Randomization scheme used for the semi-synthetic Miller data set. We started from eight non-related previously acquired exome sequences and shuffled the sample identifiers linking a variant to a sample. By doing this the total number of variants and the number of variants per consequence was kept constant but the constitution of each of the exomes changed continuously. We then assigned four exomes to both control and case groups. In the case group we added two of the reported causal variants to each of the case samples. We then analyzed each of the randomizations with Annotate-it and looked at the resulting rank of DHODH. By repeating this randomization cycle 1,000-fold we calculated rank statistics.