Polygenic Prediction of Weight and Obesity Trajectories from Birth to Adulthood.

Severe obesity is a rapidly growing global health threat. Although often attributed to unhealthy lifestyle choices or environmental factors, obesity is known to be heritable and highly polygenic; the majority of inherited susceptibility is related to the cumulative effect of many common DNA variants. Here we derive and validate a new polygenic predictor comprised of 2.1 million common variants to quantify this susceptibility and test this predictor in more than 300,000 individuals ranging from middle age to birth. Among middle-aged adults, we observe a 13-kg gradient in weight and a 25-fold gradient in risk of severe obesity across polygenic score deciles. In a longitudinal birth cohort, we note minimal differences in birthweight across score deciles, but a significant gradient emerged in early childhood and reached 12 kg by 18 years of age. This new approach to quantify inherited susceptibility to obesity affords new opportunities for clinical prevention and mechanistic assessment.

shaPRS: Leveraging shared genetic effects across traits or ancestries improves accuracy of polygenic scores.

We present shaPRS, a method that leverages widespread pleiotropy between traits or shared genetic effects across ancestries, to improve the accuracy of polygenic scores. The method uses genome-wide summary statistics from two diseases or ancestries to improve the genetic effect estimate and standard error at SNPs where there is homogeneity of effect between the two datasets. When there is significant evidence of heterogeneity, the genetic effect from the disease or population closest to the target population is maintained. We show via simulation and a series of real-world examples that shaPRS substantially enhances the accuracy of polygenic risk scores (PRSs) for complex diseases and greatly improves PRS performance across ancestries. shaPRS is a PRS pre-processing method that is agnostic to the actual PRS generation method, and as a result, it can be integrated into existing PRS generation pipelines and continue to be applied as more performant PRS methods are developed over time.

Power of inclusion: Enhancing polygenic prediction with admixed individuals.

Admixed individuals offer unique opportunities for addressing limited transferability in polygenic scores (PGSs), given the substantial trans-ancestry genetic correlation in many complex traits. However, they are rarely considered in PGS training, given the challenges in representing ancestry-matched linkage-disequilibrium reference panels for admixed individuals. Here we present inclusive PGS (iPGS), which captures ancestry-shared genetic effects by finding the exact solution for penalized regression on individual-level data and is thus naturally applicable to admixed individuals. We validate our approach in a simulation study across 33 configurations with varying heritability, polygenicity, and ancestry composition in the training set. When iPGS is applied to n = 237,055 ancestry-diverse individuals in the UK Biobank, it shows the greatest improvements in Africans by 48.9% on average across 60 quantitative traits and up to 50-fold improvements for some traits (neutrophil count, R2 = 0.058) over the baseline model trained on the same number of European individuals. When we allowed iPGS to use n = 284,661 individuals, we observed an average improvement of 60.8% for African, 11.6% for South Asian, 7.3% for non-British White, 4.8% for White British, and 17.8% for the other individuals. We further developed iPGS+refit to jointly model the ancestry-shared and -dependent genetic effects when heterogeneous genetic associations were present. For neutrophil count, for example, iPGS+refit showed the highest predictive performance in the African group (R2 = 0.115), which exceeds the best predictive performance for the White British group (R2 = 0.090 in the iPGS model), even though only 1.49% of individuals used in the iPGS training are of African ancestry. Our results indicate the power of including diverse individuals for developing more equitable PGS models.

The contributions of rare inherited and polygenic risk to ASD in multiplex families.

Autism spectrum disorder (ASD) has a complex genetic architecture involving contributions from both de novo and inherited variation. Few studies have been designed to address the role of rare inherited variation or its interaction with common polygenic risk in ASD. Here, we performed whole-genome sequencing of the largest cohort of multiplex families to date, consisting of 4,551 individuals in 1,004 families having two or more autistic children. Using this study design, we identify seven previously unrecognized ASD risk genes supported by a majority of rare inherited variants, finding support for a total of 74 genes in our cohort and a total of 152 genes after combined analysis with other studies. Autistic children from multiplex families demonstrate an increased burden of rare inherited protein-truncating variants in known ASD risk genes. We also find that ASD polygenic score (PGS) is overtransmitted from nonautistic parents to autistic children who also harbor rare inherited variants, consistent with combinatorial effects in the offspring, which may explain the reduced penetrance of these rare variants in parents. We also observe that in addition to social dysfunction, language delay is associated with ASD PGS overtransmission. These results are consistent with an additive complex genetic risk architecture of ASD involving rare and common variation and further suggest that language delay is a core biological feature of ASD.

