Contrasting genetic architectures of schizophrenia and other complex diseases using fast variance-components analysis

Relationship of Fat Mass Ratio, a Biomarker for Lipodystrophy, With Cardiometabolic Traits

Familial partial lipodystrophy (FPLD) is a heterogenous group of syndromes associated with a high prevalence of cardiometabolic diseases. Prior work has proposed DEXA-derived fat mass ratio (FMR), defined as trunk fat percentage divided by leg fat percentage, as a biomarker of FPLD, but this metric has not previously been characterized in large cohort studies. We set out to 1) understand the cardiometabolic burden of individuals with high FMR in up to 40,796 participants in the UK Biobank and 9,408 participants in the Fenland study, 2) characterize the common variant genetic underpinnings of FMR, and 3) build and test a polygenic predictor for FMR. Participants with high FMR were at higher risk for type 2 diabetes (odds ratio [OR] 2.30, P = 3.5 × 10-41) and metabolic dysfunction-associated liver disease or steatohepatitis (OR 2.55, P = 4.9 × 10-7) in UK Biobank and had higher fasting insulin (difference 19.8 pmol/L, P = 5.7 × 10-36) and fasting triglycerides (difference 36.1 mg/dL, P = 2.5 × 10-28) in the Fenland study. Across FMR and its component traits, 61 conditionally independent variant-trait pairs were discovered, including 13 newly identified pairs. A polygenic score for FMR was associated with an increased risk of cardiometabolic diseases. This work establishes the cardiometabolic significance of high FMR, a biomarker for FPLD, in two large cohort studies and may prove useful in increasing diagnosis rates of patients with metabolically unhealthy fat distribution to enable treatment or a preventive therapy.

ARTICLE HIGHLIGHTS

Genetic drivers and cellular selection of female mosaic X chromosome loss

Mosaic loss of the X chromosome (mLOX) is the most common clonal somatic alteration in leukocytes of female individuals1,2, but little is known about its genetic determinants or phenotypic consequences. Here, to address this, we used data from 883,574 female participants across 8 biobanks; 12% of participants exhibited detectable mLOX in approximately 2% of leukocytes. Female participants with mLOX had an increased risk of myeloid and lymphoid leukaemias. Genetic analyses identified 56 common variants associated with mLOX, implicating genes with roles in chromosomal missegregation, cancer predisposition and autoimmune diseases. Exome-sequence analyses identified rare missense variants in FBXO10 that confer a twofold increased risk of mLOX. Only a small fraction of associations was shared with mosaic Y chromosome loss, suggesting that distinct biological processes drive formation and clonal expansion of sex chromosome missegregation. Allelic shift analyses identified X chromosome alleles that are preferentially retained in mLOX, demonstrating variation at many loci under cellular selection. A polygenic score including 44 allelic shift loci correctly inferred the retained X chromosomes in 80.7% of mLOX cases in the top decile. Our results support a model in which germline variants predispose female individuals to acquiring mLOX, with the allelic content of the X chromosome possibly shaping the magnitude of clonal expansion.

Noninvasive assessment of organ-specific and shared pathways in multi-organ fibrosis using T1 mapping

Fibrotic diseases affect multiple organs and are associated with morbidity and mortality. To examine organ-specific and shared biologic mechanisms that underlie fibrosis in different organs, we developed machine learning models to quantify T1 time, a marker of interstitial fibrosis, in the liver, pancreas, heart and kidney among 43,881 UK Biobank participants who underwent magnetic resonance imaging. In phenome-wide association analyses, we demonstrate the association of increased organ-specific T1 time, reflecting increased interstitial fibrosis, with prevalent diseases across multiple organ systems. In genome-wide association analyses, we identified 27, 18, 11 and 10 independent genetic loci associated with liver, pancreas, myocardial and renal cortex T1 time, respectively. There was a modest genetic correlation between the examined organs. Several loci overlapped across the examined organs implicating genes involved in a myriad of biologic pathways including metal ion transport (SLC39A8, HFE and TMPRSS6), glucose metabolism (PCK2), blood group antigens (ABO and FUT2), immune function (BANK1 and PPP3CA), inflammation (NFKB1) and mitosis (CENPE). Finally, we found that an increasing number of organs with T1 time falling in the top quintile was associated with increased mortality in the population. Individuals with a high burden of fibrosis in ≥3 organs had a 3-fold increase in mortality compared to those with a low burden of fibrosis across all examined organs in multivariable-adjusted analysis (hazard ratio = 3.31, 95% confidence interval 1.77-6.19; P = 1.78 × 10-4). By leveraging machine learning to quantify T1 time across multiple organs at scale, we uncovered new organ-specific and shared biologic pathways underlying fibrosis that may provide therapeutic targets.

Dynamics of cognitive variability with age and its genetic underpinning in NIHR BioResource Genes and Cognition cohort participants

A leading explanation for translational failure in neurodegenerative disease is that new drugs are evaluated late in the disease course when clinical features have become irreversible. Here, to address this gap, we cognitively profiled 21,051 people aged 17-85 years as part of the Genes and Cognition cohort within the National Institute for Health and Care Research BioResource across England. We describe the cohort, present cognitive trajectories and show the potential utility. Surprisingly, when studied at scale, the APOE genotype had negligible impact on cognitive performance. Different cognitive domains had distinct genetic architectures, with one indicating brain region-specific activation of microglia and another with glycogen metabolism. Thus, the molecular and cellular mechanisms underpinning cognition are distinct from dementia risk loci, presenting different targets to slow down age-related cognitive decline. Participants can now be recalled stratified by genotype and cognitive phenotype for natural history and interventional studies of neurodegenerative and other disorders.

Genetic association of inflammatory marker GlycA with lung function and respiratory diseases

Association of circulating glycoprotein acetyls (GlycA), a systemic inflammation biomarker, with lung function and respiratory diseases remain to be investigated. We examined the genetic correlation, shared genetics, and potential causality of GlycA (N = 115,078) with lung function and respiratory diseases (N = 497,000). GlycA showed significant genetic correlation with FEV1 (rg = -0.14), FVC (rg = -0.18), asthma (rg = 0.21) and COPD (rg = 0.31). We consistently identified ten shared loci (including chr3p21.31 and chr8p23.1) at both SNP and gene level revealing potential shared biological mechanisms involving ubiquitination, immune response, Wnt/β-catenin signaling, cell growth and differentiation in tissues or cells including blood, epithelium, fibroblast, fetal thymus, and fetal intestine. Genetically elevated GlycA was significantly correlated with lung function and asthma susceptibility (354.13 ml decrement of FEV1, 442.28 ml decrement of FVC, and 144% increased risk of asthma per SD increment of GlycA) from MR analyses. Our findings provide insights into biological mechanisms of GlycA in relating to lung function, asthma, and COPD.

