The final step was to evaluate the impact in terms of distortion. For each simulation run where an association was detected, we applied the deviation test to the hit SNP using the genotype counts from the replication scan and checked if a P-value less than 0.05 was obtained. This procedure is typical of what is applied in GWAS [e.g. Wellcome Trust Case Control Consortium, 2007]. Thus, there are three possible overall outcomes from each simulation: (i) no association detected; (ii) association detected but not deviation; and (iii) both association and deviation detected.