Cumulative Effects of Resting-State Connectivity Across All Brain Networks Significantly Correlate with Attention-Deficit Hyperactivity Disorder Symptoms.

Identification of replicable neuroimaging correlates of attention-deficit hyperactivity disorder (ADHD) has been hindered by small sample sizes, small effects, and heterogeneity of methods. Given evidence that ADHD is associated with alterations in widely distributed brain networks and the small effects of individual brain features, a whole-brain perspective focusing on cumulative effects is warranted. The use of large, multisite samples is crucial for improving reproducibility and clinical utility of brain-wide MRI association studies. To address this, a polyneuro risk score (PNRS) representing cumulative, brain-wide, ADHD-associated resting-state functional connectivity was constructed and validated using data from the Adolescent Brain Cognitive Development (ABCD, N = 5,543, 51.5% female) study, and was further tested in the independent Oregon-ADHD-1000 case-control cohort (N = 553, 37.4% female). The ADHD PNRS was significantly associated with ADHD symptoms in both cohorts after accounting for relevant covariates (p < 0.001). The most predictive PNRS involved all brain networks, though the strongest effects were concentrated among the default mode and cingulo-opercular networks. In the longitudinal Oregon-ADHD-1000, non-ADHD youth had significantly lower PNRS (Cohen's d = -0.318, robust p = 5.5 × 10-4) than those with persistent ADHD (age 7-19). The PNRS, however, did not mediate polygenic risk for ADHD. Brain-wide connectivity was robustly associated with ADHD symptoms in two independent cohorts, providing further evidence of widespread dysconnectivity in ADHD. Evaluation in enriched samples demonstrates the promise of the PNRS approach for improving reproducibility in neuroimaging studies and unraveling the complex relationships between brain connectivity and behavioral disorders.

Advances and Applications of Polygenic Scores for Coronary Artery Disease.

Polygenic scores quantify inherited risk by integrating information from many common sites of DNA variation into a single number. Rapid increases in the scale of genetic association studies and new statistical algorithms have enabled development of polygenic scores that meaningfully measure-as early as birth-risk of coronary artery disease. These newer-generation polygenic scores identify up to 8% of the population with triple the normal risk based on genetic variation alone, and these individuals cannot be identified on the basis of family history or clinical risk factors alone. For those identified with increased genetic risk, evidence supports risk reduction with at least two interventions, adherence to a healthy lifestyle and cholesterol-lowering therapies, that can substantially reduce risk. Alongside considerable enthusiasm for the potential of polygenic risk estimation to enable a new era of preventive clinical medicine is recognition of a need for ongoing research into how best to ensure equitable performance across diverse ancestries, how and in whom to assess the scores in clinical practice, as well as randomized trials to confirm clinical utility.

Polygenic subtype identified in ACCORD trial displays a favorable type 2 diabetes phenotype in the UKBiobank population.

INTRODUCTION: We previously identified a genetic subtype (C4) of type 2 diabetes (T2D), benefitting from intensive glycemia treatment in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial. Here, we characterized the population of patients that met the C4 criteria in the UKBiobank cohort. RESEARCH DESIGN AND METHODS: Using our polygenic score (PS), we identified C4 individuals in the UKBiobank and tested C4 status with risk of developing T2D, cardiovascular disease (CVD) outcomes, and differences in T2D medications. RESULTS: C4 individuals were less likely to develop T2D, were slightly older at T2D diagnosis, had lower HbA1c values, and were less likely to be prescribed T2D medications (P < .05). Genetic variants in MAS1 and IGF2R, major components of the C4 PS, were associated with fewer overall T2D prescriptions. CONCLUSION: We have confirmed C4 individuals are a lower risk subpopulation of patients with T2D.

Evaluating interactions of polygenic risk scores and NAT2 genotypes with tobacco smoking in bladder cancer risk.

