Discovering Genomic Regions Enriched for De Novo Recombination in Autism Spectrum Disorder

Recombination is a vitally important aspect of meiosis, ensuring orderly segregation in the first meiotic division and generating genetic diversity by creating new haplotypes. Despite the essential role of recombination in the production of gametes and the potential for catastrophic chromosomal abnormalities resulting from improper crossing over, recombination rates and locations vary widely among individuals. Given the relative rarity of recombination, generally occurring only one to four times per chromosome, there exist large regions of the genome where variants and rarely separated and may become co-adapted over many generations. Should a recombination occur in one of these regions that separates co-adapted variants, it may generate a deleterious haplotype that contributes to disease. This thesis introduces methods for detecting regions enriched for de novo recombination in order to examine their effect on autism spectrum disorder. A new software package for detecting recombination in nuclear families, called inferrecom, was developed for this study. The rates and locations of recombination as inferred by the package conform closely to the distribution predicted by population genetic maps. A suite of diagnostic and visualisation tools and a novel phasing algorithm were also developed for the package and feature throughout this work. Family structures consisting of both parents and three or more children allow for localisation of crossover events and identification of the recombinant child. Comparing the distribution of recombination in autistic children with their non-autistic siblings, several regions showed a suggestive excess of recombination in the autistic children. While none of the regions were genome-wide significant after rigorous permutation-based multiple testing control, several had possible mechanistic links to autism, including hotspots near RYR3 in the oxytocin signalling pathway and DRD1 in the dopaminergic system. Nuclear families consisting of both parents and two children allow for localisation of crossover events, but do not allow for the identification of the exact recombinant child. To analyse regions of potential enrichment for recombination, it was necessary to devise a method that compared observed recombination rates to an external control derived from sex-specific genetic maps. After applying a novel bootstrap-based multiple testing control procedure, no regions remained genome-wide significant, though several suggestively enriched regions had plausible means of contribution to autism. One putatively enriched region identified in the 20q13.33 cytogenetic band in the families of three or more children also showed suggestive enrichment in the two-child families. The analyses in this study have identified several loci where rearrangement via recombination has a plausible link with autism. While no regions showed genome-wide significance, the presence of local excesses of recombination in known autism susceptibility regions lends credence to the hypothesis that de novo recombination may contribute to the genetic aetiology of autism. These regions are candidates for further investigation to determine the exact contribution and regulatory mechanism of each locus.