of frequency 0.005–0.02 is not symmetric). When baseline disease probability was 0.01, a single GRR of 2 corresponds to a variant explaining 0.15% of the variance in liability and a GRR of 6 corresponds to 1.2% of the variance in liability. For 9 causal variants the variance explained is approximately 9 times the value for a single locus, i.e., 1.4% for GRR = 2 and 13% for GRR = 6. Table 4 shows how the probability of disease in the population increases under the Dickson et al. model, which fixes the baseline (no risk alleles) probability of disease; for the model with baseline probability of disease of 10% for 9 causal variants each of GRR = 6, the actual probability of disease in the population is 19%. For more natural benchmarking, we have undertaken the calculations fixing the probability of disease in the population. Assuming a disease with a heritability of 0.8, we calculate the maximum number of these 100 kb loci possible in the genome if all the genetic liability variance were attributable to loci like this, as this provides an upper limit to the contribution of such loci. The maximum number of loci can be high (maximum 543,