MPH: fast REML for large-scale genome partitioning of quantitative genetic variation

MOTIVATION

Genome partitioning of quantitative genetic variation is useful for dissecting the genetic architecture of complex traits. However, existing methods, such as Haseman-Elston regression and linkage disequilibrium score regression, often face limitations when handling extensive farm animal datasets, as demonstrated in this study.

RESULTS

To overcome this challenge, we present MPH, a novel software tool designed for efficient genome partitioning analyses using restricted maximum likelihood. The computational efficiency of MPH primarily stems from two key factors: the utilization of stochastic trace estimators and the comprehensive implementation of parallel computation. Evaluations with simulated and real datasets demonstrate that MPH achieves comparable accuracy and significantly enhances convergence, speed, and memory efficiency compared to widely used tools like GCTA and LDAK. These advancements facilitate large-scale, comprehensive analyses of complex genetic architectures in farm animals.

AVAILABILITY AND IMPLEMENTATION

The MPH software is available at https://jiang18.github.io/mph/.

The cliff edge model of the evolution of schizophrenia: Mathematical, epidemiological, and genetic evidence

How has schizophrenia, a condition that significantly reduces an individual's evolutionary fitness, remained common across generations and cultures? Numerous theories about the evolution of schizophrenia have been proposed, most of which are not consistent with modern epidemiological and genetic evidence. Here, we briefly review this evidence and explore the cliff edge model of schizophrenia. It suggests that schizophrenia is the extreme manifestation of a polygenic trait or a combination of traits that, within a normal range of variation, confer cognitive, linguistic, and/or social advantages. Only beyond a certain threshold, these traits precipitate the onset of schizophrenia and reduce fitness. We provide the first mathematical model of this qualitative concept and show that it requires only very weak positive selection of the underlying trait(s) to explain today's schizophrenia prevalence. This prediction, along with expectations about the effect size of schizophrenia risk alleles, are surprisingly well matched by empirical evidence. The cliff edge model predicts a dynamic change of selection of risk alleles, which explains the contradictory findings of evolutionary genetic studies.

Distinct explanations underlie gene-environment interactions in the UK Biobank

The role of gene-environment (GxE) interaction in disease and complex trait architectures is widely hypothesized, but currently unknown. Here, we apply three statistical approaches to quantify and distinguish three different types of GxE interaction for a given trait and E variable. First, we detect locus-specific GxE interaction by testing for genetic correlation r g < 1 across E bins. Second, we detect genome-wide effects of the E variable on genetic variance by leveraging polygenic risk scores (PRS) to test for significant PRSxE in a regression of phenotypes on PRS, E, and PRSxE, together with differences in SNP-heritability across E bins. Third, we detect genome-wide proportional amplification of genetic and environmental effects as a function of the E variable by testing for significant PRSxE with no differences in SNP-heritability across E bins. Simulations show that these approaches achieve high sensitivity and specificity in distinguishing these three GxE scenarios. We applied our framework to 33 UK Biobank traits (25 quantitative traits and 8 diseases; average N = 325 K ) and 10 E variables spanning lifestyle, diet, and other environmental exposures. First, we identified 19 trait-E pairs with r g significantly < 1 (FDR<5%) (average r g = 0.95 ); for example, white blood cell count had r g = 0.95 (s.e. 0.01) between smokers and non-smokers. Second, we identified 28 trait-E pairs with significant PRSxE and significant SNP-heritability differences across E bins; for example, BMI had a significant PRSxE for physical activity (P=4.6e-5) with 5% larger SNP-heritability in the largest versus smallest quintiles of physical activity (P=7e-4). Third, we identified 15 trait-E pairs with significant PRSxE with no SNP-heritability differences across E bins; for example, waist-hip ratio adjusted for BMI had a significant PRSxE effect for time spent watching television (P=5e-3) with no SNP-heritability differences. Across the three scenarios, 8 of the trait-E pairs involved disease traits, whose interpretation is complicated by scale effects. Analyses using biological sex as the E variable produced additional significant findings in each of the three scenarios. Overall, we infer a significant contribution of GxE and GxSex effects to complex trait and disease variance.

Heritability of Gene Expression Measured from Peripheral Blood in Older Adults

The contributions of genetic variation and the environment to gene expression may change across the lifespan. However, few studies have investigated the heritability of blood gene expression in older adults. The current study therefore aimed to investigate this question in a community sample of older adults. A total of 246 adults (71 MZ and 52 DZ twins, 69.91% females; mean age-75.79 ± 5.44) were studied. Peripheral blood gene expression was assessed using Illumina microarrays. A heritability analysis was performed using structural equation modelling. There were 5269 probes (19.9%) from 4603 unique genes (23.9%) (total 26,537 probes from 19,256 genes) that were significantly heritable (mean h2 = 0.40). A pathway analysis of the top 10% of significant genes showed enrichment for the immune response and ageing-associated genes. In a comparison with two other gene expression twin heritability studies using adults from across the lifespan, there were 38 out of 9479 overlapping genes that were significantly heritable. In conclusion, our study found ~24% of the available genes for analysis were heritable in older adults, with only a small number common across studies that used samples from across adulthood, indicating the importance of examining gene expression in older age groups.

Identification of potential drug targets for allergic diseases from a genetic perspective: A mendelian randomization study

BACKGROUND

Allergic diseases typically refer to a heterogeneous group of conditions primarily caused by the activation of mast cells or eosinophils, including atopic dermatitis (AD), allergic rhinitis (AR), and asthma. Asthma, AR, and AD collectively affect approximately one-fifth of the global population, imposing a significant economic burden on society. Despite the availability of drugs to treat allergic diseases, they have been shown to be insufficient in controlling relapses and halting disease progression. Therefore, new drug targets are needed to prevent the onset of allergic diseases.

METHOD

We employed a Mendelian randomization approach to identify potential drug targets for the treatment of allergic diseases. Leveraging 1798 genetic instruments for 1537 plasma proteins from the latest reported Genome-Wide Association Studies (GWAS), we analyzed the GWAS summary statistics of Ferreira MA et al. (nCase = 180,129, nControl = 180,709) using the Mendelian randomization method. Furthermore, we validated our findings in the GWAS data from the FinnGen and UK Biobank cohorts. Subsequently, we conducted sensitivity tests through reverse causal analysis, Bayesian colocalization analysis, and phenotype scanning. Additionally, we performed protein-protein interaction analysis to determine the interaction between causal proteins. Finally, based on the potential protein targets, we conducted molecular docking to identify potential drugs for the treatment of allergic diseases.

