As a third, independent method to estimate burden of de novo variants in TD probands versus SSC Siblings, we performed a Fisher exact test that “normalizes” by the number of de novo synonymous mutations (Sanders et al., 2012). We chose this method because we hypothesized that the number of de novo synonymous mutations per individual should be unrelated to phenotypic status, and therefore, could potentially control for batch effects affecting the rate of de novo variant detection. For this analysis, we constructed a 2×2 contingency table from the counts of de novo mutations of a particular class in probands and SSC sibling controls, and the counts of de novo synonymous mutations in probands and SSC sibling controls. A one-sided Fisher exact test then estimated the odds ratio and p value for each class of mutation. For example, for de novo LGD variants in the combined cohort the contingency table is:
