Table 1.
Parameters of the agent-based model.
| Parameter | Value |
|---|---|
| Time step | 1 week |
| Simulation time | 5 million years |
| Host parametersa | |
| Maximal population size Nmax | 5000 individuals |
| MHC diversity | 2 loci, each with 15 alleles |
| Maximal number of NKR loci per haplotype | 5 |
| Bit string length | 16 bitsb |
| Host mutation rate μ (i.e., point mutation) | 0.00005 per gene per birth event |
| Probability of generating a random novel NKR | 0.1 per mutation event |
| Infectionc | |
| Infection state i | 1 (acute), 2 (chronic) |
| Effect of viral load on the death rate VLi | 0.1 (for i = 1), 0.06 (for i = 2) per year |
| Probability of viral transmission during acute phase pac | 0.85 per contact |
| Probability of viral transmission during chronic phase pch | 0.15 per contac |
| Probability of clearing the infection pcl | 0 if missing-self is not detected, or 0.6 if missing self is detected |
| Immunity time ti | 10 years |
| Acute infection time tinf | 4 weeks |
| Initial conditions | |
| Initial population size Ninit | 4500 individuals with homozygous hosts, encoding two copies of the same NKR in their gene cluster) |
aThe death and birth rate parameters are age-dependent and have been chosen according to a human population (46). For a full description of the age-dependency of birth and death rate, see Ref. (14).
bBy using 16-bit strings, a large enough theoretical repertoire of 65,536 sequences is represented.
cWe choose the parameters used for the infection such that the epidemic can be maintained. Changing the length of the acute phase or the probabilities of clearance do not affect our results on the evolution of the NKRs qualitatively (results not shown).