Tobacco smoking is the most important risk factor for bladder cancer. Previous studies have identified the N-acetyltransferase (NAT2) gene in association with bladder cancer risk. The NAT2 gene encodes an enzyme that metabolizes aromatic amines, carcinogens commonly found in tobacco smoke. In our study, we evaluated potential interactions of tobacco smoking with NAT2 genotypes and polygenic risk score (PRS) for bladder cancer, using data from the UK Biobank, a large prospective cohort study. We used Cox proportional hazards models to measure the strength of the association. The PRS was derived using genetic risk variants identified by genome-wide association studies for bladder cancer. With an average of 10.1 years of follow-up of 390 678 eligible participants of European descent, 769 incident bladder cancer cases were identified. Current smokers with a PRS in the highest tertile had a higher risk of developing bladder cancer (HR: 6.45, 95% CI: 4.51-9.24) than current smokers with a PRS in the lowest tertile (HR: 2.41, 95% CI: 1.52-3.84; P for additive interaction = <.001). A similar interaction was found for genetically predicted metabolizing NAT2 phenotype and tobacco smoking where current smokers with the slow NAT2 phenotype had an increased risk of developing bladder cancer (HR: 5.70, 95% CI: 2.64-12.30) than current smokers with the fast NAT2 phenotype (HR: 3.61, 95% CI: 1.14-11.37; P for additive interaction = .100). Our study provides support for considering both genetic and lifestyle risk factors in developing prevention measures for bladder cancer.

A polygenic-score-based approach for identification of gene-drug interactions stratifying breast cancer risk.

An individual's genetics can dramatically influence breast cancer (BC) risk. Although clinical measures for prevention do exist, non-invasive personalized measures for reducing BC risk are limited. Commonly used medications are a promising set of modifiable factors, but no previous study has explored whether a range of widely taken approved drugs modulate BC genetics. In this study, we describe a quantitative framework for exploring the interaction between the genetic susceptibility of BC and medication usage among UK Biobank women. We computed BC polygenic scores (PGSs) that summarize BC genetic risk and find that the PGS explains nearly three-times greater variation in disease risk within corticosteroid users compared to non-users. We map 35 genes significantly interacting with corticosteroid use (FDR < 0.1), highlighting the transcription factor NRF2 as a common regulator of gene-corticosteroid interactions in BC. Finally, we discover a regulatory variant strongly stratifying BC risk according to corticosteroid use. Within risk allele carriers, 18.2% of women taking corticosteroids developed BC, compared to 5.1% of the non-users (with an HR = 3.41 per-allele within corticosteroid users). In comparison, there are no differences in BC risk within the reference allele homozygotes. Overall, this work highlights the clinical relevance of gene-drug interactions in disease risk and provides a roadmap for repurposing biobanks in drug repositioning and precision medicine.

PGS-server: accuracy, robustness and transferability of polygenic score methods for biobank scale studies.

Polygenic scores (PGS) are important tools for carrying out genetic prediction of common diseases and disease related complex traits, facilitating the development of precision medicine. Unfortunately, despite the critical importance of PGS and the vast number of PGS methods recently developed, few comprehensive comparison studies have been performed to evaluate the effectiveness of PGS methods. To fill this critical knowledge gap, we performed a comprehensive comparison study on 12 different PGS methods through internal evaluations on 25 quantitative and 25 binary traits within the UK Biobank with sample sizes ranging from 147 408 to 336 573, and through external evaluations via 25 cross-study and 112 cross-ancestry analyses on summary statistics from multiple genome-wide association studies with sample sizes ranging from 1415 to 329 345. We evaluate the prediction accuracy, computational scalability, as well as robustness and transferability of different PGS methods across datasets and/or genetic ancestries, providing important guidelines for practitioners in choosing PGS methods. Besides method comparison, we present a simple aggregation strategy that combines multiple PGS from different methods to take advantage of their distinct benefits to achieve stable and superior prediction performance. To facilitate future applications of PGS, we also develop a PGS webserver (http://www.pgs-server.com/) that allows users to upload summary statistics and choose different PGS methods to fit the data directly. We hope that our results, method and webserver will facilitate the routine application of PGS across different research areas.

Genetic basis of STEM occupational choice and regional economic performance: a UK biobank genome-wide association study.

