Genetic etiologies of Autism Spectrum Disorder

Abstract

Autism spectrum disorder (ASD) is a common, heritable neurodevelopmental disorder. In this thesis, I examine how different genetic etiologies, mutation types and specific genes contribute to the risk of ASD, and how these factors can be used to expand our understanding of the neurobiological underpinnings of ASD. I develop a new bioinformatics method (CoNIFER: Copy Number Inference from Exome Reads) to identify Copy Number Variants (CNVs) using exome sequencing data, enabling much more sensitive identification of a previously under-ascertained class of small CNVs (<100kb in size). I estimate the precision of the algorithm using 366 exomes and show that this method can be used to reliably predict both <italic>de novo</italic> and inherited rare CNVs and can predict absolute copy number for loci with fewer than eight copies. Next, I searched for disruptive, genic rare CNVs among 411 families with sporadic ASD from the Simons Simplex Collection and identified additional small genic rare CNVs compared to high-density SNP microarrays (~2x higher yield). I found that affected probands inherit more CNVs than their siblings (p=0.004; OR=1.19), and these CNVs affect more genes, are enriched for brain-expressed genes, and are transmitted preferentially from the mother. I found that the excess burden of inherited CNVs among probands is driven primarily by sib-pairs with discordant social behavior phenotypes. Next, I created a combined set of both inherited and <italic>de novo</italic> Single Nucleotide Variants (SNVs) and CNVs across 2,377 Simons Simplex Collection (SSC) ASD families, including 1,786 families with both an affected and unaffected child. I compared the burden of inherited and <italic>de novo</italic> mutations between affected and unaffected siblings and found that private inherited truncating SNV mutations in conserved genes are significantly enriched in probands (OR=1.14, p < 0.0002)-- an effect that became more pronounced with increasing gene conservation. I quantified ASD risk for <italic>de novo</italic> and inherited CNVs and SNVs by using a conditional logistic regression model. Independent from <italic>de novo</italic> mutations, private truncating SNVs and rare inherited CNVs contribute an increase in risk of 1.11 (p=0.0002) and 1.23 (p = 0.01), respectively. These results confirm a statistically independent role for inherited mutations in ASD risk and identify additional candidate genes (eg. <italic>RIMS1</italic>, <italic>CUL7</italic> and <italic>CSMD1</italic>) where inherited and <italic>de novo</italic> burden converge.