RESULTS

At Bonferroni significance (p < 3.25 × 10-5), the Mendelian randomization analysis revealed 11 significantly associated protein-allergic disease pairs. Among these, the increased levels of TNFAIP3, ERBB3, TLR1, and IL1RL2 proteins were associated with a reduced risk of allergic diseases, with corresponding odds ratios of 0.82 (0.76-0.88), 0.74 (0.66-0.82), 0.49 (0.45-0.55), and 0.81 (0.75-0.87), respectively. Conversely, increased levels of IL6R, IL1R1, ITPKA, IL1RL1, KYNU, LAYN, and LRP11 proteins were linked to an elevated risk of allergic diseases, with corresponding odds ratios of 1.04 (1.03-1.05), 1.25 (1.18-1.34), 1.48 (1.25-1.75), 1.14 (1.11-1.18), 1.09 (1.05-1.12), 1.96 (1.56-2.47), and 1.05 (1.03-1.07), respectively. Bayesian colocalization analysis suggested that LAYN (coloc.abf-PPH4 = 0.819) and TNFAIP3 (coloc.abf-PPH4 = 0.930) share the same variant associated with allergic diseases.

CONCLUSIONS

Our study demonstrates a causal association between the expression levels of TNFAIP3 and LAYN and the risk of allergic diseases, suggesting them as potential drug targets for these conditions, warranting further clinical investigation.

A mediation analysis framework based on variance component to remove genetic confounding effect

Identification of pleiotropy at the single nucleotide polymorphism (SNP) level provides valuable insights into shared genetic signals among phenotypes. One approach to study these signals is through mediation analysis, which dissects the total effect of a SNP on the outcome into a direct effect and an indirect effect through a mediator. However, estimated effects from mediation analysis can be confounded by the genetic correlation between phenotypes, leading to inaccurate results. To address this confounding effect in the context of genetic mediation analysis, we propose a restricted-maximum-likelihood (REML)-based mediation analysis framework called REML-mediation, which can be applied to either individual-level or summary statistics data. Simulations demonstrated that REML-mediation provides unbiased estimates of the true cross-trait causal effect, assuming certain assumptions, albeit with a slightly inflated standard error compared to traditional linear regression. To validate the effectiveness of REML-mediation, we applied it to UK Biobank data and analyzed several mediator-outcome trait pairs along with their corresponding sets of pleiotropic SNPs. REML-mediation successfully identified and corrected for genetic confounding effects in these trait pairs, with correction magnitudes ranging from 7% to 39%. These findings highlight the presence of genetic confounding effects in cross-trait epidemiological studies and underscore the importance of accounting for them in data analysis.

Distinguishing different psychiatric disorders using DDx-PRS

Despite great progress on methods for case-control polygenic prediction (e.g. schizophrenia vs. control), there remains an unmet need for a method that genetically distinguishes clinically related disorders (e.g. schizophrenia (SCZ) vs. bipolar disorder (BIP) vs. depression (MDD) vs. control); such a method could have important clinical value, especially at disorder onset when differential diagnosis can be challenging. Here, we introduce a method, Differential Diagnosis-Polygenic Risk Score (DDx-PRS), that jointly estimates posterior probabilities of each possible diagnostic category (e.g. SCZ=50%, BIP=25%, MDD=15%, control=10%) by modeling variance/covariance structure across disorders, leveraging case-control polygenic risk scores (PRS) for each disorder (computed using existing methods) and prior clinical probabilities for each diagnostic category. DDx-PRS uses only summary-level training data and does not use tuning data, facilitating implementation in clinical settings. In simulations, DDx-PRS was well-calibrated (whereas a simpler approach that analyzes each disorder marginally was poorly calibrated), and effective in distinguishing each diagnostic category vs. the rest. We then applied DDx-PRS to Psychiatric Genomics Consortium SCZ/BIP/MDD/control data, including summary-level training data from 3 case-control GWAS ( N =41,917-173,140 cases; total N =1,048,683) and held-out test data from different cohorts with equal numbers of each diagnostic category (total N =11,460). DDx-PRS was well-calibrated and well-powered relative to these training sample sizes, attaining AUCs of 0.66 for SCZ vs. rest, 0.64 for BIP vs. rest, 0.59 for MDD vs. rest, and 0.68 for control vs. rest. DDx-PRS produced comparable results to methods that leverage tuning data, confirming that DDx-PRS is an effective method. True diagnosis probabilities in top deciles of predicted diagnosis probabilities were considerably larger than prior baseline probabilities, particularly in projections to larger training sample sizes, implying considerable potential for clinical utility under certain circumstances. In conclusion, DDx-PRS is an effective method for distinguishing clinically related disorders.

Pervasive correlations between causal disease effects of proximal SNPs vary with functional annotations and implicate stabilizing selection

The genetic architecture of human diseases and complex traits has been extensively studied, but little is known about the relationship of causal disease effect sizes between proximal SNPs, which have largely been assumed to be independent. We introduce a new method, LD SNP-pair effect correlation regression (LDSPEC), to estimate the correlation of causal disease effect sizes of derived alleles between proximal SNPs, depending on their allele frequencies, LD, and functional annotations; LDSPEC produced robust estimates in simulations across various genetic architectures. We applied LDSPEC to 70 diseases and complex traits from the UK Biobank (average N=306K), meta-analyzing results across diseases/traits. We detected significantly nonzero effect correlations for proximal SNP pairs (e.g., -0.37±0.09 for low-frequency positive-LD 0-100bp SNP pairs) that decayed with distance (e.g., -0.07±0.01 for low-frequency positive-LD 1-10kb), varied with allele frequency (e.g., -0.15±0.04 for common positive-LD 0-100bp), and varied with LD between SNPs (e.g., +0.12±0.05 for common negative-LD 0-100bp) (because we consider derived alleles, positive-LD and negative-LD SNP pairs may yield very different results). We further determined that SNP pairs with shared functions had stronger effect correlations that spanned longer genomic distances, e.g., -0.37±0.08 for low-frequency positive-LD same-gene promoter SNP pairs (average genomic distance of 47kb (due to alternative splicing)) and -0.32±0.04 for low-frequency positive-LD H3K27ac 0-1kb SNP pairs. Consequently, SNP-heritability estimates were substantially smaller than estimates of the sum of causal effect size variances across all SNPs (ratio of 0.87±0.02 across diseases/traits), particularly for certain functional annotations (e.g., 0.78±0.01 for common Super enhancer SNPs)-even though these quantities are widely assumed to be equal. We recapitulated our findings via forward simulations with an evolutionary model involving stabilizing selection, implicating the action of linkage masking, whereby haplotypes containing linked SNPs with opposite effects on disease have reduced effects on fitness and escape negative selection.

Chargaff's second parity rule lies at the origin of additive genetic interactions in quantitative traits to make omnigenic selection possible

Background

Francis Crick's central dogma provides a residue-by-residue mechanistic explanation of the flow of genetic information in living systems. However, this principle may not be sufficient for explaining how random mutations cause continuous variation of quantitative highly polygenic complex traits. Chargaff's second parity rule (CSPR), also referred to as intrastrand DNA symmetry, defined as near-exact equalities G ≈ C and A ≈ T within a single DNA strand, is a statistical property of cellular genomes. The phenomenon of intrastrand DNA symmetry was discovered more than 50 years ago; at present, it remains unclear what its biological role is, what the mechanisms are that force cellular genomes to comply strictly with CSPR, and why genomes of certain noncellular organisms have broken intrastrand DNA symmetry. The present work is aimed at studying a possible link between intrastrand DNA symmetry and the origin of genetic interactions in quantitative traits.

