Females with higher-risk genotypes may encounter difficulties at later stages of their lives that manifest as a different diagnostic category, or that reduces fecundity. If true, the disorder would most likely be one with a gender bias opposite that of ASDs, such as anorexia nervosa (Fairburn and Harrison, 2003). Our genetic theory of autism, as discussed above, largely depends on dominant acting genetic variants of variable

penetrance. We think the theory is sufficient to explain most of the genetic basis of autism, both simplex and multiplex, but certainly not all. For example, the role of recessive mutations in individuals from consanguineous marriages has been demonstrated (Morrow et al., 2008). We have observed only a single case of inheritance of a rare homozygous null state. A striking finding of all the studies of de novo mutation

in children with ASDs is the apparent number of distinct target loci. Even discounting 25% of events PI3K Inhibitor Library as incidental (based on a 2% frequency in sibs and 8% in probands), there are a large number of target regions and few recurrences. Only CNVs at 16p11.2 are present in more than 1% of cases (ten out of 858 children). We can make an estimate of the minimum number of target regions by analysis of recurrence. Combining two large studies (ours and that of Pinto et al., 2010), we observe 39 overlaps at 12 recurrent loci in 121 events. Excluding the highly recurrent 16p11.2 locus (with 13 hits in the combined dataset) and Selleck LY2835219 discounting

one-quarter of the remaining 108 events as incidental, we observe 11 recurrent loci in approximately 80 causal events. If we assume a uniform rate of copy-number almost mutation, we estimate the number of target loci at 250–300. However, targets do not have a uniform rate of copy-number mutation, so this figure would be an underestimate of total targets. We derive a similar estimate for target size by a completely different method, based on many assumptions including the rate of new mutations that damage a gene in humans (about one gene per three births), the incidence of ASDs among males (approximately 1 in 100), a genetic model that predicts that about half of ASDs result from new mutations (Zhao et al., 2007), and high penetrance of a select set of single mutational hits. The latter assumption is based on the observation of dominant transmission in multiplex families (Zhao et al., 2007). An organism will be vulnerable to a single mutational hit at only a small subset of its genetic elements. We imagine that vulnerable targets may arise by two distinct cellular mechanisms: insufficient or uncorrectable dosage compensation resulting from (for example) altered stoichiometry of protein complexes; and monoallelic gene expression, which could result in subpopulations of functionally null neurons, perhaps confined to specific subtypes (Gimelbrant et al., 2007 and Gregg et al., 2010).