7th Conference

L5 Timing the Origin of the HIV-1 Pandemic. B. KORBER^*1, M. MULDOON², J. THEILER¹, F. GAO³, A. LAPEDES¹, R. GUPTA¹, B. HAHN³, S. WOLINSKY⁴, T. BHATTACHARYA¹. ¹Los Alamos Natl. Lab., NM; ²Univ. of Manchester Inst. of Technology, UK; ³Univ. of Alabama, Birmingham and ⁴Northwestern Univ., Chicago, IL.

HIV-1 circulating in the world today is extremely diverse. Genetically associated forms of the virus called subtypes, as well as intersubtype recombinant viruses, can dominate the epidemic in particular geographic regions. We have extended phylogenetic analysis methods to estimate the timing of the origin of the HIV-1 pandemic. It is important to establish the onset of the AIDS epidemic in order to better understand the possible routes and circumstances of zoonotic transmission, as well as how rapidly HIV-1 evolves in human populations. An implementation of maximum likelihood tree building code designed for parallel supercomputers was developed to allow optimization of the evolutionary model for comprehensive sets of sequences with known years of sampling (160 env and 70 gag sequences). Two basic approaches were employed to estimate the time of origin from these trees. The first assumes a constant rate of evolution throughout the period under consideration, incorporates biologically motivated error models for time and branch length, and uses a bootstrap for estimating confidence intervals. The second strategy allows for variation in the rate of evolution. We validated our methodologies by using them to accurately estimate the timing of two historically documented points in the trees: the sampling time of oldest available HIV sequence (1959), and the origin of the HIV epidemic in Thailand (1986-1987). Applying these methods to both the gag and env trees, we estimated the time of origin of the HIV-1 Main (M) group to be near 1930, with 95% CIs essentially spanning the period from 1910 to 1950.