To investigate convergence in the genital tract, Satish and his colleagues first needed to determine whether genital HIV strains form distinct lineages — that is, in the phylogeny of all the HIV viruses that live within an infected individual, do genital strains form a clade?

The researchers focused on 12 people for whom both blood and semen-derived viral sequences had been collected and were available on GenBank (a public database of genetic sequences) — about 30 blood sequences and 20 seminal sequences per person. The focus was on the env sequence (short for envelope glycoprotein), a stretch of HIV RNA that encodes part of the mushroom-shaped spikes on the surface of the virus. They reconstructed the phylogenetic relationships among viral strains, building a tree of env sequences for each patient.

On some patients’ HIV trees, it was easy to see clades of viruses from semen — for example on Individual A’s tree below. In contrast, Individual J’s genital viruses are scattered over the tree. To make the comparisons rigorous, the researchers performed some statistical calculations. On each tree, they mapped which viruses were from semen and counted the number of times that viruses must have moved between the genital tract and the blood. Then they randomly generated 1000 trees, mapped on the same data about which viruses came from semen, and counted the number of migration events on each. If the real tree had fewer migrations than 99% of the random trees, the researchers took that as evidence of restricted gene flow between the semen and the blood. All in all, in half of the patients, HIV in the genital tract was isolated from that in the blood. Why would the there be a division between blood and semen in some individuals but not others? Satish suspects that infections that trigger immune system cells to move into the genital tract (e.g., gonorrhea) are likely to cause mixing between HIV strains from the semen and blood.