Parameters and variables of the (standard) model.

Notes: (1) is chosen larger than observed numbers of immune responses [25], [27], [43], since we predict that viruses have escape mutations at infection, and do not escape all CTL responses. is chosen to get reasonable variance in spVL, while limiting individuals with a very small binding repertoire. (2) About of all possible peptides from HIV-1's proteome of a.a. (3) During the chronic phase, the escape rate slows down markedly [27], [60], hence we take . The AIDS phase is sometimes preceded by escape from critical immune responses [75], and modeling suggests that escape rate speeds up towards the late disease phase [46]. Therefore we set . (4) Both reports on fast [76] and very slow [77] reversion exist. We choose in the order of magnitude of the ratio fitness cost and escape benefit. (5) The model for virus load was taken from [35]. During the acute phase, merely represents the virus fitness. (6) The magnitude is chosen to be in estimated ranges [31], [32]. Since escape appears to be faster than reversion, we choose . Although several studies find that a CTL response to Gag gives a fold higher fitness cost than [29], [32], we take as an average fitness cost. (7) The parameters and were taken from [17]. The parameters for the Hill functions and are: , , , , , .Parameters and variables of the (standard) model.