Since resequencing of candidate genes in a larger cohort is an important validation step in evaluation of any candidate gene, we screened a larger independent cohort of whole exome data from 418 autism cases and 371 controls, sequenced as part of the ARRA Autism Sequencing Consortium. DNA from these cases and controls underwent whole exome capture, cloning and sequencing in the same fashion that our 16 cases did at the Broad Institute. For all four genes, we compared the rate of mutations under a recessive model, looking for either homozygous or compound heterozygous mutations in cases versus controls. As a group, the 4 genes showed a higher number of recessive mutations (homozygous or compound heterozygous) in cases (24/418, 5.7%) compared to controls (11/371, 3.0%) (P = 0.042, Fisher's exact test, one-tailed). These mutations were all missense changes and were relatively rare, all with allele frequencies of ≤5% (Table 3). One gene, CLTCL1, especially stood out compared to the other four genes, having 17 mutations in cases versus 6 mutations in controls (Table 3).