BACKGROUND: Science, technology, engineering, and mathematics (STEM) professionals are regarded as the highly skilled labor force that fosters economic productivity, enterprise innovation, and international competitiveness of a country. This study aims to understand the genetic predisposition to STEM occupations and investigate its associations with regional economic performance. We conducted a genome-wide association study on the occupational choice of STEM jobs based on a sample of 178,976 participants from the UK Biobank database. RESULTS: We identified two genetic loci significantly associated with participants' STEM job choices: rs10048736 on chromosome 2 and rs12903858 on chromosome 15. The SNP heritability of STEM occupations was estimated to be 4.2%. We also found phenotypic and genetic evidence of assortative mating in STEM occupations. At the local authority level, we found that the average polygenic score of STEM is significantly and robustly associated with several metrics of regional economic performance. CONCLUSIONS: The current study expands our knowledge of the genetic basis of occupational choice and potential regional disparities in socioeconomic developments.

Examining associations between genetic and neural risk for externalizing behaviors in adolescence and early adulthood.

BACKGROUND: Researchers have identified genetic and neural risk factors for externalizing behaviors. However, it has not yet been determined if genetic liability is conferred in part through associations with more proximal neurophysiological risk markers. METHODS: Participants from the Collaborative Study on the Genetics of Alcoholism, a large, family-based study of alcohol use disorders were genotyped and polygenic scores for externalizing (EXT PGS) were calculated. Associations with target P3 amplitude from a visual oddball task (P3) and broad endorsement of externalizing behaviors (indexed via self-report of alcohol and cannabis use, and antisocial behavior) were assessed in participants of European (EA; N = 2851) and African ancestry (AA; N = 1402). Analyses were also stratified by age (adolescents, age 12-17 and young adults, age 18-32). RESULTS: The EXT PGS was significantly associated with higher levels of externalizing behaviors among EA adolescents and young adults as well as AA young adults. P3 was inversely associated with externalizing behaviors among EA young adults. EXT PGS was not significantly associated with P3 amplitude and therefore, there was no evidence that P3 amplitude indirectly accounted for the association between EXT PGS and externalizing behaviors. CONCLUSIONS: Both the EXT PGS and P3 amplitude were significantly associated with externalizing behaviors among EA young adults. However, these associations with externalizing behaviors appear to be independent of each other, suggesting that they may index different facets of externalizing.

Mental illness and cardiovascular health: observational and polygenic score analyses in a population-based cohort study.

BACKGROUND: Individuals with serious mental illness have a markedly shorter life expectancy. A major contributor to premature death is cardiovascular disease (CVD). We investigated associations of (genetic liability for) depressive disorder, bipolar disorder and schizophrenia with a range of CVD traits and examined to what degree these were driven by important confounders. METHODS: We included participants of the Dutch Lifelines cohort (N = 147 337) with information on self-reported lifetime diagnosis of depressive disorder, bipolar disorder, or schizophrenia and CVD traits. Employing linear mixed-effects models, we examined associations between mental illness diagnoses and CVD, correcting for psychotropic medication, demographic and lifestyle factors. In a subsample (N = 73 965), we repeated these analyses using polygenic scores (PGSs) for the three mental illnesses. RESULTS: There was strong evidence that depressive disorder diagnosis is associated with increased arrhythmia and atherosclerosis risk and lower heart rate variability, even after confounder adjustment. Positive associations were also found for the depression PGSs with arrhythmia and atherosclerosis. Bipolar disorder was associated with a higher risk of nearly all CVD traits, though most diminished after adjustment. The bipolar disorder PGSs did not show any associations. While the schizophrenia PGSs was associated with increased arrhythmia risk and lower heart rate variability, schizophrenia diagnosis was not. All mental illness diagnoses were associated with lower blood pressure and a lower risk of hypertension. CONCLUSIONS: Our study shows widespread associations of (genetic liability to) mental illness (primarily depressive disorder) with CVD, even after confounder adjustment. Future research should focus on clarifying potential causal pathways between mental illness and CVD.

The risk of venous thromboembolism in oral contraceptive users: the role of genetic factors-a prospective cohort study of 240,000 women in the UK Biobank.

