Behavior Genetics

Mendelian randomization (MR) uses genetic variants as instrumental variables to enhance causal inference. Studies have identified genetic variants related to childhood maltreatment, but interpreting the effects of these variants or assessing the plausibility of MR assumptions is complex. We aim to investigate the feasibility of applying MR to complex social traits using the association between childhood maltreatment and mental health and behavioral outcomes as an example. We explore four potential key concerns: confounding by population phenomena, horizontal and vertical pleiotropy, reverse causality, and selection. For each concern, we demonstrate scenarios where MR studies of childhood maltreatment may be biased using DAGs and critical appraisal of previous MR analyses. For confounding by population phenomena, we further perform within-family genetic analyses in 42,101 parent-offspring trios from the Norwegian Mother, Father and Child Cohort Study (MoBa) to address bias due to family-level processes since childhood maltreatment often occurs within households. Our results showed same-trait shrinkage (11% attenuation of the association between childrens polygenic risk scores of childhood maltreatment (PRSCM) and mothers report of childrens physical abuse) but not cross-trait shrinkage (childrens PRSCM and childrens mental health and behavioral outcomes) after adjusting for parental PRSCM. The lack of cross-trait shrinkage suggests that genetic variants related to childhood maltreatment may be capturing other child-level phenotypes, after adjusting for family-level processes. Mothers PRSCM were also associated with mothers own maltreatment experiences in childhood and adulthood with similar magnitudes, suggesting these genetic effects are not specific to childhood maltreatment. Due to the complexity involved in the causal chain of childhood maltreatment and it being reported, the interpretation of MR studies for childhood maltreatment is challenging. Other causal approaches should be considered for observational studies of complex social traits.

Earlier research has shown observational associations of early pubertal timing and poor mental health. Mendelian randomization (MR) studies demonstrated a transient effect of pubertal timing on mental health during adolescence, but not later in life. MR studies also showed that there is a likely causal association of pubertal timing with life history traits. However, the strongest causal effects and genetic correlations with age of menarche have been found for Body Mass Index (BMI). As high BMI is associated with lower socioeconomic status and with poor mental health, the shared genetic etiology of socioeconomic status, BMI and poor mental health is not yet fully understood. BMI correlates negatively with socioeconomic status and several mental health outcomes. Despite their substantial genetic overlap, the underlying genetic etiology of these phenotypes remains unclear. In this study we applied Linkage Disequi-librium score regression to test genetic correlations of age of menarche with 33 socioeconomic, life history, social interaction, personality and psychiatric traits, and BMI. We further applied spectral decomposition and hierarchical clustering to the genetic correlation matrix. After controlling for multiple testing, we could only identify significant genetic correlations with BMI and three socioeconomic traits (household income, deprivation and parental longevity). The results suggest that genome-wide association studies on age of menarche also contain socioeconomic information. Future MR studies aiming to test the unconfounded effect of pubertal timing should make sure that genetic instruments have no pleiotropic effect on socioeconomic variables, or (if possible) also control for socioeconomic status on the observational level.

Genetic associations and correlations are perceived as confirmation that genotype influences one or more phenotypes respectively. However, genetic correlations can arise from non-biological or indirect mechanisms including population stratification, dynastic effects, and assortative mating. In this paper, we outline these mechanisms and demonstrate available tools and analytic methods that can be used to assess their presence in estimates of genetic correlations and genetic associations. Using educational attainment and parental socioeconomic position data as an exemplar, we demonstrate that both heritability and genetic correlation estimates may be inflated by these indirect mechanisms. The results highlight the limitations of between-individual estimates obtained from samples of unrelated individuals and the potential value of family-based studies. Use of the highlighted tools in combination with within-sibling or mother-father-offspring trio data may offer researchers greater opportunity to explore the underlying mechanisms behind genetic associations and correlations and identify the underlying causes of estimate inflation.

