We also note that the apparent size of the disease class is not a good guide as to the number of causal variants responsible. Even if the disease class is quite large, it is easily possible that it consists of only rare variants if there are a relatively large number of rare causal variants and these could be spread very broadly over genome regions stretching into the megabases. With respect to the size of the disease class, our simulations highlight the counterintuitive result that, under some genetic models, increases in the number of causal mutations at a locus can increase the probability of a synthetic association.