Methods

Computational analysis of single-nucleotide polymorphisms in human and mouse populations and of nucleotide composition biases at different codon positions in bacterial and human proteomes.

Results

The analysis of mutation spectra inferred from single-nucleotide polymorphisms observed in murine and human populations revealed near-exact equalities of numbers of reverse complementary mutations, indicating that random genetic variations obey CSPR. Furthermore, nucleotide compositions of coding sequences proved to be statistically interwoven via CSPR because pyrimidine bias at the 3rd codon position compensates purine bias at the 1st and 2nd positions.

Conclusions

According to Fisher's infinitesimal model, we propose that accumulation of reverse complementary mutations results in a continuous phenotypic variation due to small additive effects of statistically interwoven genetic variations. Therefore, additive genetic interactions can be inferred as a statistical entanglement of nucleotide compositions of separate genetic loci. CSPR challenges the neutral theory of molecular evolution-because all random mutations participate in variation of a trait-and provides an alternative solution to Haldane's dilemma by making a gene function diffuse. We propose that CSPR is symmetry of Fisher's infinitesimal model and that genetic information can be transferred in an implicit contactless manner.

The correlates of neonatal complement component 3 and 4 protein concentrations with a focus on psychiatric and autoimmune disorders

Complement components have been linked to schizophrenia and autoimmune disorders. We examined the association between neonatal circulating C3 and C4 protein concentrations in 68,768 neonates and the risk of six mental disorders. We completed genome-wide association studies (GWASs) for C3 and C4 and applied the summary statistics in Mendelian randomization and phenome-wide association studies related to mental and autoimmune disorders. The GWASs for C3 and C4 protein concentrations identified 15 and 36 independent loci, respectively. We found no associations between neonatal C3 and C4 concentrations and mental disorders in the total sample (both sexes combined); however, post-hoc analyses found that a higher C3 concentration was associated with a reduced risk of schizophrenia in females. Mendelian randomization based on C4 summary statistics found an altered risk of five types of autoimmune disorders. Our study adds to our understanding of the associations between C3 and C4 concentrations and subsequent mental and autoimmune disorders.

Polygenic prediction of human longevity on the supposition of pervasive pleiotropy

The highly polygenic nature of human longevity renders cross-trait pleiotropy an indispensable feature of its genetic architecture. Leveraging the genetic correlation between the aging-related traits (ARTs), we sought to model the additive variance in lifespan as a function of cumulative liability from pleiotropic segregating variants. We tracked allele frequency changes as a function of viability across different age bins and prioritized 34 variants with an immediate implication on lipid metabolism, body mass index (BMI), and cognitive performance, among other traits, revealed by PheWAS analysis in the UK Biobank. Given the highly complex and non-linear interactions between the genetic determinants of longevity, we reasoned that a composite polygenic score would approximate a substantial portion of the variance in lifespan and developed the integrated longevity genetic scores (iLGSs) for distinguishing exceptional survival. We showed that coefficients derived from our ensemble model could potentially reveal an interesting pattern of genomic pleiotropy specific to lifespan. We assessed the predictive performance of our model for distinguishing the enrichment of exceptional longevity among long-lived individuals in two replication cohorts and showed that the median lifespan in the highest decile of our composite prognostic index is up to 4.8 years longer. Finally, using the proteomic correlates of i L G S , we identified protein markers associated with exceptional longevity irrespective of chronological age and prioritized drugs with repurposing potentials for gerotherapeutics. Together, our approach demonstrates a promising framework for polygenic modeling of additive liability conferred by ARTs in defining exceptional longevity and assisting the identification of individuals at higher risk of mortality for targeted lifestyle modifications earlier in life. Furthermore, the proteomic signature associated with i L G S highlights the functional pathway upstream of the PI3K-Akt that can be effectively targeted to slow down aging and extend lifespan.

Inferring disease architecture and predictive ability with LDpred2-auto

LDpred2 is a widely used Bayesian method for building polygenic scores (PGSs). LDpred2-auto can infer the two parameters from the LDpred model, the SNP heritability h2 and polygenicity p, so that it does not require an additional validation dataset to choose best-performing parameters. The main aim of this paper is to properly validate the use of LDpred2-auto for inferring multiple genetic parameters. Here, we present a new version of LDpred2-auto that adds an optional third parameter α to its model, for modeling negative selection. We then validate the inference of these three parameters (or two, when using the previous model). We also show that LDpred2-auto provides per-variant probabilities of being causal that are well calibrated and can therefore be used for fine-mapping purposes. We also introduce a formula to infer the out-of-sample predictive performance r2 of the resulting PGS directly from the Gibbs sampler of LDpred2-auto. Finally, we extend the set of HapMap3 variants recommended to use with LDpred2 with 37% more variants to improve the coverage of this set, and we show that this new set of variants captures 12% more heritability and provides 6% more predictive performance, on average, in UK Biobank analyses.

Pervasive correlations between causal disease effects of proximal SNPs vary with functional annotations and implicate stabilizing selection

The genetic architecture of human diseases and complex traits has been extensively studied, but little is known about the relationship of causal disease effect sizes between proximal SNPs, which have largely been assumed to be independent. We introduce a new method, LD SNP-pair effect correlation regression (LDSPEC), to estimate the correlation of causal disease effect sizes of derived alleles between proximal SNPs, depending on their allele frequencies, LD, and functional annotations; LDSPEC produced robust estimates in simulations across various genetic architectures. We applied LDSPEC to 70 diseases and complex traits from the UK Biobank (average N=306K), meta-analyzing results across diseases/traits. We detected significantly nonzero effect correlations for proximal SNP pairs (e.g., -0.37±0.09 for low-frequency positive-LD 0-100bp SNP pairs) that decayed with distance (e.g., -0.07±0.01 for low-frequency positive-LD 1-10kb), varied with allele frequency (e.g., -0.15±0.04 for common positive-LD 0-100bp), and varied with LD between SNPs (e.g., +0.12±0.05 for common negative-LD 0-100bp) (because we consider derived alleles, positive-LD and negative-LD SNP pairs may yield very different results). We further determined that SNP pairs with shared functions had stronger effect correlations that spanned longer genomic distances, e.g., -0.37±0.08 for low-frequency positive-LD same-gene promoter SNP pairs (average genomic distance of 47kb (due to alternative splicing)) and -0.32±0.04 for low-frequency positive-LD H3K27ac 0-1kb SNP pairs. Consequently, SNP-heritability estimates were substantially smaller than estimates of the sum of causal effect size variances across all SNPs (ratio of 0.87±0.02 across diseases/traits), particularly for certain functional annotations (e.g., 0.78±0.01 for common Super enhancer SNPs)-even though these quantities are widely assumed to be equal. We recapitulated our findings via forward simulations with an evolutionary model involving stabilizing selection, implicating the action of linkage masking, whereby haplotypes containing linked SNPs with opposite effects on disease have reduced effects on fitness and escape negative selection.