BACKGROUND: More than 150 million women worldwide use oral contraceptives. Women with inherited thrombophilia and carriers of certain thrombophilia gene variants, such as factor V Leiden and the prothrombin, are at an increased risk for venous thromboembolism, especially when combined with oral contraceptive use. Venous thromboembolism is a complex disorder involving many genetic risk factors, and recently, polygenic risk scores have been proposed to capture a significant proportion of the genetic risk of venous thromboembolism. OBJECTIVE: The aim of this study was to estimate the risk for developing venous thromboembolism when initiating oral contraceptive use (first 2 years) and during continued use among women with a high genetic liability. STUDY DESIGN: We used a prospective study design in which 244,420 participants from the UK Biobank were followed from birth. The effect of oral contraceptive use during the first 2 years and in the remaining years of oral contraceptive use on the risk of developing venous thromboembolism was estimated using a Cox regression with a time-dependent exposure variable. Women were stratified according to their polygenic risk scores and whether they were carriers of factor V Leiden and/or prothrombin variants. RESULTS: When genetic risk was not considered, an increased risk for venous thromboembolism was observed during the first 2 years of oral contraceptive use (hazard ratio, 3.09; 95% confidence interval, 3.00-3.20) but not during continued use (hazard ratio, 0.92; 95% confidence interval, 0.80-1.05). However, when genetic risk was considered, women in the highest polygenic risk score category had a more pronounced risk of developing a venous thromboembolism during the first 2 years of oral contraceptive use (hazard ratio, 6.35; 95% confidence interval, 4.98-8.09), and a high risk was also observed among factor V Leiden (hazard ratio, 5.73; 95% confidence interval, 5.31-6.17) and prothrombin variant carriers (hazard ratio, 5.23; 95% confidence interval, 4.67 - 5.87). A high polygenic risk score in combination with being a factor V Leiden and prothrombin variant carrier conferred the highest risk for developing a venous thromboembolism during the first 2 years of oral contraceptive use (hazard ratio, 14.8; 95% confidence interval, 9.28-23.6). Women with a high genetic liability also had an increased risk during continued use but it was less pronounced, and the highest risk was conferred to carriers of both factor V Leiden and the prothrombin variant (hazard ratio, 4.93; 95% confidence interval, 3.16-7.7). CONCLUSION: Evaluating polygenic risk can identify additional venous thromboembolism risk that is not captured in the commonly investigated genes for inherited thrombophilia. Our results indicate that oral contraceptive use is associated with an increased risk for developing a venous thromboembolism, particularly among women with a high genetic predisposition, and that oral contraceptive use dramatically increases the risk thereof short after initiation of use, which decreases with continued use. This suggests that the polygenic risk score could be used to identify women who are at high risk for developing a venous thromboembolism and advise them on alternative methods of contraception.

The interaction between polygenic risk and environmental influences: A direct test of the 3P model of insomnia in adolescents.

BACKGROUND: Stress is a universal phenomenon and one of the most common precipitants of insomnia. However, not everyone develops insomnia after experiencing a stressful life event. This study aims to test aspects of Spielman's '3P model of insomnia' (during adolescence) by exploring the extent to which: (a) insomnia symptoms are predicted by polygenic scores (PGS); (b) life events predict insomnia symptoms; (c) the interaction between PGS and life events contribute to the prediction of insomnia symptoms; (d) gene-environment interaction effects remain after controlling for sex. METHODS: The sample comprised 4,629 twins aged 16 from the Twin Early Development Study who reported on their insomnia symptoms and life events. PGS for insomnia were calculated. In order to test the main hypothesis of this study (a significant interaction between PGS and negative life events), we fitted a series of mixed effect regressions. RESULTS: The best fit was provided by the model including sex, PGS for insomnia, negative life events, and their interactions (AIC = 26,158.7). Our results show that the association between insomnia symptoms and negative life events is stronger for those with a higher genetic risk for insomnia. CONCLUSIONS: This work sheds light on the complex relationship between genetic and environmental factors implicated for insomnia. This study has tested for the first time the interaction between genetic predisposition (PGS) for insomnia and environmental stressors (negative life events) in adolescents. This work represents a direct test of components of Spielman's 3P model for insomnia which is supported by our results.