Randomized experiments are the "gold standard" for inferring causation and are designed to collect covariate information. Yet, many questions cannot be answered with experiments practically or ethically. Often, potential confounds are controlled statistically as covariates in quasi-experimental designs. The typical use of covariates does not control for many systematically confounded gene-and-environmental effects. Poverty, health, and individual differences all covary with gene-and-environmental effects, so much so that using covariates can create bias. We advocate for using genetically informed designs, which strengthen causal inference by controlling for major genetic and environmental confounds even in the absence of random assignment. We adapted the reciprocal standard dyad model into an analytic method to facilitate sibling comparisons. Differences between kin pairs explicitly distinguish within-family variance from between-family and control for all background variance linked to gene-and-environmental differences. We present four vignettes with individual differences and health outcomes. Although all four illustrations found significant associations when using covariate-based approaches, results diverged after addressing familial confounding with the discordant-kinship model. This divergence highlights the importance of considering familial influences in psychological research, demonstrating the versatility and efficacy of our method in different contexts.

Social contagion research suggests that health behaviors (BMI, smoking, drinking, etc.) spread through social networks, including dyadic ties such as between married/cohabiting partners. However, separating contagion from assortative mating ( like seeks like) and shared environmental factors remains notoriously difficult in observational studies. It is not possible to obtain exogenous variation in long-term partnerships ( random mating), but genetic approaches can offer a novel way to examine partner similarity and the role of social contagion. This paper explores possible social genetic effects among partners, i.e., effects of the partners genes on ones own behavior. We use the longitudinal Health and Retirement Study with data on health behavior and genomic data for both ego and his/her partner to examine social genetic effects for BMI, drinking, and smoking behavior. For each outcome, we find support for social genetic effects. Americans of European descent were more overweight if they had partners with higher polygenic scores for BMI net of their own polygenic score. Similar findings were found for the number of drinks per week and cigarettes per day. Longitudinal analyses that conditioned on past health behavior of both spouses confirmed these findings. We further explored whether susceptibility to the partners influence differed between men and women, but did not find consistent differences across outcomes. Findings are further discussed in the light of ramifications of social genetic effects for the social and biological sciences.

BackgroundResearch has begun to explore the effects of parental genetic nurturing on offspring outcomes using polygenic scores (PGSs). However, there are concerns regarding potential biases due to confounding when mediating parental phenotypes are included. MethodsDepression, educational attainment and height PGSs were generated for 2680 biological parent-offspring trios using genome-wide association study (GWAS) meta-analysis summary statistics in a large population study: Generation Scotland. Regression and pathway models were estimated incorporating PGSs for both parents and offspring to explore direct (offspring PGS) and genetic nurturing (parental PGS) effects on psychological distress, educational attainment and height. Genetic nurturing via parental phenotypes were incorporated into the models. To explore sources of bias we conducted simulation analyses of 10,000 trios using combinations of PGS predictive accuracy and accounted variance. ResultsModels incorporating both offspring and parental PGSs suggested positive parental genetic nurturing effects on offspring educational attainment, but not psychological distress or height. In contrast, models additionally incorporating parental phenotypic information suggested positive parent phenotype mediated genetic nurturing effects were at play for all phenotypes explored as well as negative residual genetic nurturing effects for height. 10,000 parent-offspring trio effects (without genetic nurturing effects) were simulated. Simulations demonstrated that models incorporating parent and offspring PGSs resulted in genetic nurturing effects that were unbiased. However, adding parental phenotypes as mediating variables results in biased positive estimates of parent phenotype mediated genetic nurturing effects and negative estimates of residual genetic nurturing effects. Biased effects increased in magnitude as PGS accuracy and accounted variance decreased. These biases were only eliminated when PGSs were simulated to capture the entirety of trait genetic variance. ConclusionResults suggest that in the absence of PGSs that capture all genetic variance, parental phenotypes act as colliders in the same way as heritable environments. Relatively simple models combining parental and offspring PGSs can be used to detect genetic nurturing effects in complex traits. However, our findings suggest alternative methods should be utilised when aiming to identify mediating phenotypes and potentially modifiable parental nurturing effects.

