Phylogenetic analyses

HIV-1 genetic subtype was defined by REGA automated genotyping tool (version 2.0) and by phylogenetic inference [34,35]. Study sequences were aligned with reference set from the Los Alamos HIV database (ww.hiv.lanl.gov) using the Clustal X software and adjusted manually by the BioEdit software [33,36]. Phylogenetic inferences were performed by the Neighbor-Joining (NJ) method under Kimura’s two-parameter correction using MEGA6 software [37,38]. Bootstrap values (1000 replicates) above 70% were considered significant. Isolates with discordant PR/RT subtypes were analyzed by SIMPLOT 3.5.1 software (200 bp sliding window, advanced in 20 bp steps size increments, 1000 replicates) [39]. Sequences were gap-stripped, the transversion-to-transition ratio was set to 2.0, and distances were calculated according to Kimura’s two-parameter mode. The clustering pattern of Northern Brazilian sequences analyzed here and reference sequences was investigated by performing maximum-likelihood (ML) phylogenetic analyses. HIV-1 subtype B pol sequences from Northern Brazil were aligned with subtype B references sequences representative of the B_PANDEMIC clade and major B_CAR clades circulating in the Caribbean and South America described previously [13,40–42]. The HIV-1 subtype F1 pol sequences from Northern Brazil were aligned with subtype F1 references sequences representative of Brazilian (F1_BR), Romanian and Central African clades described previously [43]. The HIV-1 subtype C pol sequences from Northern Brazil were aligned with subtype C references sequences representative of Brazilian (C_BR), Eastern African, Southern African and Central African clades described previously [44,45] The BF1 recombinants sequences here identified were aligned with all URFs_BF1 and CRFs_BF1 of Brazilian origin available at Los Alamos HIV Database. ML phylogenetic trees were reconstructed with the PhyML 3.0 program [46] under the best nucleotide substitution model, selected by the SMS (Smart Model Selection) software [47] integrated into the PhyML web server. The SPR branch-swapping algorithm was selected for heuristic tree search and the approximate likelihood-ratio test (aLRT) [48] to estimate the reliability of the tree topology obtained. The GenBank accession numbers of the sequences presented in this study are {"type":"entrez-nucleotide-range","attrs":{"text":"MH673055-MH673281","start_term":"MH673055","end_term":"MH673281","start_term_id":"1573780433","end_term_id":"1573780881"}}MH673055-MH673281.