Accurate and efficient estimation of local heritability using summary statistics and the linkage disequilibrium matrix

Existing SNP-heritability estimators that leverage summary statistics from genome-wide association studies (GWAS) are much less efficient (i.e., have larger standard errors) than the restricted maximum likelihood (REML) estimators which require access to individual-level data. We introduce a new method for local heritability estimation-Heritability Estimation with high Efficiency using LD and association Summary Statistics (HEELS)-that significantly improves the statistical efficiency of summary-statistics-based heritability estimator and attains comparable statistical efficiency as REML (with a relative statistical efficiency >92%). Moreover, we propose representing the empirical LD matrix as the sum of a low-rank matrix and a banded matrix. We show that this way of modeling the LD can not only reduce the storage and memory cost, but also improve the computational efficiency of heritability estimation. We demonstrate the statistical efficiency of HEELS and the advantages of our proposed LD approximation strategies both in simulations and through empirical analyses of the UK Biobank data.

Genome-Wide Assessment of Shared Genetic Architecture Between Rheumatoid Arthritis and Cardiovascular Diseases

Background Patients with rheumatoid arthritis (RA) have a 2- to 10-fold increased risk of cardiovascular disease (CVD), but the biological mechanisms and existence of causality underlying such associations remain to be investigated. We aimed to investigate the genetic associations and underlying mechanisms between RA and CVD by leveraging large-scale genomic data and genetic cross-trait analytic approaches. Methods and Results Within UK Biobank data, we examined the genetic correlation, shared genetics, and potential causality between RA (Ncases=6754, Ncontrols=452 384) and cardiovascular diseases (CVD, Ncases=44 238, Ncontrols=414 900) using linkage disequilibrium score regression, cross-trait meta-analysis, and Mendelian randomization. We observed significant genetic correlations of RA with myocardial infarction (rg:0.40 [95% CI, 0.24-0.56), angina (rg:0.42 [95% CI, 0.28-0.56]), coronary heart diseases (rg:0.41 [95% CI, 0.27-0.55]), and CVD (rg:0.43 [95% CI, 0.29-0.57]) after correcting for multiple testing (P<0.05/5). When stratified by frequent use of analgesics, we found increased genetic correlation between RA and CVD among participants without aspirin usage (rg:0.54 [95% CI, 0.30-0.78] for angina; Pvalue=6.69×10-6) and among participants with paracetamol usage (rg:0.75 [95% CI, 0.20-1.29] for myocardial infarction; Pvalue=8.90×10-3), whereas others remained similar. Cross-trait meta-analysis identified 9 independent shared loci between RA and CVD, including PTPN22 at chr1p13.2, BCL2L11 at chr2q13, and CCR3 at chr3p21.31 (Psingle trait<1×10-3 and Pmeta<5×10-8), highlighting potential shared pathogenesis including accelerating atherosclerosis, upregulating oxidative stress, and vascular permeability. Finally, Mendelian randomization estimates showed limited evidence of causality between RA and CVD. Conclusions Our results supported shared genetic pathogenesis rather than causality in explaining the observed association between RA and CVD. The identified shared genetic factors provided insights into potential novel therapeutic target for RA-CVD comorbidities.

Speos: an ensemble graph representation learning framework to predict core gene candidates for complex diseases

Understanding phenotype-to-genotype relationships is a grand challenge of 21st century biology with translational implications. The recently proposed "omnigenic" model postulates that effects of genetic variation on traits are mediated by core-genes and -proteins whose activities mechanistically influence the phenotype, whereas peripheral genes encode a regulatory network that indirectly affects phenotypes via core gene products. Here, we develop a positive-unlabeled graph representation-learning ensemble-approach based on a nested cross-validation to predict core-like genes for diverse diseases using Mendelian disorder genes for training. Employing mouse knockout phenotypes for external validations, we demonstrate that core-like genes display several key properties of core genes: Mouse knockouts of genes corresponding to our most confident predictions give rise to relevant mouse phenotypes at rates on par with the Mendelian disorder genes, and all candidates exhibit core gene properties like transcriptional deregulation in disease and loss-of-function intolerance. Moreover, as predicted for core genes, our candidates are enriched for drug targets and druggable proteins. In contrast to Mendelian disorder genes the new core-like genes are enriched for druggable yet untargeted gene products, which are therefore attractive targets for drug development. Interpretation of the underlying deep learning model suggests plausible explanations for our core gene predictions in form of molecular mechanisms and physical interactions. Our results demonstrate the potential of graph representation learning for the interpretation of biological complexity and pave the way for studying core gene properties and future drug development.

Estimating heritability explained by local ancestry and evaluating stratification bias in admixture mapping from summary statistics

The heritability explained by local ancestry markers in an admixed population (hγ2) provides crucial insight into the genetic architecture of a complex disease or trait. Estimation of hγ2 can be susceptible to biases due to population structure in ancestral populations. Here, we present heritability estimation from admixture mapping summary statistics (HAMSTA), an approach that uses summary statistics from admixture mapping to infer heritability explained by local ancestry while adjusting for biases due to ancestral stratification. Through extensive simulations, we demonstrate that HAMSTA hγ2 estimates are approximately unbiased and are robust to ancestral stratification compared to existing approaches. In the presence of ancestral stratification, we show a HAMSTA-derived sampling scheme provides a calibrated family-wise error rate (FWER) of ∼5% for admixture mapping, unlike existing FWER estimation approaches. We apply HAMSTA to 20 quantitative phenotypes of up to 15,988 self-reported African American individuals in the Population Architecture using Genomics and Epidemiology (PAGE) study. We observe hˆγ2 in the 20 phenotypes range from 0.0025 to 0.033 (mean hˆγ2 = 0.012 ± 9.2 × 10-4), which translates to hˆ2 ranging from 0.062 to 0.85 (mean hˆ2 = 0.30 ± 0.023). Across these phenotypes we find little evidence of inflation due to ancestral population stratification in current admixture mapping studies (mean inflation factor of 0.99 ± 0.001). Overall, HAMSTA provides a fast and powerful approach to estimate genome-wide heritability and evaluate biases in test statistics of admixture mapping studies.