Mapping potential pathways from polygenic liability through brain structure to psychological problems across the transition to adolescence.

BACKGROUND: We used a polygenic score for externalizing behavior (extPGS) and structural MRI to examine potential pathways from genetic liability to conduct problems via the brain across the adolescent transition. METHODS: Three annual assessments of child conduct problems, attention-deficit/hyperactivity problems, and internalizing problems were conducted across across 9-13 years of age among 4,475 children of European ancestry in the Adolescent Brain Cognitive DevelopmentSM Study (ABCD Study®). RESULTS: The extPGS predicted conduct problems in each wave (R2 = 2.0%-2.9%). Bifactor models revealed that the extPRS predicted variance specific to conduct problems (R2 = 1.7%-2.1%), but also variance that conduct problems shared with other measured problems (R2 = .8%-1.4%). Longitudinally, extPGS predicted levels of specific conduct problems (R2 = 2.0%), but not their slope of change across age. The extPGS was associated with total gray matter volume (TGMV; R2 = .4%) and lower TGMV predicted both specific conduct problems (R2 = 1.7%-2.1%) and the variance common to all problems in each wave (R2 = 1.6%-3.1%). A modest proportion of the polygenic liability specific to conduct problems in each wave was statistically mediated by TGMV. CONCLUSIONS: Across the adolescent transition, the extPGS predicted both variance specific to conduct problems and variance shared by all measured problems. The extPGS also was associated with TGMV, which robustly predicted conduct problems. Statistical mediation analyses suggested the hypothesis that polygenic variation influences individual differences in brain development that are related to the likelihood of conduct problems during the adolescent transition, justifying new research to test this causal hypothesis.

Brain volumes, thicknesses, and surface areas as mediators of genetic factors and childhood adversity on intelligence.

Although genetic and environmental factors influence general intelligence (g-factor), few studies examined the neuroanatomical measures mediating environmental and genetic effects on intelligence. Here, we investigate the brain volumes, cortical mean thicknesses, and cortical surface areas mediating the effects of the g-factor polygenic score (gPGS) and childhood adversity on the g-factor in the UK Biobank. We first examined the global and regional brain measures that contribute to the g-factor. Most regions contributed to the g-factor through global brain size. Parieto-frontal integration theory (P-FIT) regions were not more associated with the g-factor than non-PFIT regions. After adjusting for global brain size and regional associations, only a few regions predicted intelligence and were included in the mediation analyses. We conducted mediation analyses on global measures, regional volumes, mean thicknesses, and surface areas, separately. Total brain volume mediated 7.04% of the gPGS' effect on the g-factor and 2.50% of childhood adversity's effect on the g-factor. In comparison, the fraction of the gPGS and childhood adversity's effects mediated by individual regional volumes, surfaces, and mean thicknesses was 10-15 times smaller. Therefore, genetic and environmental effects on intelligence may be mediated to a larger extent by other brain properties.

Using polygenic scores in combination with symptom rating scales to identify attention-deficit/hyperactivity disorder.

BACKGROUND: The inclusion of biomarkers could improve diagnostic accuracy of attention-deficit/hyperactivity disorder (ADHD). One potential biomarker is the ADHD polygenic score (PGS), a measure of genetic liability for ADHD. This study aimed to investigate if the ADHD PGS can provide additional information alongside ADHD rating scales and examination of family history of ADHD to distinguish between ADHD cases and controls. METHODS: Polygenic scores were calculated for 576 adults with ADHD and 530 ethnically matched controls. ADHD PGS was used alongside scores from the Wender-Utah Rating Scale (WURS) and the Adult ADHD Self-Report Scale (ASRS) as predictors of ADHD diagnosis in a set of nested logistic regression models. These models were compared by likelihood ratio (LR) tests, Akaike information criterion corrected for small samples (AICc), and Lee R². These analyses were repeated with family history of ADHD as a covariate in all models. RESULTS: The ADHD PGS increased the variance explained of the ASRS by 0.58% points (pp) (R2ASRS = 61.11%, R2ASRS + PGS=61.69%), the WURS by 0.61pp (R2WURS = 77.33%, R2WURS + PGS= 77.94%), of ASRS and WURS together by 0.57pp (R2ASRS + WURS=80.84%, R2ASRS + WURS+PGS=81.40%), and of self-reported family history by 1.40pp (R2family = 28.06%, R2family + PGS=29.46%). These increases were statistically significant, as measured by LR tests and AICc. CONCLUSION: We found that the ADHD PGS contributed additional information to common diagnostic aids. However, the increase in variance explained was small, suggesting that the ADHD PGS is currently not a clinically useful diagnostic aid. Future studies should examine the utility of ADHD PGS in ADHD prediction alongside non-genetic risk factors, and the diagnostic utility of the ADHD PGS should be evaluated as more genetic data is accumulated and computational tools are further refined.

