Directional dominance on stature and cognition in diverse human populations

Homozygosity has long been associated with rare, often devastating, Mendelian disorders, and Darwin was one of the first to recognize that inbreeding reduces evolutionary fitness. However, the effect of the more distant parental relatedness that is common in modern humanpopulations is lesswell understood.Genomic data now allow us to investigate the effects of homozygosity on traits of public health importance by observing contiguous homozygous segments (runs of homozygosity), which are inferred to be homozygous along their complete length. Given the low levels of genome-wide homozygosity prevalent in most human populations, information is required on very large numbers of people to provide sufficient power. Here we use runs of homozygosity to study 16 healthrelated quantitative traits in 354,224 individuals from 102 cohorts, and find statistically significant associations between summed runs of homozygosity and four complex traits: height, forced expiratory lung volume in one second, general cognitive ability and educational attainment (P, 13 10, 2.13 10, 2.53 10 and 1.83 10, respectively). In each case, increased homozygosity was associated with decreased trait value, equivalent to the offspring of first cousins being 1.2 cm shorter and having 10 months’ less education. Similar effect sizes were found across four continental groups and populations with different degrees of genome-wide homozygosity, providing evidence that homozygosity, rather than confounding, directly contributes to phenotypic variance. Contrary to earlier reports in substantially smaller samples, no evidencewas seen of an influence of genome-wide homozygosity on blood pressure and low density lipoprotein cholesterol, or ten other cardio-metabolic traits. Since directional dominance is predicted for traits under directional evolutionary selection, this study provides evidence that increased stature and cognitive function have been positively selected in human evolution, whereas many important risk factors for late-onset complex diseases may not have been. Inbreeding influences complex traits through increases in homozygosity and corresponding reductions in heterozygosity, most likely resulting from the action of deleterious (partially) recessive mutations. For polygenic traits, a systematic association with genome-wide homozygosity is not expected when dominant alleles at some loci increase the trait value while others decrease it. Rather, dominance must be biased in one direction on average over all causal loci, for instance to decrease the trait. Such directional dominance is expected to arise in evolutionary fitness-related traits due to directional selection. Studies of genome-wide homozygosity thus have the potential to reveal the non-additive allelic architecture of a trait and its evolutionary history. Historically, inbreeding has been measured using pedigrees. However, such techniques cannot account for the stochastic nature of inheritance, nor are they practical for the capture of the distant parental relatedness present in most modern-day populations. High-density genome-wide single nucleotide polymorphism (SNP) array data can now be used to assess genome-wide homozygosity directly, using genomic runs of homozygosity (ROH). Such runs are inferred to be homozygous-by-descent and are common in human populations. Summed ROH (SROH) is the sum of the length of these ROH, in megabases of DNA. FROH is the ratio of SROH to the total length of the genome. Like pedigree-based F (with which it is highly correlated), FROH estimates the probability of being homozygous at any site in the genome. FROH has been shown to vary widely within and between populations and is a powerful method of detecting genome-wide homozygosity effects. We found marked differences by geography and demographic history in both the population mean SROH and the relationship between SROH and NROH (the numbers of separate runs of homozygosity) (Fig. 1). As observed previously, isolated populations have a higher burden of ROH, whereas African heritage populations have the least homozygosity. We studied bFROH , defined as the effect of FROH on 16 complex traits of biomedical importance (Fig. 2). For height, FEV1 (forced expiratory volume in one second, ameasure of lung function), educational attainment, and g (a measure of general cognitive ability derived from scores on several diverse cognitive tests), we found the effect sizes were greater than two intra-sex standard deviations, with P values all less than 10. Thus the associations could not plausibly be explained by chance alone (Table 1; see Extended Data Figs 1–4 for Forest plots of individual traits; Supplementary Table 1 for s.d.). To ensure that the results were not driven by a few outliers, we repeated the analysis excluding extreme sub-cohort trait results. In all cases the effect sizes and their significance remained similar or increased (see Supplementary Table 2 for comparisons with and without outliers). After exclusion of outliers, these effect sizes translate into a reduction of 1.2 cm in height and 137 ml in FEV1 for the offspring of first cousins, and into a decrease of 0.3 s.d. in g and 10 months’ less educational attainment. We performed a number of analyses to exclude confounding.While SROH iswholly a genetic effect, its inheritance is entirely non-additive. Therefore, unlike in genome-wide association, an association with population genetic structure or co-segregation of additive genomewide polygenic effects and SROH (as opposed to SNPs in a genomewide association study) are not expected as a matter of course, except in the case of siblings. However, confounding could still theoretically arise, as discussed below. We therefore assessed this by conducting stratified and covariate analyses. We found effects of similar magnitude and in the same direction for all four traits across isolated and non-isolated European, Finnish, African, Hispanic, East Asian and South and Central Asian populations (Extended Data Fig. 5a and Supplementary Table 3).We further testedwhether the effect sizes were similar when cohorts were split into more and less homozygous groups. The effect sizes were very similar, even though the degree of homozygosity (and variation in homozygosity) varied 3–10-fold between the two strata (depending on which cohorts contributed to the trait; Extended Data Fig. 5b). This suggests a broadly linear relationshipwith SROH. In general, confidence intervals overlap for stratified estimates, suggesting that differences arose due to sampling variance. Larger confidence intervals for some estimates reflect the lower power of some strata, in turn reflecting the sample size and degree of homozygosity of those strata (for example, the wider confidence intervals for estimates of educational attainment bFROH for Finnish and African strata). Finally, we fitted educational attainment