Evolutionary rate shifts suggest species-specific adaptation events in HIV-1 and SIV

Maoz Gelbart; Adi Stern

harnessed to identify adaptation events of emerging pathogens to their new host species. Our 260 results suggest that innate immunity serves as a strong barrier for cross-species transmission 261 events, and that this barrier imposed a strong selective pressure for viruses to adapt as they 262 crossed these barriers with increasing efficiency. 263 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/190769 doi: bioRxiv preprint Methods 264 In order to collect sequences for this study, the Los Alamos HIV sequence database (available 265 online at http://www.hiv.lanl.gov [19] ) was queried for HIV-1 sequences from the same strain 266 that spanned all nine HIV-1 open reading frames (ORFs: gag, pol, vif, vpr, tat, rev, vpu, env Initial multiple sequence alignments of the nine proteins were performed using PRANK and 281 iteratively improved until convergence [59] . In order to reconstruct of the phylogenetic 282 relationship between the sequences, we concatenated the alignments of Gag, Pol, Vif, Vpr, Tat 283 and Env and provided this as input for PhyML [60] . We next used the reconstructed phylogeny 284 as a guide tree to realign each protein with PRANK. JpHMM was used to validate that the strains 285 used in the analysis are not inter-group recombinants [61] . 286 . CC-BY-NC-ND 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi. org/10.1101/190769 doi: bioRxiv preprint In order to identify evolutionary rate shifts, we used RASER [18] to analyze each of the nine 287 proteins separately, with the proteome-based phylogeny as input. RASER is a likelihood-based 288 phylogenetic method for detecting a change in site-specific evolutionary rates. First, a likelihood 289 ratio test against a null model of no rate shifts is performed in order to assess if a model 290 enabling rate shifts better fits the data. Next, the posterior probability of rate-shift is calculated 291 at each site and sites with a probability higher than 0.6 are considered here as significant. 292