BackgroundCognitive, language, and social abilities are complex, heritable and intertwined traits shaping childrens development and later mental health. To better understand cross-trait interrelationships, we model here the structures of shared genomic and shared non-genomic/residual (i.e. broadly environmental) influences, and their correlation (rGE), investigating cognitive, language, and social behavioural/communication measures. MethodsData were obtained for unrelated children (8-13 years) from two population-based cohorts: the UK Avon Longitudinal Study of Parents and Children (ALSPAC, N[≤]6,543) and the US Adolescent Brain Cognitive DevelopmentSM (ABCD) Study (N[≤]4,412), and analyses were carried out implementing an extended data-driven genetic-relationship-matrix structural equation modelling (GRM-SEM) approach. ResultsIn ALSPAC, we identified two independent phenotypic domains, each captured by a structurally matching pair consisting of a genomic (A) and a non-genomic/residual (E) factor. The first domain reflected cognitive/language difficulties, with the largest genomic and residual factor loadings ({lambda}A and {lambda}E, respectively) for verbal IQ ({lambda}A=0.73(SE=0.05); {lambda}E=0.57(SE=0.07)). The second domain captured social difficulties, with the largest {lambda}A and {lambda}E for social communication measures ({lambda}A=0.39(SE=0.10); {lambda}E=0.82(SE=0.10)). We identified trait-specific rGE between pairs of A and E factors with different directions of effect (cognition/language rGE=0.89(SE=0.18), social rGE=-0.62(SE=0.17)). rGE patterns were linked to increased measurable A and E contributions for cognition/language difficulties, but decreased contributions for social problems. Analyses in ABCD confirmed the two domains for E and phenotypic structures, although genomic contributions were low. ConclusionsIn childhood, cognitive/language abilities versus social abilities are influenced by distinct genomic and/or environmental factors, potentially interlinked through trait-specific rGE, suggesting differences in developmental processes.

Suboptimal school performance is often seen in children with psychiatric disorders and is influenced by both genetics and the environment. Educational attainment polygenic score (EA-PGS) has been shown to significantly predict school performance in the general population. Here we analyze the association of EA-PGS with school performance in 18,495 children with and 12,487, without one or more of six psychiatric disorders and show that variance explained in the school performance by the EA-PGS is substantially lower in children with attention deficit hyperactivity disorder (ADHD) and autism spectrum disorder (ASD). Accounting for parents socioeconomic status obliterated the variance difference between ADHD-but not ASD-and controls. Given that a large proportion of the prediction performance of EA-PGS originate from family environment, our findings hint that family environmental influences on school performance might differ between ADHD and controls; studying the same further will open new avenues to improve the school performance of children with ADHD.

Twin and family studies have historically aimed to partition phenotypic variance into components corresponding to additive genetic effects (A), common environment (C), and unique environment (E). Here we present the ACE Model and several extensions in the Adolescent Brain Cognitive Development Study (ABCD Study(R)), employed using the new Fast Efficient Mixed Effects Analysis (FEMA) package. In the twin sub-sample (n = 924; 462 twin pairs), heritability estimates were similar to those reported by prior studies for height (twin heritability = 0.86) and cognition (twin heritability between 0.00 and 0.61), respectively. Incorporating SNP-derived genetic relatedness and using the full ABCD Study(R) sample (n = 9,742) led to narrower confidence intervals for all parameter estimates. By leveraging the sparse clustering method used by FEMA to handle genetic relatedness only for participants within families, we were able to take advantage of the diverse distribution of genetic relatedness within the ABCD Study(R) sample.

It is well-established that (1) there is a large genetic component to mental health, and (2) higher schooling attainment is associated with better mental health. Given these two observations, we test the hypothesis that schooling may attenuate the genetic predisposition to poor mental health. Specifically, we estimate associations between a polygenic score (PGS) for depressive symptoms, schooling attainment and gene-environment (GxE) interactions with mental health (depressive symptoms and depression), in two distinct United States datasets at different adult ages-29 years old in the National Longitudinal Study of Adolescent Health (Add Health) and 54 years old in the Wisconsin Longitudinal Study (WLS). OLS results indicate that the association of the PGS with mental health is similar in Add Health and the WLS, but the association of schooling attainment is much larger in Add Health than in the WLS. There is some suggestive evidence that the association of the PGS with mental health is lower for more-schooled older individuals in the WLS, but there is no evidence of any significant GxE associations in Add Health. Quantile regression estimates also show that in the WLS the GxE associations are statistically significant only in the upper parts of the conditional depressive symptoms score distribution. We assess the robustness of the OLS results to omitted variable bias by using the siblings samples in both datasets to estimate sibling fixed-effect regressions. The sibling fixed-effect results must be qualified, in part due to low statistical power. However, the sibling fixed-effect estimates show that college education is associated with fewer depressive symptoms in both datasets.