Pragmatic Approach to Applying Polygenic Risk Scores to Diverse Populations

Polygenic risk scores (PRS) estimate genetic susceptibility of an individual to disease and have the potential of providing utility in multiple clinical contexts. However, their performance, computation, and reporting in diverse populations remain challenging. Here, we present a pragmatic approach to optimize a PRS for a population of interest that leverages publicly available data and methods and consists of seven steps that are easily implemented without the requirement of expertise in complex genetics: step 1, selecting source genome-wide association studies (GWAS) and imputation; step 2, selecting methods to compute polygenic score; step 3, adjusting scores using principal components of genetic ancestry; step 4, selecting the best performing score; step 5, defining percentiles of a population distribution; step 6, validating performance of the optimized polygenic score; and step 7, implementing the optimized polygenic score in clinical practice. © 2023 Wiley Periodicals LLC.

Genetic influences on alcohol flushing in East Asian populations

BACKGROUND

Although it is known that variation in the aldehyde dehydrogenase 2 (ALDH2) gene family influences the East Asian alcohol flushing response, knowledge about other genetic variants that affect flushing symptoms is limited.

METHODS

We performed a genome-wide association study meta-analysis and heritability analysis of alcohol flushing in 15,105 males of East Asian ancestry (Koreans and Chinese) to identify genetic associations with alcohol flushing. We also evaluated whether self-reported flushing can be used as an instrumental variable for alcohol intake.

RESULTS

We identified variants in the region of ALDH2 strongly associated with alcohol flushing, replicating previous studies conducted in East Asian populations. Additionally, we identified variants in the alcohol dehydrogenase 1B (ADH1B) gene region associated with alcohol flushing. Several novel variants were identified after adjustment for the lead variants (ALDH2-rs671 and ADH1B-rs1229984), which need to be confirmed in larger studies. The estimated SNP-heritability on the liability scale was 13% (S.E. = 4%) for flushing, but the heritability estimate decreased to 6% (S.E. = 4%) when the effects of the lead variants were controlled for. Genetic instrumentation of higher alcohol intake using these variants recapitulated known associations of alcohol intake with hypertension. Using self-reported alcohol flushing as an instrument gave a similar association pattern of higher alcohol intake and cardiovascular disease-related traits (e.g. stroke).

CONCLUSION

This study confirms that ALDH2-rs671 and ADH1B-rs1229984 are associated with alcohol flushing in East Asian populations. Our findings also suggest that self-reported alcohol flushing can be used as an instrumental variable in future studies of alcohol consumption.

Bayesian multivariate genetic analysis improves translational insights

While lipid traits are known essential mediators of cardiovascular disease, few approaches have taken advantage of their shared genetic effects. We apply a Bayesian multivariate size estimator, mash, to GWAS of four lipid traits in the Million Veterans Program (MVP) and provide posterior mean and local false sign rates for all effects. These estimates borrow information across traits to improve effect size accuracy. We show that controlling local false sign rates accurately and powerfully identifies replicable genetic associations and that multivariate control furthers the ability to explain complex diseases. Our application yields high concordance between independent datasets, more accurately prioritizes causal genes, and significantly improves polygenic prediction beyond state-of-the-art methods by up to 59% for lipid traits. The use of Bayesian multivariate genetic shrinkage has yet to be applied to human quantitative trait GWAS results, and we present a staged approach to prediction on a polygenic scale.

Genome-wide polygenic risk scores for hypertensive disease during pregnancy can also predict the risk for long-term cardiovascular disease

BACKGROUND

Hypertensive disorders during pregnancy are associated with the risk of long-term cardiovascular disease after pregnancy, but it has not yet been determined whether genetic predisposition for hypertensive disorders during pregnancy can predict the risk for long-term cardiovascular disease.

OBJECTIVE

This study aimed to evaluate the risk for long-term atherosclerotic cardiovascular disease according to polygenic risk scores for hypertensive disorders during pregnancy.

STUDY DESIGN

Among UK Biobank participants, we included European-descent women (n=164,575) with at least 1 live birth. Participants were divided according to genetic risk categorized by polygenic risk scores for hypertensive disorders during pregnancy (low risk, score ≤25th percentile; medium risk, score 25th∼75th percentile; high risk, score >75th percentile), and were evaluated for incident atherosclerotic cardiovascular disease, defined as the new occurrence of one of the following: coronary artery disease, myocardial infarction, ischemic stroke, or peripheral artery disease.

RESULTS

Among the study population, 2427 (1.5%) had a history of hypertensive disorders during pregnancy, and 8942 (5.6%) developed incident atherosclerotic cardiovascular disease after enrollment. Women with high genetic risk for hypertensive disorders during pregnancy had a higher prevalence of hypertension at enrollment. After enrollment, women with high genetic risk for hypertensive disorders during pregnancy had an increased risk for incident atherosclerotic cardiovascular disease, including coronary artery disease, myocardial infarction, and peripheral artery disease, compared with those with low genetic risk, even after adjustment for history of hypertensive disorders during pregnancy.

CONCLUSION

High genetic risk for hypertensive disorders during pregnancy was associated with increased risk for atherosclerotic cardiovascular disease. This study provides evidence on the informative value of polygenic risk scores for hypertensive disorders during pregnancy in prediction of long-term cardiovascular outcomes later in life.

A mixed-model approach for estimating drivers of microbiota community composition and differential taxonomic abundance

Next-generation sequencing (NGS) and metabarcoding approaches are increasingly applied to wild animal populations, but there is a disconnect between the widely applied generalized linear mixed model (GLMM) approaches commonly used to study phenotypic variation and the statistical toolkit from community ecology typically applied to metabarcoding data. Here, we describe the suitability of a novel GLMM-based approach for analyzing the taxon-specific sequence read counts derived from standard metabarcoding data. This approach allows decomposition of the contribution of different drivers to variation in community composition (e.g., age, season, individual) via interaction terms in the model random-effects structure. We provide guidance to implementing this approach and show how these models can identify how responsible specific taxonomic groups are for the effects attributed to different drivers. We applied this approach to two cross-sectional data sets from the Soay sheep population of St. Kilda. GLMMs showed agreement with dissimilarity-based approaches highlighting the substantial contribution of age and minimal contribution of season to microbiota community compositions, and simultaneously estimated the contribution of other technical and biological factors. We further used model predictions to show that age effects were principally due to increases in taxa of the phylum Bacteroidetes and declines in taxa of the phylum Firmicutes. This approach offers a powerful means for understanding the influence of drivers of community structure derived from metabarcoding data. We discuss how our approach could be readily adapted to allow researchers to estimate contributions of additional factors such as host or microbe phylogeny to answer emerging questions surrounding the ecological and evolutionary roles of within-host communities. IMPORTANCE NGS and fecal metabarcoding methods have provided powerful opportunities to study the wild gut microbiome. A wealth of data is, therefore, amassing across wild systems, generating the need for analytical approaches that can appropriately investigate simultaneous factors at the host and environmental scale that determine the composition of these communities. Here, we describe a generalized linear mixed-effects model (GLMM) approach to analyze read count data from metabarcoding of the gut microbiota, allowing us to quantify the contributions of multiple host and environmental factors to within-host community structure. Our approach provides outputs that are familiar to a majority of field ecologists and can be run using any standard mixed-effects modeling packages. We illustrate this approach using two metabarcoding data sets from the Soay sheep population of St. Kilda investigating age and season effects as worked examples.

