The impact of assortative mating, participation bias and socioeconomic status on the polygenic risk of behavioural and psychiatric traits

Abstract

To investigate assortative mating (AM), participation bias and socioeconomic status (SES) with respect to the genetics of behavioural and psychiatric traits, we estimated AM signatures using gametic phase disequilibrium and within-spouses and within-siblings polygenic risk score correlation analyses, also performing a SES conditional analysis. The cross-method meta-analysis identified AM genetic signatures for multiple alcohol-related phenotypes, bipolar disorder, major depressive disorder, schizophrenia and Tourette syndrome. Here, after SES conditioning, we observed changes in the AM genetic signatures for maximum habitual alcohol intake, frequency of drinking alcohol and Tourette syndrome. We also observed significant gametic phase disequilibrium differences between UK Biobank mental health questionnaire responders versus non-responders for major depressive disorder and alcohol use disorder. These results highlight the impact of AM, participation bias and SES on the polygenic risk of behavioural and psychiatric traits, particularly in alcohol-related traits.

Figure 1.

Genetic signatures of assortative mating (AM) across behavioral traits and psychiatric disorders after Bonferroni multiple testing correction. Genome-wide association statistics generated by the Psychiatric Genomics Consortium (PGC) and the Million Veteran Program (MVP) were used as discovery datasets to calculate polygenic risk scores in the UK Biobank (target cohort). AM estimates were obtained from gametic phase disequilibrium (GPD; 271,957 unrelated individuals of European descent), within-siblings (WSib; 17,910 sibling pairs of European descent), and within-spouses (WSps; 38,457 spouse pairs of European descent), and analyses (Supplementary Tables 3, 4, and 5, respectively). The AM estimates were combined through meta-analysis (Supplementary Table 2). 95% confidence intervals (CI) are reported for each estimate obtained from the corresponding regression model. All estimates are reported as percentages to facilitate their interpretation. Method-specific estimates were scaled to reflect the same underlying parameter: 2×(WSib-PRS rho-0.5Δ), 2×GPD, and WSps-PRS rho.

Figure 2.

Genetic signatures of assortative mating (AM) across UK Biobank (UKB) mental health questionary (MHQ) traits after Bonferroni multiple testing correction. Polygenic risk scoring was conducted using UKB MHQ responders as discovery cohort and UKB non-responders as target cohort. AM estimates were obtained from gametic phase disequilibrium (GPD; 199,414 unrelated individuals of European descent), within-siblings (WSib; 8,542 sibling pairs of European descent), and within-spouses (WSps; 20,971 spouse pairs of European descent) analyses (Supplementary Tables 11, 12, and 13, respectively). The AM estimates were combined through meta-analysis (Supplementary Table 10). 95% confidence intervals (CI) are reported for each estimate obtained from the corresponding regression model. All estimates are reported as percentages to facilitate their interpretation. Method-specific estimates were scaled to reflect the same underlying parameter: 2×(WSib-PRS rho-0.5Δ), 2×GPD, and WSps-PRS rho.

Figure 3.

Genetic signatures of assortative mating (AM) across behavioral traits and psychiatric disorders with significant changes after SES conditioning. 95% confidence intervals (CI) are reported for each estimate. The sample size of the Within-siblings (WSib), Within-spouses (WSps), and Gametic Phase Disequilibrium (GPD) analyses are indicated in Figure 1 legend for Maximum habitual alcohol intake and Tourette syndrome and in Figure 2 legend for Frequency of drinking alcohol. GPD, WSib, and WSps estimates were combined through meta-analysis (Supplementary Tables 2–5 and 10–13). The estimates are reported as percentages to facilitate their interpretation. 95% confidence intervals (CI) are reported for each estimate obtained from the corresponding regression model. Method-specific estimates were scaled to reflect the same underlying parameter: 2×(WSib-PRS rho-0.5Δ), 2×GPD, and WSps-PRS. Abbreviations: Household income (HI); Townsend Deprivation Index (TDI).