, !T415. This association was significant only for a grouping of the earliest samples, from Fiebig stages 1 and 2. Three sites in addition to site 12 (already included in Table 2) were strongly supported signatures of recurrent change in the chronic subjects using full tree analyses. One combination of sites was found through more intensive examination of the functional domain sets. It was found in the CCR5 CoRbs model, defined based on a heavy-atom based distance criterion to identify the proximal amino acids to the CCR5 CoRbs.bRegion explored for Ref-1 HXB2 amino acid and positions, complex signature positions in bold; Ref-2 refers to the alignment position given in parenthesis. Q114 (133), L122 (141), I201 (271), Q203 (273), A204 (274), S209 (279), N377 (463), Y384 (470), A436 (546) and P437 (547)