Genetic Susceptibility to Atrial Fibrillation Identified via Deep Learning of 12-Lead Electrocardiograms

BACKGROUND

Artificial intelligence (AI) models applied to 12-lead ECG waveforms can predict atrial fibrillation (AF), a heritable and morbid arrhythmia. However, the factors forming the basis of risk predictions from AI models are usually not well understood. We hypothesized that there might be a genetic basis for an AI algorithm for predicting the 5-year risk of new-onset AF using 12-lead ECGs (ECG-AI)-based risk estimates.

METHODS

We applied a validated ECG-AI model for predicting incident AF to ECGs from 39 986 UK Biobank participants without AF. We then performed a genome-wide association study (GWAS) of the predicted AF risk and compared it with an AF GWAS and a GWAS of risk estimates from a clinical variable model.

RESULTS

In the ECG-AI GWAS, we identified 3 signals (P<5×10-8) at established AF susceptibility loci marked by the sarcomeric gene TTN and sodium channel genes SCN5A and SCN10A. We also identified 2 novel loci near the genes VGLL2 and EXT1. In contrast, the clinical variable model prediction GWAS indicated a different genetic profile. In genetic correlation analysis, the prediction from the ECG-AI model was estimated to have a higher correlation with AF than that from the clinical variable model.

CONCLUSIONS

Predicted AF risk from an ECG-AI model is influenced by genetic variation implicating sarcomeric, ion channel and body height pathways. ECG-AI models may identify individuals at risk for disease via specific biological pathways.

Intermediate Molecular Phenotypes to Identify Genetic Markers of Anthracycline-Induced Cardiotoxicity Risk

Cardiotoxicity due to anthracyclines (CDA) affects cancer patients, but we cannot predict who may suffer from this complication. CDA is a complex trait with a polygenic component that is mainly unidentified. We propose that levels of intermediate molecular phenotypes (IMPs) in the myocardium associated with histopathological damage could explain CDA susceptibility, so variants of genes encoding these IMPs could identify patients susceptible to this complication. Thus, a genetically heterogeneous cohort of mice (n = 165) generated by backcrossing were treated with doxorubicin and docetaxel. We quantified heart fibrosis using an Ariol slide scanner and intramyocardial levels of IMPs using multiplex bead arrays and QPCR. We identified quantitative trait loci linked to IMPs (ipQTLs) and cdaQTLs via linkage analysis. In three cancer patient cohorts, CDA was quantified using echocardiography or Cardiac Magnetic Resonance. CDA behaves as a complex trait in the mouse cohort. IMP levels in the myocardium were associated with CDA. ipQTLs integrated into genetic models with cdaQTLs account for more CDA phenotypic variation than that explained by cda-QTLs alone. Allelic forms of genes encoding IMPs associated with CDA in mice, including AKT1, MAPK14, MAPK8, STAT3, CAS3, and TP53, are genetic determinants of CDA in patients. Two genetic risk scores for pediatric patients (n = 71) and women with breast cancer (n = 420) were generated using machine-learning Least Absolute Shrinkage and Selection Operator (LASSO) regression. Thus, IMPs associated with heart damage identify genetic markers of CDA risk, thereby allowing more personalized patient management.

Characterizing the polygenic architecture of complex traits in populations of East Asian and European descent

To investigate the polygenicity of complex traits in populations of East Asian (EAS) and European (EUR) descents, we leveraged genome-wide data from Biobank Japan, UK Biobank, and FinnGen cohorts. Specifically, we analyzed up to 215 outcomes related to 18 health domains, assessing their polygenic architecture via descriptive statistics, such as the proportion of susceptibility SNPs per trait (πc). While we did not observe EAS-EUR differences in the overall distribution of polygenicity parameters across the phenotypes investigated, there were ancestry-specific patterns in the polygenicity differences between health domains. In EAS, pairwise comparisons across health domains showed enrichment for πc differences related to hematological and metabolic traits (hematological fold-enrichment = 4.45, p = 2.15 × 10-7; metabolic fold-enrichment = 4.05, p = 4.01 × 10-6). For both categories, the proportion of susceptibility SNPs was lower than that observed for several other health domains (EAS-hematological median πc = 0.15%, EAS-metabolic median πc = 0.18%) with the strongest πc difference with respect to respiratory traits (EAS-respiratory median πc = 0.50%; hematological-p = 2.26 × 10-3; metabolic-p = 3.48 × 10-3). In EUR, pairwise comparisons showed multiple πc differences related to the endocrine category (fold-enrichment = 5.83, p = 4.76 × 10-6), where these traits showed a low proportion of susceptibility SNPs (EUR-endocrine median πc = 0.01%) with the strongest difference with respect to psychiatric phenotypes (EUR-psychiatric median πc = 0.50%; p = 1.19 × 10-4). Simulating sample sizes of 1,000,000 and 5,000,000 individuals, we also showed that ancestry-specific polygenicity patterns translate into differences across health domains in the genetic variance explained by susceptibility SNPs projected to be genome-wide significant (e.g., EAS hematological-neoplasm p = 2.18 × 10-4; EUR endocrine-gastrointestinal p = 6.80 × 10-4). These findings highlight that traits related to the same health domains may present ancestry-specific variability in their polygenicity.

Shared genetic architecture of blood eosinophil counts and asthma in UK Biobank

Rationale

Asthma is a complex, heterogeneous disease strongly associated with type 2 inflammation, and blood eosinophil counts guide therapeutic interventions in moderate and severe asthma. Eosinophils are leukocytes involved in type 2 immune responses. Despite these critical associations between asthma and blood eosinophil counts, the shared genetic architecture of these two traits remains unknown. The objective of the present study was to characterise the genetic architecture of blood eosinophil counts and asthma in the UK Biobank.

Methods

We performed genome-wide association studies (GWAS) of doctor-diagnosed asthma, blood eosinophil, neutrophil, lymphocyte and monocyte counts in the UK Biobank. Genetic correlation analysis was performed on GWAS results and validated in the Trans-National Asthma Genetic Consortium (TAGC) study of asthma.

Results

GWAS of doctor-diagnosed asthma and blood eosinophil counts in the UK Biobank identified 585 and 3429 significant variants, respectively. STAT6, a transcription factor involved in interleukin-4 signalling, was a key shared pathway between asthma and blood eosinophil counts. Genetic correlation analysis demonstrated a positive correlation between doctor-diagnosed asthma and blood eosinophil counts (r=0.38±0.10, correlation±se; p=4.7×10-11). As a validation of this association, we found a similar correlation between TAGC and blood eosinophil counts in the UK Biobank (0.37±0.08, correlation±se; p=1.2×10-6).

