Yield of genetic association signals from genomes, exomes and imputation in the UK Biobank

Gaynor SM; Joseph T; Bai X; Zou Y; outkov B; Maxwell EK; Delaneau O; Hofmeister RJ; Krasheninina O; Balasubramanian S; Marcketta A; Backman J; C Regeneron Genetics; Reid JG; Overton JD; Lotta LA; Marchini J; Salerno WJ; Baras A; Abecasis GR; Thornton TA

doi:10.1530/ey.22.15.4

15.4

Prev Next

Section Contents Cite

ESPE Yearbook of Paediatric Endocrinology (2025) 22 15.4 | DOI: 10.1530/ey.22.15.4

ESPEYB25 15. Editors' Choice Genetics (9 abstracts)

15.4. Yield of genetic association signals from genomes, exomes and imputation in the UK Biobank

Gaynor SM , Joseph T , Bai X , Zou Y , Boutkov B , Maxwell EK , Delaneau O , Hofmeister RJ , Krasheninina O , Balasubramanian S , Marcketta A , Backman J , Regeneron Genetics C , Reid JG , Overton JD , Lotta LA , Marchini J , Salerno WJ , Baras A , Abecasis GR & Thornton TA

Nat Genet 2024; 56:2345-2351. PMID: 39322778 doi: 10.1038/s41588-024-01930-4

In Brief: This population-based study in up to 468,169 UK Biobank participants compared the genetic contributions to 100 complex traits arising from different types of data: whole-genome sequencing (WGS), whole-exome sequencing (WES) and array genotyping with imputation (IMP). Compared to WES+IMP, WGS generated data on ~5-times more genetic variants but resulted in a very modest increase in the number of robust genetic associations (only 1% more).

Comment: Substantial public, charitable and commercial funding continues to sustain a remarkable pace of growth in genetic research data, with much of it being available in the public domain. WGS is by definition the most complete source of germline genetic information. WES (which sequences only protein coding domains) and IMP (genotyping of a subset of tagging variants then statistical estimation of the others) were designed as affordable alternatives while the cost of WGS was prohibitive.

In an era when all 3 sources are now available (WGS, WES and IMP), it may seem surprising then that WES+IMP are the preferred options to WGS. There are several reasons for this. The authors focussed on complex traits, which are likely to be polygenic in origin. By contrast, WGS is likely most informative for studies of rare monogenic disorders. Second, current affordable WGS uses short-read sequencing and has lower coverage than clinical sequencing. This reduces its accuracy and it does not detect large structural variations. Third, WGS provides the most complete ‘search space’ (compared to WES and IMP) but in doing so it requires the most stringent statistical thresholds to control against multiple testing and false positive findings. WGS will (by definition) provide the most complete genetic picture but (at least for complex traits) will require far larger samples sizes than we have now.