Both direct genetic effects (effects of alleles in an individual on that individual) and indirect genetic effects -- effects of alleles in an individual (e.g. parents) on another individual (e.g. offspring) -- can contribute to phenotypic variation and genotype-phenotype associations. Here, we consider a phenotype affected by direct and parental indirect genetic effects under assortative mating at equilibrium. We generalize classical theory to derive a decomposition of the equilibrium phenotypic variance in terms of direct and indirect genetic effect components. We extend this theory to show that popular methods for estimating indirect genetic effects or genetic nurture through analysis of parental and offspring polygenic predictors (called polygenic indices or scores -- PGIs or PGSs) are substantially biased by assortative mating. We propose an improved method for estimating indirect genetic effects while accounting for assortative mating that can also correct heritability estimates for bias due to assortative mating. We validate our method in simulations and apply it to PGIs for height and educational attainment (EA), estimating that the equilibrium heritability of height is 0.699 (S.E. = 0.075) and finding no evidence for indirect genetic effects on height. We estimate a very high correlation between parents underlying genetic components for EA, 0.755 (S.E. = 0.035), which is inconsistent with twin based estimates of the heritability of EA, possibly due to confounding in the EA PGI and/or in twin studies. We implement our method in the software package snipar, enabling researchers to apply the method to data including observed and/or imputed parental genotypes. We provide a theoretical framework for understanding the results of PGI analyses and a practical methodology for estimating heritability and indirect genetic effects while accounting for assortative mating.

Attention-Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder, which in some cases occurs comorbid with aggressive and antisocial behavior (AGG; ASB). The three externalizing behaviors are moderately to highly heritable and are genetically correlated. However, the genomic regions underlying this correlation are unknown. In this study, we aimed to localize genetic loci shared between ADHD, AGG, and ASB, using two complementary approaches. GWAS summary statistics for ADHD, AGG, and ASB were used for (1) cross-trait gene-based meta-analysis association analyses and (2) local genetic correlation analyses to identify shared genetic loci. Results of both complementary methods were combined to retrieve overlapping genes. Biological functionality of prioritized genes was assessed by exploring gene expression patterns in brain tissues and testing for gene-based association with (subcortical) brain regions. We confirmed previous findings that ADHD, AGG, and ASB were positively genetically correlated at a global level. We identified eleven significant genes in cross-trait gene-based meta-analyses, 31 loci shared between traits; 34 genes were identified when both approaches were combined. This study emphasizes the complex genetic architecture underlying global genetic correlations at the locus level. Converging evidence from these cross-trait analyses highlights novel candidate genes underlying biological mechanisms shared by ADHD, AGG, and ASB.

Polygenic scores for educational attainment and intelligence (eduPGS), genetic ancestry, and cognitive ability have been found to be inter-correlated in some admixed American populations. We argue that this could either be due to causally-relevant genetic differences between ancestral groups or be due to population stratification-related confounding. Moreover, we argue that it is important to determine which scenario is the case so to better assess the validity of eduPGS. We investigate the confounding vs. causal concern by examining, in detail, the relation between eduPGS, ancestry, and general cognitive ability in East Coast Hispanic and non-Hispanic samples. European ancestry was correlated with g in the admixed Hispanic (r = .30, N = 506), European-African (r = .26, N = 228), and African (r = .084, N = 2,179) American samples. Among Hispanics and the combined sample, these associations were robust to controls for racial / ethnic self-identification, genetically predicted color, and parental education. Additionally, eduPGS predicted g among Hispanics (B = 0.175, N = 506) and all other groups (European: B = 0.230, N = 4914; European-African: B = 0.215, N = 228; African: B = 0.126, N = 2179) with controls for ancestry. Path analyses revealed that eduPGS, but not color, partially statistically explained the association between g and European ancestry among both Hispanics and the combined sample. Of additional note, we were unable to account for eduPGS differences between ancestral populations using common tests for ascertainment bias and confounding related to population stratification. Overall, our results suggest that eduPGS derived from European samples can be used to predict g in American populations. However, owing to the uncertain cause of the differences in eduPGS, it is not yet clear how the effect of ancestry should be handled. We argue that more research is needed to determine the source of the relation between eduPGS, genetic ancestry, and cognitive ability.

