Imprint of Assortative Mating on the Human Genome

Abstract

Non-random mate-choice with respect to complex traits is widely observed in humans, but whether this reflects true phenotypic assortment, environment (social homogamy) or convergence after choosing a partner is not known. Understanding the causes of mate choice is important, because assortative mating (AM) if based upon heritable traits, has genetic and evolutionary consequences. AM is predicted under Fisher's classical theory1 to induce a signature in the genome at trait-associated loci that can be detected and quantified. Here, we develop and apply a method to quantify AM on a specific trait by estimating the correlation (θ) between genetic predictors of the trait from SNPs on odd versus even chromosomes. We show by theory and simulation that the effect of AM can be distinguished from population stratification. We applied this approach to 32 complex traits and diseases using SNP data from ~400,000 unrelated individuals of European ancestry. We found significant evidence of AM for height (θ=3.2%) and educational attainment (θ=2.7%), both consistent with theoretical predictions. Overall, our results imply that AM involves multiple traits, affects the genomic architecture of loci that are associated with these traits and that the consequence of mate choice can be detected from a random sample of genomes.. CC-BY-NC-ND 4.0 International license It is made available under a (which was not peer-reviewed) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.The copyright holder for this preprint . http://dx.doi.org/10.1101/300020 doi: bioRxiv preprint first posted online Apr. 13, 2018; 3 Non-random mating in natural populations has short and long-term evolutionary consequences. In many species, including humans, mate choice is often associated with phenotypic similarities between mates 2,3 . Such phenotypic similarities have multiple sources, for example social homogamy, the preference for a mate from the same environment, or because of primary assortment on certain traits observable at the time of mate choice. Contrary to social homogamy, primary phenotypic assortment, here referred to as assortative mating (AM), has genetic and evolutionary consequences and therefore is the focus of our study. In humans, AM involves multiple complex traits 4-8 and can sometimes lead to similar susceptibility to diseases [9][10][11][12] . The genetic effects of AM were first studied in the seminal articles of Fisher (1918) 1 and Wright (1921) 13 . Those two founding contributions, further complemented by Crow & Kimura (1970) 14 and others [15][16][17] have set the basis of the theory of AM on complex traits. AM theory predicts three main genetic consequences of a positive correlation between the phenotypes of mates in a population: (i) an increase of the genetic variance in the population, (ii) an increase in the correlation between relatives and (iii) an increase of homozygosity (deviation from HardyWeinberg Equilibrium; HWE), in particular at causal loci. These seemingly distinct c...

Familial Influences on Neuroticism and Education in the UK Biobank

“…Columns 2-4 of Table 1 (bold) show how many pairs of the three types of family members we identified with available data on neuroticism and years of education. The numbers of family relationships closely matched findings from previous publications (Yengo et al 2018;Bycroft et al 2018). Column 5 contains the sample sizes of family pairs (the total of couple, sibling and parent-offspring pairs).…”

“…Couples were identified as pairs of unrelated oppositesex individuals matching exactly on a string of household variables: social deprivation (Townsend Deprivation Index), assessment centre, income, time at address, smoker in household, type of accommodation, relatives in household, number in household. This approach of matching on household variables was used in a recent study of assortative mating in the UK Biobank (Yengo et al 2018). We note that there is potential for type 1 error: it is possible, especially in densely populated areas, that people could match on all eight variables by chance.…”

Genetic Associations Between Childhood Psychopathology and Adult Depression and Associated Traits in 42 998 Individuals

“…In each cohort, PGS were constructed for the following adult traits: MD, 35 BD, 37 SWB, 41 neuroticism, 41 insomnia, 42 EA, 38 and BMI. 39 Height 39 was included as a control phenotype (eTable 1 in the Supplement contains the GWAS discovery sample size for each trait). To avoid overlap between discovery and target samples, summary statistics omitting the target cohort or cohorts were used.…”

“…The last 2 years have seen ever-larger GWAS for traits, including major depression (MD), 35,36 BD, 37 EA, 38 and BMI, 39 consequently increasing accuracy of PGS. 40 Combined with the substantial increase in individuals genotyped in large longitudinal childhood cohorts that assess psychopathology, this provides an opportunity to rigorously investigate whether genetic factors underlie the associations between childhood psychopathology and adult mood disorders and associated nonpsychiatric traits (EA, insomnia, SWB, neuroticism, and BMI) and determine whether this association depends on age.…”

A critique of life history approaches to human trait covariation

“…Covariation between variants at different loci can be generated by non-random mating (Fisher 1918;Yengo et al 2018). For example if tall men tend to mate with intelligent women (or vice versa), the variants influencing the two traits will tend to covary in the population (which appears to be the case; Keller et al 2013).…”