John Mellors paper– set point viral load is an important predictor of progression to disease.
What determines it? Researchers have done GWAS and found that 20% Viral factors– some things have been found, but no comprehensive screen.
What about heritability? Define it as a quantitative measure of the statistical associate of set point virus load in donor-recipient pairs. In evolutionary genetics, h^2 = the proportion of phenotypic variance that can be explained by additive
Heritability: donor-recipient correlation. Using index-to-secondary transmission studies, get basically zero heritability.
Using phylogenetic contrasts: using Bloomberg’s K, Pagel’s lambda, and heritability h^2, show a much higher degree of heritability. Citing Alizon et al 2010 PLOS Pathogens 2010.
Setpoint virus load and its evolution. Patients vary 3-4x in their set point load. Very fast turnover of virus. High viral load kills off patient but is more infectious.
We have all the prerequisites for fast evolution: phenotypic variation, link to fitness, and high heritability. Paradox: little intrahost evolution.
Analytical model: are observed levels of mean and variance of virus load compatible with observed high heritability? Population of HIV carriers, -> donors selected for transmission -> recipients at transmission -> population. Seems that we should be progressing towards higher viral load. Work of Christoph Fraser, PNAS 2007.
How is the set point VL distributed in carriers? Log set point VL is approximately normally distributed. Mean is 4-5 logs, variance is 0.5-0.75 logs. Have weak selection.
Have maximal transmission at intermediate level of viral load. A subset of the viruses in the host are transmitted.
Model: decompose phenotype into host and virus effects. Model each with normal distributions.
Change of virus load over course of disease. In Shankarappa, increasing viral load through time, though effect is weak. Question of McVean -> perhaps this was an unusual subset?
Where are we with respect to the overall picture. Can solve for mean and variance of the equilibrium viral load, and for heritability.
Which parameter regions are compatible with observed range of heritability, and observations of mean and variance of viral load? There is a subset, and considering the effect of the strength of selection and the intrahost evolution.
Search for compatible parameters.
Generally estimates are hard to reconcile with each other. Conclusion: controlled to a certain amount by host genotype, but weak selection.
A model of target cell activation (Bonhoeffer, Trends Microbiol, 2003). Something about the virus activates target cells to a different extent.
[To be honest, I couldn’t completely follow their lines of argument, which is why the notes are incomplete here.]