Search

Fully understanding the genetic factors involved in Autism Spectrum Disorder (ASD) requires whole-genome sequencing (WGS), which theoretically allows the detection of all types of genetic variants. With the aim of generating an unprecedented resource for resolving the genomic architecture underlying ASD, we analyzed genome sequences and phenotypic data from 5,100 individuals with ASD and 6,212 additional parents and siblings (total n=11,312) in the Autism Speaks MSSNG Project, as well as additional individuals from other WGS cohorts. WGS data and autism phenotyping were based on high-quality short-read sequencing (>30x coverage) and clinically accepted diagnostic measures for ASD, respectively. For initial discovery of ASD-associated genes, we used exonic sequence-level variants from MSSNG as well as whole-exome sequencing-based ASD data from SPARK and the Autism Sequencing Consortium (>18,000 trios plus additional cases and controls), identifying 135 ASD-associated protein-coding genes with false discovery rate SCN2A and a nuclear mitochondrial insertion impacting SYNGAP1. Polygenic risk scores did not differ between children with ASD in multiplex families versus simplex, and rare, damaging recessive events were significantly depleted in multiplex families, collectively suggesting that rare, dominant variation plays a predominant role in multiplex ASD. Our study provides a guidebook for exploring genotype-phenotype correlations in the 15-20% of ASD families who carry ASD-associated rare variants, as well as an entry point to the larger and more diverse studies that will be required to dissect the etiology in the >80% of the ASD population that remains idiopathic. All data resulting from this study are available to the medical genomics research community in an open but protected manner.

Autism Spectrum Disorder (ASD) is genetically complex, but specific copy number variants (CNVs; e.g., 1q21.1, 16p11.2) and genes (e.g., NRXN1, NLGN4) have been identified as penetrant susceptibility factors, and all of these demonstrate pleiotropy. Many ASD-associated CNVs are, in fact, genomic disorder loci where flanking segmental duplications lead to recurrent deletion and duplication events of the same region in unrelated individuals, but these lesions are large and involve multiple genes. To identify opportunities to establish a more specific genotype and phenotype correlation in ASD, we searched genomic data, and the literature, for recurrent predicted damaging sequence-level variants affecting single genes. We identified 17 individuals from 15 unrelated families carrying a heterozygous guanine duplication (rs797044936; NM_033517.1; c.3679dup; p.Ala1227Glyfs*69) occurring within a string of 8 guanines (at genomic location [hg38]g.50,721,512dup) affecting SHANK3, a prototypical ASD gene (6/7,521 or 0.08% of ASD-affected individuals studied by whole genome sequencing carried the p.Ala1227Glyfs*69 variant). This variant, which is predicted to cause a frameshift leading to a premature stop codon truncating the C-terminal region of the corresponding protein, was not reproducibly found in any of the control groups we analyzed. All probands identified carried de novo mutations with the exception of five individuals in three families who inherited it through somatic mosaicism. This same heterozygous variant in published mouse models leads to an ASD-like phenotype. We scrutinized the phenotype of p.Ala1227Glyfs*69 carriers, and while everyone (16/16) formally tested for ASD carried a diagnosis, there was variable expression of core ASD features both within families and between families, underscoring the impact of as yet unknown modifiable factors affecting expressivity in autism.