Discovering mechanisms underlying mental illness requires disentangling genetic and environmental factors influencing mental health. Researchers have started investigating the brains role as a potential intermediate biomarker linking genes and environment to mental health; however, understanding how much genetics shapes adolescent brain function remains elusive. Using data from the ABCD study (5,247 unrelated, 330 dizygotic, 248 monozygotic twin subjects), we estimated the heritability of functional connectivity and topography using both SNP and twin data. SNP-based heritability was calculated using genetic correlation and the recently developed AdjHE-RE estimator. We found low SNP heritability for brain functional connectivity (median = 2e-10% Gordon, 5.8e-7% probabilistic) and topography (max = 2%). Twin estimates using ACE models replicated prior findings from the literature (median = 8.9e-6% Gordon, 6% probabilistic, 27% topography). This suggest that additive genetic effects are minimally associated with functional brain features in adolescents highlighting the importance of considering both genetic and environmental factors when studying the development of functional brain networks and their relevance to mental health.

Wellbeing is commonly defined as the combination of feeling good and functioning well and typically conceptualized as two related but distinct components. Hedonic wellbeing emphasizes pleasure, happiness, and life satisfaction, while eudaimonic wellbeing focuses on meaning, personal growth, flourishing, and the realization of ones potential. The Mental Health Continuum-Short Form was developed as a comprehensive measure of wellbeing and includes three subscales assessing emotional, social, and psychological wellbeing. Although the Mental Health Continuum total score is often interpreted as an indicator of overall wellbeing, the underlying genetic structure of its three subscales and its genetic overlap with other commonly used wellbeing measures remains unclear. Using data from 5,212 individuals from the Netherlands Twin Register (72% female, mean age 36.4), we fitted multivariate twin models to examine the genetic architecture of the Mental Health Continuum and its associations with other wellbeing measures (quality of life, life satisfaction, subjective happiness, and flourishing). Results indicate that, at the genetic level, the Mental Health Continuum is best explained by its three distinct subscales rather than by a latent factor. When considering the Mental Health Continuum together with the other wellbeing measures, we found moderate to high genetic correlations (r = 0.52 - 0.83), indicating substantial overlap in the genetics underlying the wellbeing constructs. However, we did not find evidence for a single common genetic factor underlying all constructs. These findings highlight the multidimensional structure of wellbeing, but the moderate to high genetic correlations across measures suggest that it is important to align the level of measurement (phenotypic vs genetic) with the research question.

Researchers often claim that sibling analysis can be used to separate causal genetic effects from the assortment of biases that contaminate most downstream genetic studies. Indeed, typical results from sibling models show large (>50%) attenuations in the associations between polygenic scores and phenotypes compared to non-sibling models, consistent with researchers expectations about bias reduction. This paper explores these expectations by using family (quad) data and simulations that include indirect genetic effect processes and evaluates the ability of sibling models to uncover direct genetic effects. We find that sibling models, in general, fail to uncover direct genetic effects; indeed, these models have both upward and downward biases that are difficult to sign in typical data. When genetic nurture effects exist, sibling models create "measurement error" that attenuate associations between polygenic scores and phenotypes. As the correlation between direct and indirect effect changes, this bias can increase or decrease. Our findings suggest that interpreting results from sibling analysis aimed at uncovering direct genetic effects should be treated with caution.

