This is an archived static version of the original discussion site.

[Paper] The Genealogical Population Dynamics of HIV-1 in a Large Transmission Chain: Bridging within and among Host Evolutionary Rates


Very interesting work from @arambaut, @alexei_drummond, @guy_baele, and @philippe_Lemey.

Methodologically, this paper provides long-overdue inferential methods that distinguish (the rather different) within-host and between-host evolutionary processes, as well as integrate epidemiological information. I was wondering if the authors could say if this method would improve inference with data sets for which we have no epidemiological information, just stratification of sequences by host.

Biologically, it has been known for a while that viruses from chronic infections have certain mutations that are not present in primary infection. Here the authors provide further evidence that infectious HIV derives from “stored” virions rather than mutated virion lineages which have reverted. This builds on the work of:


Here is a note from the first author Bram Vrancken from @philippe_Lemey’s group responding to my question:

We think modeling the transmission bottleneck does not really improve inferences, and certainly not when it cannot be pinpointed who infected who (a situation in which we cannot specify the transmission model and thus fall back on other standard coalescent models). If there is improvement, most of it will likely come from the rate modeling (without epidemiological info we can set up a similar backbone effect model between patients): the improved rate variation modeling should in principle result in more accurate inferences on the divergence times and perhaps even clustering. I’ve discussed this with Philippe and he doesn’t expect a large effect for both though.


I just wanted to note that the above paper has interesting consequences for this one:

as in changes our interpretation of what coalescent trees look like in transmission.