We present here the results from a large international collaborative GWAS meta-analysis and follow up of 16539 ASD cases and 157234 controls. Despite the considerable increase in the sample size and statistical power of these new analyses to identify associations, we do not observe individual variants that exceed the accepted GWS threshold (P ≤ 5 × 10−8) in the discovery GWAS (n = 7387 ASD cases and 8567 controls). This does not, however, disqualify the loci which fall within the upper-ranked associations from further interrogation, through replication or through supporting biology. There is evidence to support the hypothesis that a substantial proportion of ‘borderline’ association represent genuine associations and deserve an attempt at replication [41]. A PGC schizophrenia GWAS [40], which was of comparable size to this, yielded 5 GWS loci and 183 non-GWS loci at P < 10−4. In a follow-up study [19], 20 of these markers were elevated beyond GWS; these were not limited to the most highly ranked markers, as the newly GWS markers ranged from 7 to 188 in the original study.
