[Paper Review] Spin models inferred from patient data faithfully describe HIV fitness landscapes and enable rational vaccine design

Karthik Shekhar, Claire Ruberman+4 more|arXiv (Cornell University)|Jun 9, 2013

HIV Research and Treatment41 references87 citations

TL;DR

This paper demonstrates that spin models inferred from patient-derived HIV sequences—despite being based on non-equilibrium, immune-driven evolution—faithfully reproduce the intrinsic fitness landscape of HIV, enabling accurate prediction of viral fitness rank order. The study combines simulations and variational theory to show that population-level immune diversity enables comprehensive sampling of sequence space, making prevalence-based models effective for rational vaccine design by identifying mutational vulnerabilities.

ABSTRACT

Mutational escape from vaccine induced immune responses has thwarted the development of a successful vaccine against AIDS, whose causative agent is HIV, a highly mutable virus. Knowing the virus' fitness as a function of its proteomic sequence can enable rational design of potent vaccines, as this information can focus vaccine induced immune responses to target mutational vulnerabilities of the virus. Spin models have been proposed as a means to infer intrinsic fitness landscapes of HIV proteins from patient-derived viral protein sequences. These sequences are the product of non-equilibrium viral evolution driven by patient-specific immune responses, and are subject to phylogenetic constraints. How can such sequence data allow inference of intrinsic fitness landscapes? We combined computer simulations and variational theory \'{a} la Feynman to show that, in most circumstances, spin models inferred from patient-derived viral sequences reflect the correct rank order of the fitness of mutant viral strains. Our findings are relevant for diverse viruses.

Motivation & Objective

• To determine whether fitness landscapes inferred from patient-derived viral sequences—shaped by non-equilibrium immune selection—can accurately reflect intrinsic viral fitness.
• To resolve the paradox of how non-equilibrium sequence data can yield reliable fitness predictions despite lack of equilibrium sampling.
• To establish conditions under which spin models based on mutational correlations in patient sequences faithfully represent the true fitness landscape of HIV.
• To provide a mechanistic explanation for the success of maximum entropy models in predicting in vitro fitness and clinical outcomes.

Proposed method

• The authors use computer simulations of viral evolution in a host network, modeling transmission between genetically diverse hosts with patient-specific immune pressures.
• Each host's viral quasispecies evolves via a non-equilibrium mutation-selection process over discrete generations, with fitness determined by an effective Hamiltonian combining intrinsic fitness and host-specific immune fields.
• The maximum entropy principle is applied to infer spin models from multiple sequence alignments (MSAs), using one- and two-site mutational probabilities to construct a Boltzmann distribution with a Hamiltonian of the form H₀[⃗s] = ΣJij s_i s_j + Σh_i s_i.
• Variational theory based on Feynman’s approach is used to derive self-consistent equations for model parameters, ensuring consistency between inferred model statistics and simulated data.
• The fitness landscape is evaluated by comparing the rank order of predicted strain energies (H₀[⃗s]) with actual in vitro fitness measurements and simulated intrinsic fitness.
• The method accounts for phylogenetic correlations between generations using a time-averaged effective Hamiltonian, enabling inference from non-equilibrium data.

Experimental results

Research questions

• RQ1 Can spin models inferred from non-equilibrium, patient-derived viral sequences accurately reflect the intrinsic fitness landscape of HIV?
• RQ2 Why does the prevalence landscape—derived from immune-driven, non-equilibrium evolution—correlate with intrinsic fitness measured in vitro?
• RQ3 Under what conditions does the maximum entropy model fail to recover the true fitness rank order?
• RQ4 How does population-level genetic diversity in immune responses enable comprehensive sampling of viral sequence space?

Key findings

• Spin models inferred from patient-derived HIV sequences accurately reproduce the rank order of intrinsic viral fitness across mutant strains, even though the data arise from non-equilibrium, immune-driven evolution.
• The robust correlation between model-predicted energies (H₀[⃗s]) and in vitro fitness measurements is explained by the fact that diverse host immune pressures collectively sample the full sequence space of viral proteins.
• The presence of genetically diverse immune responses across the population ensures that mutational correlations in patient data reflect underlying fitness constraints rather than local immune selection alone.
• The study shows that maximum entropy models are not merely statistical tools but can capture the true fitness landscape when the population-wide immune pressure is sufficiently diverse and non-uniform.
• The variational theory framework successfully reconciles non-equilibrium sequence data with equilibrium statistical mechanics, validating the use of H₀[⃗s] as a proxy for intrinsic fitness.
• The model’s predictive power is maintained even when compensatory mutations exist, as the collective fitness penalty of multiple mutations is captured by the inferred J_ij couplings.