A mean-field computational approach to intra-host HIV mutational dynamics

During HIV infection, the intra-host population is attacked by host cytotoxic T lymphocyte (CTL) responses. CTLs kill HIV-infected cells by recognizing certain HIV peptides presented on their surface. HIV strains with mutations in these regions escape CTL recognition and hence gain a relative fitness advantage. Can the dynamics of HIV mutations during intra-host infection be predicted? The intrinsic fitness landscapes of various HIV proteins, describing replicative capacity as a function of amino acid sequence, have been inferred from the global prevalence of strains infecting diverse hosts. Here, we present a new method to compute intra-host HIV mutational dynamics given an intrinsic fitness landscape and CTL responses, which we designate the evolutionary mean-field (EMF) method. EMF is a high-recombination-rate model of HIV dynamics that outputs effective fields and frequencies of mutations at each residue over time. We show via an example how intrinsic fitness costs and epistatic effects, skewed CTL responses, etc. impact the identities and time course of HIV escape mutations. We also explain features of longer-term dynamics using the effective fitnesses yielded by EMF. Finally, we extend EMF to stochastic population dynamics and quantify stochasticity in infection outcomes.