Identification of Genetic Copy Number Variants in Neurodevelopmental Disorders from Genome Sequence Data

Abstract

Neurodevelopmental disorders (NDDs) are a group of heterogeneous conditions that effect brain function with onset in the developmental period. While individually rare, collectively NDDs are present in 7.0-13.6% of the population. Genetics (including copy number variants, CNVs) plays a central role in many of these disorders, but is yet to be fully understood. Many bioinformatic CNV detection tools have been developed to identify CNVs from whole exome or whole genome sequence reads. However, there are no established best practices for CNV detection, and the use in molecular diagnostics is not routine. This thesis aimed to establish a bioinformatic framework for the identification of causative CNVs from aligned genetic sequence reads. The performance of whole exome sequencing based CNV detection tools were found to be ineffective without an a priori assumption of the genes involved. Therefore, a quantitative comparison of whole genome sequencing based CNV detection tools was conducted, comparing the performance of one tool from each of the primary approaches: read depth, read pair, split read, assembly-based, and a combinatorial approach. The most balanced performance was observed from Break Dancer, which was subsequently integrated into a bioinformatic pipeline to identify CNVs. Copy number variant calls were then filtered and prioritised, incorporating CNV size and quality, population frequency, and predicted effect on gene expression and function. Prioritisation also incorporated a custom software package (RBV, Read Balance Validator) that calculated the probability of the presence of a CNV based upon the allele balance. Thirty one New Zealand families with NDDs were analysed using this custom pipeline, identifying causative mutations from two families (encompassing genes TANGO2 and SLC19A3), resulting in molecular diagnoses and subsequent life-saving clinical management. The strength of the pipeline lies in its ability to be applied to disorders beyond NDDs, and provides a validated CNV detection framework which can be transferred to a clinical setting. This thesis demonstrates the utility of WGS data for the detection of causative CNVs in rare NDDs. An early and accurate genetic diagnosis is critical for disorders which affect development, and can significantly improve health outcomes for patients and families.