Figure 3. Fitness loss and recognition losses of escape mutations are predicted to correlate positively, with a slope that deceases in time.

Sites are randomly sampled from the simulation described in the caption to Figure 2 in order to simulate the effect of acute samples (high ranking sites), late chronic samples (low ranking sites) or patient samples from random times (random ranks). (A) No CTL decay. The slope of the correlation from acute (blue) or late chronic (red) sampled escape mutations is the same, however, it is lower for escape mutations sampled at random times (green). (B) CTL decay causes the slope of the correlation to decrease in time, due to decrease of CTL selection pressure. CTL decay is introduced as described in the caption to Figure 2 and Model.