Discrimination between healthy participants and people with panic disorder based on polygenic scores for psychiatric disorders and for intermediate phenotypes using machine learning.

OBJECTIVE: Panic disorder is a modestly heritable condition. Currently, diagnosis is based only on clinical symptoms; identifying objective biomarkers and a more reliable diagnostic procedure is desirable. We investigated whether people with panic disorder can be reliably diagnosed utilizing combinations of multiple polygenic scores for psychiatric disorders and their intermediate phenotypes, compared with single polygenic score approaches, by applying specific machine learning techniques. METHODS: Polygenic scores for 48 psychiatric disorders and intermediate phenotypes based on large-scale genome-wide association studies (n = 7556-1,131,881) were calculated for people with panic disorder (n = 718) and healthy controls (n = 1717). Discrimination between people with panic disorder and healthy controls was based on the 48 polygenic scores using five methods for classification: logistic regression, neural networks, quadratic discriminant analysis, random forests and a support vector machine. Differences in discrimination accuracy (area under the curve) due to an increased number of polygenic score combinations and differences in the accuracy across five classifiers were investigated. RESULTS: All five classifiers performed relatively well for distinguishing people with panic disorder from healthy controls by increasing the number of polygenic scores. Of the 48 polygenic scores, the polygenic score for anxiety UK Biobank was the most useful for discrimination by the classifiers. In combinations of two or three polygenic scores, the polygenic score for anxiety UK Biobank was included as one of polygenic scores in all classifiers. When all 48 polygenic scores were used in combination, the greatest areas under the curve significantly differed among the five classifiers. Support vector machine and logistic regression had higher accuracy than quadratic discriminant analysis and random forests. For each classifier, the greatest area under the curve was 0.600 ± 0.030 for logistic regression (polygenic score combinations N = 14), 0.591 ± 0.039 for neural networks (N = 9), 0.603 ± 0.033 for quadratic discriminant analysis (N = 10), 0.572 ± 0.039 for random forests (N = 25) and 0.617 ± 0.041 for support vector machine (N = 11). The greatest areas under the curve at the best polygenic score combination significantly differed among the five classifiers. Random forests had the lowest accuracy among classifiers. Support vector machine had higher accuracy than neural networks. CONCLUSIONS: These findings suggest that increasing the number of polygenic score combinations up to approximately 10 effectively improved the discrimination accuracy and that support vector machine exhibited greater accuracy among classifiers. However, the discrimination accuracy for panic disorder, when based solely on polygenic score combinations, was found to be modest.

A polygenic score for educational attainment partially predicts voter turnout.

Twin and adoption studies have shown that individual differences in political participation can be explained, in part, by genetic variation. However, these research designs cannot identify which genes are related to voting or the pathways through which they exert influence, and their conclusions rely on possibly restrictive assumptions. In this study, we use three different US samples and a Swedish sample to test whether genes that have been identified as associated with educational attainment, one of the strongest correlates of political participation, predict self-reported and validated voter turnout. We find that a polygenic score capturing individuals' genetic propensity to acquire education is significantly related to turnout. The strongest associations we observe are in second-order midterm elections in the United States and European Parliament elections in Sweden, which tend to be viewed as less important by voters, parties, and the media and thus present a more information-poor electoral environment for citizens to navigate. A within-family analysis suggests that individuals' education-linked genes directly affect their voting behavior, but, for second-order elections, it also reveals evidence of genetic nurture. Finally, a mediation analysis suggests that educational attainment and cognitive ability combine to account for between 41% and 63% of the relationship between the genetic propensity to acquire education and voter turnout.