Both genes and the neighborhood are important for life satisfaction; however, there is little research on gene-environment interactions (GxE) that examines how the effect of genetic endowments varies as a function of the environmental context with life satisfaction as the outcome. This study investigated how neighborhood deprivation moderates the effects of genetic predisposition on life satisfaction. Using data from the German Twin Family Panel (TwinLife), we identified 760 dizygotic (DZ) twins and employed twin fixed-effect models to assess the GxE effects on life satisfaction. The findings reveal that the polygenic score (PGS) for subjective well-being is positively associated with life satisfaction. The effect of PGS for subjective well-being on life satisfaction is strongest for individuals living in moderately deprived areas, while it is weaker for those living in highly deprived and less deprived areas. Thus, there are signs of compensation in less deprived areas and, particularly, diathesis-stress/triggering in highly deprived areas.

The German Twin Family Panel TwinLife is a nationwide longitudinal study of twins and their family members. Primarily focusing on the development of social inequalities over the life course, TwinLife has been collecting data since October 2014 starting with 4,096 twin families (Ntotal = 16,951 individuals). As Germanys largest twin study to date, TwinLife has been surveying four birth cohorts of monozygotic and dizygotic same-sex twin pairs (initially [~]5, 11, 17, and 23 years old) and their families for 11 years. Survey data have been collected through five biennial face-to-face interviews with four computer-assisted telephone interviews in the years between. In addition, saliva samples were collected before the COVID-19 pandemic (2018-2020), during the pandemic (2021), and after (2022-2024). In this Cohort Profile, we describe the curation and initial analyses of molecular genetic and epigenetic data from the two TwinLife satellite projects TwinSNPs and TECS. Together, these projects currently comprise 12,108 processed DNA samples from 6,450 participants, extracted from the first two saliva collections before and during the COVID-19 pandemic. We compared the subsamples with the overall TwinLife sample and provide an overview of derived polygenic scores (PGS), epigenetic clocks and other methylation profile scores (MPS). We found that PGS predicted sample attrition in TwinLife, with small but significant associations between higher PGS for educational attainment and continued participation. Epigenetic clocks derived from saliva were highly correlated with chronological age (r = .71 to r = .94) and were generally more stable over time than other MPS. PGS for epigenetic clocks were associated with the respective clock only during but not before the start of the pandemic. We discuss opportunities of combining prospectively assessed molecular (epi)genetic data in within-family designs such as TwinLife and its implications and avenues for future research.

Gene-environment interaction with regards to educational attainment has received increasing attention during the last few years. However, the potential interdependence between different types of environments in gene-environment interaction models has mostly been neglected. Using high-quality register data for an extensive panel of Swedish twins, born during most of the twentieth century, this study explores how genetic propensities for educational attainment, as measured by a polygenic index, interact with both macro-level institutional and sociopolitical context, and with socioeconomic background. The analyses, which combine between-family and causally robust within-family models, suggest that the average association between genetic propensities and educational attainment has increased in Sweden during the twentieth century, along with the expansion of the educational system and decreased economic inequality. There is also evidence of a positive interaction between genetic propensities and socioeconomic background, but only in the oldest cohorts in the sample, and that were born before the Swedish welfare state had been fully established. This implies that micro-level gene-environment interactions can be significantly dependent on macro-level context, an insight that has arguably not yet been given sufficient attention in the literature. Acknowledging limitations of polygenic indices, and the arbitrariness of the genetic lottery, the results may nevertheless indicate a development towards higher equality of opportunity in Sweden during the twentieth century.

The developmental origins of health and disease (DOHaD) hypothesis posits that early childhood stressors disproportionately impact adult health. Numerous studies have found adult mental health to be associated with childhood adversities and genetic variants, particularly in genes related to neurochemistry. However, few studies have examined the way interactive effects may manifest over time and fewer still include protective factors, like resilience. Our group has previously found associations between the monoamine oxidase A gene, MAOA, and a contextually-specific measure of resilience with a measure of perceived psychosocial stress over time in Syrian refugee youth. In this study, we work with the same sample of adolescents to test genetic variants in three additional candidate genes (FAAH, the 5-HTTLPR region of SLC6A4, and BDNF) for associations with six psychosocial stress and mental health outcomes. Using multi-level modeling, we find no association between variants in these candidate genes and psychosocial stress or mental health outcomes. Our analysis included tests for both direct genetic effects and interactions with lifetime trauma and resilience. Negative results, such as the lack of genetic associations with outcome measures, provides a more complete framework in which to better understand positive results and associations.