For definiteness, in the results presented below we simulate data under the additive disease model, and in analysis of the data consider each SNP separately and apply the so-called trend, or Cochran-Armitage test [16], a chi-squared test with one degree of freedom. We fix a significance level of 5×10−7, and vary the number of cases and controls in the simulated study. There are various other versions of these assumptions which could be made. We explicitly look at one set of multi-marker tests below and also carry-out a limited set of simulations to assess the impact of genotype imputation.
