Parallel Monte Carlo Simulation of HIV Molecular Evolution in Response to Immune Surveillance - Parallel Computing for Bioinformatics and Computational Biology - page 53

high degree of confidence in the parameter values and relatively complete knowledge
of the interacting components, then this approach may also be used to predict the
evolution of HIV-1 in response to specific scenarios of immune surveillance. Unan-
swered questions that may be addressed with this approach include: (1) What is the
form of the neutralizing antibody response to V3? (2) Does adaptation to chemokine
coreceptors impede adaptation to antibody surveillance? (3) Does adaptation to anti-
body surveillance change the coreceptor usage phenotype? (4) DoesV3 always evolve
to escape neutralization regardless of the cost to function? (5) Are cell superinfec-
tion and recombination important determinants of the rate of adaptation? For this
approach to be successful, values for several parameters or parameter sets must be
refined.
2.7.1 Refining Parameter Values
2.7.1.1 Intrahost HIV-1 Population Size The question of what is the popula-
tion size of HIV-1 infecting a patient, either in terms of infected cells or proviruses, is
an important one. This is related to the question of whether there is substantial popu-
lation structure or compartmentalization within a host. A large population with little
structure is conducive to adaptive evolution because such a population will produce
a large number of competing mutations, the raw material for selection.
2.7.1.2 Functional Selection Coefficients The functional selection coeffi-
cients of amino acids at specific sites of V3 are unknown. In this version of the model,
we assume that the relative sizes of these coefficients are related to the observed
frequencies of the amino acids at specific sites in each phenotype. This assumption,
however, does not address the magnitudes of these coefficients. The magnitudes of
selection coefficients will have to be measured empirically, and until such measures
become available, a scaling factor can be applied to coefficients to produce reason-
able rates of adaptation to coreceptors, perhaps based on data on adaptation to a new
coreceptor in culture.
2.7.1.3 The Form of the Neutralizing Antibody Response The dynam-
ics of the neutralizing antibody response to HIV-1 are poorly understood. It seems
reasonable to assume that the maturation of antibody affinity is described by a curve
resembling logistic growth. The two unknown parameters of the logistic equation, the
population size (potency) at time zero and the intrinsic growth rate (for the purposes
of this model, the carrying capacity is 1 by definition), could in principle be measured
in vivo. For now, these parameters may be adjusted to give an antibody response
whose tempo matches that observed in patients [23, 24]. Other questions about the
form of the neutralizing antibody response also remain: Is there an epitope frequency
threshold above which antibody production is stimulated? How frequently must this
stimulation occur? How narrow is the antibody response?And, probably the most vex-
ing question: What is the form of humoral memory of HIV-1 epitopes? Answers to
these questions obviously require empirical investigation, but again parameter values