Detection of clinically relevant copy-number variants from short-read sequencing data for genomic diagnostics

Ramakrishnan Rajagopalan

doi:10.17918/00000332

The overall goal of clinical genomic diagnostics is to identify pathogenic genetic variants that cause disease. Genetic variants can be classified into different subtypes based on the nature of the variant (sequence, copy-number, and structural variants), and current standard of care protocols require different diagnostic assays to detect these subtypes. Patients with suspected genetic disease often receive both chromosomal microarray array (CMA) testing for genome-wide copy number variant detection and exome sequencing for sequence variant detection to cover the spectrum of variant subtypes and sizes during their diagnostic odyssey. While short-read NGS testing, such as exome and genome, can detect both copy number and sequence variants, its implementation into the clinical lab has been hampered by both the lack of clinical standards and technical challenges concerning performance. Although not a currently offered clinical test, the identification of both copy number and sequencing variants from the same data would reduce the time to diagnosis, the costs involved, and significantly impact patient care. Using a cohort of 307 samples with clinical CMA and exome sequencing data, this thesis presents high quality, clinical-grade technical validation study and a new approach to tackling the major issue of false positives with current algorithms for CNV detection from exomes. A novel reproducibility framework was developed to assess the effect of control cohorts, and an R package for scalability in high-performance computing environments to analyze large exome sequencing cohorts. Application of the tools developed in this thesis to a cohort of 546 patients with rare pediatric disorders revealed eight novel diagnoses (1.5%) over their standard of care testing, and application to routine clinical epilepsy and hearing loss next-generation sequencing panels provided a minimum additional diagnostic yield of 2%. Over the next few years, genome sequencing is poised to become a first-tier diagnostic test, with the capability to detect all major classes of variation, including structural variants. Building on the exome work, I have developed a clinical-grade analytical workflow to integrate copy number and structural variants using 48 index samples with genome sequencing data and improved the false-positive rate. Application of this workflow to a cohort of 14 patients with clinically diagnosed Alagille Syndrome, but without a molecular diagnosis, revealed four novel diagnoses that were not detectable by the prior standard of care tests. These findings included a submicroscopic inversion in the gene JAG1 and a deletion in NOTCH2, the first-ever pathogenic copy number variant identified in this gene. Further, applying these methods to a cohort of 15 patients with nonsyndromic hearing loss revealed three novel diagnoses. The case series presented in this work argues the advantage of genome sequencing over the current standard of care tests for variants undetectable by their standard of care testing. Together, the frameworks presented in this thesis improved the current standard of care exome test and laid the groundwork for a future clinical test, genome sequencing. In addition, I demonstrated the clinical utility of the tools developed in this work showing how these approaches resulted in novel diagnoses and developed recommendations for broader application in clinical diagnostic settings. Keywords: copy-number variants, next-generation sequencing, Short-read sequencing, structural variation

Detection of clinically relevant copy-number variants from short-read sequencing data for genomic diagnostics

Files and links (1)

Abstract

Metrics

Details

Detection of clinically relevant copy-number variants from short-read sequencing data for genomic diagnostics

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media