CNest: A Novel Copy Number Association Discovery Method Uncovers 862 New Associations from 200,629 Whole Exome Sequence Datasets in the UK Biobank

Abstract

Copy number variation (CNV) has long been known to influence human traits having a rich history of research into common and rare genetic disease and although CNV is accepted as an important class of genomic variation, progress on copy number (CN) phenotype associations from Next Generation Sequencing data (NGS) has been limited, in part, due to the relative difficulty in CNV detection and an enrichment for large numbers of false positives. To date most successful CN genome wide association studies (CN-GWAS) have focused on using predictive measures of dosage intolerance or gene burden tests to gain sufficient power for detecting CN effects. Here we present a novel method for large scale CN analysis from NGS data generating robust CN estimates and allowing CN-GWAS to be performed genome wide in discovery mode. We provide a detailed analysis in the large scale UK BioBank resource and a specifically designed software package for deriving CN estimates from NGS data that are robust enough to be used for CN-GWAS. We use these methods to perform genome wide CN-GWAS analysis across 78 human traits discovering 862 genetic associations that are likely to contribute strongly to trait distributions based solely on their CN or by acting in concert with other genetic variation. Finally, we undertake an analysis comparing CNV and SNP association signals across the same traits and samples, defining specific CNV association classes based on whether they could be detected using standard SNP-GWAS in the UK Biobank.

The individual and global impact of copy-number variants on complex human traits

“…Finally, we focused on the most frequent CNV in our cohort (frequency = 3.76%; Figure 1 A), the 50 kb 1p36.11 deletion (chr1: 25,599,041–25,648,747), which encompasses RHD (Rhesus [Rh] blood group D antigen [MIM: 111680 ]) and RSRP1 and associated with increased reticulocyte count (β del = 2.7 × 10 9 cells/L; p = 7.8 × 10 −14 ), decreased platelet count (β del = −3.7 × 10 9 cells/L; p = 1.4 × 10 −12 ), and decreased HbA1c (β del = −0.3 mmol/mol; p = 9.3 × 10 −8 ) ( Figure 4 C). Overlap with SNP-GWAS signals for various hematological traits 78 , 79 combined with subsequent replication of the reticulocyte count association based on whole-exome sequencing CNV calls 80 prompted the investigation of the expression of these genes in whole blood. Tissue-specific transcriptomic data from the GTEx project v8 81 ( web resources ) revealed that RHD , a protein whose presence/absence on erythrocyte cell membranes is critical in determining an individual’s Rh blood group, 82 was almost exclusively expressed in whole blood ( Figure 4 D), whereas RSRP1 was ubiquitously expressed, with lower expression in whole blood ( Figure 4 E).…”

“…Future release of large sequencing datasets combined to progress in CNV detection tools could resolve these issues and lead to novel discoveries. 21 , 80 , 132 , 133 Second, despite substantial evidence of CNV- and SNP-GWAS signal colocalization, we did not perform robust enrichment analyses, as the non-random genomic distribution and complex nature of CNVs renders simulating the null scenario beyond the scope of this paper. Signal colocalization is likely to be underestimated, as manual literature searches revealed overlaps missed by our annotation pipeline (e.g., 16p13.11 age at menarche signal 101 ) and we obtained a 7% increase in signal colocalization by using GWAS Catalog annotation 6 months apart (31% April 2021 → 38% October 2021).…”