Multi-ancestry genome-wide analysis identifies shared genetic effects and common genetic variants for self-reported sleep duration

Abstract

Both short (≤6 h per night) and long sleep duration (≥9 h per night) are associated with increased risk of chronic diseases. Despite evidence linking habitual sleep duration and risk of disease, the genetic determinants of sleep duration in the general population are poorly understood, especially outside of European (EUR) populations. Here, we report that a polygenic score of 78 European ancestry sleep duration single-nucleotide polymorphisms (SNPs) is associated with sleep duration in an African (n = 7288; P = 0.003), an East Asian (n = 13 618; P = 6 × 10-4) and a South Asian (n = 7485; P = 0.025) genetic ancestry cohort, but not in a Hispanic/Latino cohort (n = 8726; P = 0.71). Furthermore, in a pan-ancestry (N = 483 235) meta-analysis of genome-wide association studies (GWAS) for habitual sleep duration, 73 loci are associated with genome-wide statistical significance. Follow-up of five loci (near HACD2, COG5, PRR12, SH3RF1 and KCNQ5) identified expression-quantitative trait loci for PRR12 and COG5 in brain tissues and pleiotropic associations with cardiovascular and neuropsychiatric traits. Overall, our results suggest that the genetic basis of sleep duration is at least partially shared across diverse ancestry groups.

Figure 1

Transferability of genome-wide EUR sleep duration polygenic score to AFR and SAS cohorts. PGS quartile plots for (A) AFR and (B) SAS cohorts; the results are displayed as the change in sleep duration (minutes) by genome-wide PGS quartile.

Figure 2

Manhattan and Q-Q plots for the sleep duration multi-ancestry meta-analysis (n = 483,235). The Manhattan plot highlights 3 novel loci (rs11928693, rs7689303, and rs6909225), 2 loci (rs75639044 and rs12462756) reported by Jansen et al. (28) (n = 384,317; (yellow) and 68 loci from Dashti et al. (10) (n = 483,235; hg19 genome build). Genomic inflation for the meta-analysis was λ = 1.351. This is comparable to inflation values observed for other polygenic complex traits and is also due to the large sample size of the EUR GWAS (λ = 1.200) and the meta-analysis structure of the study.

Figure 3

Brain tissue–specific eQTLs at PRR12 and COG5 sleep duration loci from multi-ancestry analysis. The eQTLs were examined using GTex V8 resource, and the results are displayed as normalized gene expression for each locus; the number below the box plot is the sample size. The plot shows significant eQTLs from Table S5 for tissues previously reported to be involved in sleep duration. (A) The sleep duration increasing A allele at rs12462756 significantly increases PRR12 gene expression in the cerebellum, cortex and putamen. (B) The rs12462756 A allele significantly increases IRF3 gene expression in the cerebellum. (C) The sleep duration increasing C allele at rs75639044 significantly decreases COG5 gene expression in the spinal cord.