Using meta-analysis, we examined association from 14 independent cohorts contributed by eight academic studies (see Table 1). Each contributing site confirmed all affected individuals had an ASD diagnosis; details of diagnostic processes are provided in the Additional file 1 and where available, study specific details are described elsewhere [7, 12, 16, 17, 23–25]. Where data permitted, we excluded individuals assessed at under 36 months of age or if there was any evidence of diagnostic criteria not being met from either the Autism Diagnostic Interview-Revised (ADI-R) [26] or the Autism Diagnostic Observation Schedule (ADOS) [27]. The primary meta-analysis (Worldwide ancestry (WW)) was based on data from 7387 ASD cases and 8567 controls. An additional meta-analysis on a more ancestrally homogenous subset (see Additional file 1) was also performed; this subset included data from 6197 ASD cases and 7377 controls of ‘European ancestry’.Table 1Study design and sample size of the contributing ASD collections. For some collections, more than one genotyping panel was used or the study design differed, i.e., trios or case-control; in such cases, the sample was split into ‘sets’ based