Conclusions

These findings define the shared genetic architecture between blood eosinophil counts and asthma risk in subjects of European ancestry and point to a genetic link to the STAT6 signalling pathway in these two traits.

Pan-cancer and cross-population genome-wide association studies dissect shared genetic backgrounds underlying carcinogenesis

Integrating genomic data of multiple cancers allows de novo cancer grouping and elucidating the shared genetic basis across cancers. Here, we conduct the pan-cancer and cross-population genome-wide association study (GWAS) meta-analysis and replication studies on 13 cancers including 250,015 East Asians (Biobank Japan) and 377,441 Europeans (UK Biobank). We identify ten cancer risk variants including five pleiotropic associations (e.g., rs2076295 at DSP on 6p24 associated with lung cancer and rs2525548 at TRIM4 on 7q22 nominally associated with six cancers). Quantifying shared heritability among the cancers detects positive genetic correlations between breast and prostate cancer across populations. Common genetic components increase the statistical power, and the large-scale meta-analysis of 277,896 breast/prostate cancer cases and 901,858 controls identifies 91 newly genome-wide significant loci. Enrichment analysis of pathways and cell types reveals shared genetic backgrounds across said cancers. Focusing on genetically correlated cancers can contribute to enhancing our insights into carcinogenesis.

Characterizing the polygenic architecture of complex traits in populations of East Asian and European descent

To investigate the polygenicity of complex traits in populations of East Asian (EAS) and European (EUR) descents, we leveraged genome-wide data from Biobank Japan, UK Biobank, and FinnGen cohorts. Specifically, we analyzed up to 215 outcomes related to 18 health domains, assessing their polygenic architecture via descriptive statistics, such as the proportion of susceptibility SNPs per trait (π _c ). While we did not observe EAS-EUR differences in the overall distribution of polygenicity parameters across the phenotypes investigated, there were ancestry-specific patterns in the polygenicity differences between health domains. In EAS, pairwise comparisons across health domains showed enrichment for π _c differences related to hematological and metabolic traits (hematological fold-enrichment=4.45, p=2.15×10 ^-7 ; metabolic fold-enrichment=4.05, p=4.01×10 ^-6 ). For both categories, the proportion of susceptibility SNPs was lower than that observed for several other health domains (EAS-hematological median π _c =0.15%, EAS-metabolic median π _c =0.18%) with the strongest π _c difference with respect to respiratory traits (EAS-respiratory median π _c =0.50%; Hematological-p=2.26×10 ^-3 ; Metabolic-p=3.48×10 ^-3 ). In EUR, pairwise comparisons showed multiple π _c differences related to the endocrine category (fold-enrichment=5.83, p=4.76×10 ^-6 ), where these traits showed a low proportion of susceptibility SNPs (EUR-endocrine median π _c =0.01%) with the strongest difference with respect to psychiatric phenotypes (EUR-psychiatric median π _c =0.50%; p=1.19×10 ^-4 ). Simulating sample sizes of 1,000,000 and 5,000,000 individuals, we also showed that ancestry-specific polygenicity patterns translate into differences across health domains in the genetic variance explained by susceptibility SNPs projected to be genome-wide significant (e.g., EAS hematological-neoplasm p=2.18×10 ^-4 ; EUR endocrine-gastrointestinal p=6.80×10 ^-4 ). These findings highlight that traits related to the same health domains may present ancestry-specific variability in their polygenicity.

Identification of new risk loci shared across systemic vasculitides points towards potential target genes for drug repurposing

OBJECTIVES

The number of susceptibility loci currently associated with vasculitis is lower than in other immune-mediated diseases due in part to small cohort sizes, a consequence of the low prevalence of vasculitides. This study aimed to identify new genetic risk loci for the main systemic vasculitides through a comprehensive analysis of their genetic overlap.

METHODS

Genome-wide data from 8467 patients with any of the main forms of vasculitis and 29 795 healthy controls were meta-analysed using ASSET. Pleiotropic variants were functionally annotated and linked to their target genes. Prioritised genes were queried in DrugBank to identify potentially repositionable drugs for the treatment of vasculitis.

RESULTS

Sixteen variants were independently associated with two or more vasculitides, 15 of them representing new shared risk loci. Two of these pleiotropic signals, located close to CTLA4 and CPLX1, emerged as novel genetic risk loci in vasculitis. Most of these polymorphisms appeared to affect vasculitis by regulating gene expression. In this regard, for some of these common signals, potential causal genes were prioritised based on functional annotation, including CTLA4, RNF145, IL12B, IL5, IRF1, IFNGR1, PTK2B, TRIM35, EGR2 and ETS2, each of which has key roles in inflammation. In addition, drug repositioning analysis showed that several drugs, including abatacept and ustekinumab, could be potentially repurposed in the management of the analysed vasculitides.

CONCLUSIONS

We identified new shared risk loci with functional impact in vasculitis and pinpointed potential causal genes, some of which could represent promising targets for the treatment of vasculitis.

Biobank-scale inference of ancestral recombination graphs enables genealogical analysis of complex traits

Genome-wide genealogies compactly represent the evolutionary history of a set of genomes and inferring them from genetic data has the potential to facilitate a wide range of analyses. We introduce a method, ARG-Needle, for accurately inferring biobank-scale genealogies from sequencing or genotyping array data, as well as strategies to utilize genealogies to perform association and other complex trait analyses. We use these methods to build genome-wide genealogies using genotyping data for 337,464 UK Biobank individuals and test for association across seven complex traits. Genealogy-based association detects more rare and ultra-rare signals (N = 134, frequency range 0.0007-0.1%) than genotype imputation using ~65,000 sequenced haplotypes (N = 64). In a subset of 138,039 exome sequencing samples, these associations strongly tag (average r = 0.72) underlying sequencing variants enriched (4.8×) for loss-of-function variation. These results demonstrate that inferred genome-wide genealogies may be leveraged in the analysis of complex traits, complementing approaches that require the availability of large, population-specific sequencing panels.

Genetics of myocardial interstitial fibrosis in the human heart and association with disease

Myocardial interstitial fibrosis is associated with cardiovascular disease and adverse prognosis. Here, to investigate the biological pathways that underlie fibrosis in the human heart, we developed a machine learning model to measure native myocardial T1 time, a marker of myocardial fibrosis, in 41,505 UK Biobank participants who underwent cardiac magnetic resonance imaging. Greater T1 time was associated with diabetes mellitus, renal disease, aortic stenosis, cardiomyopathy, heart failure, atrial fibrillation, conduction disease and rheumatoid arthritis. Genome-wide association analysis identified 11 independent loci associated with T1 time. The identified loci implicated genes involved in glucose transport (SLC2A12), iron homeostasis (HFE, TMPRSS6), tissue repair (ADAMTSL1, VEGFC), oxidative stress (SOD2), cardiac hypertrophy (MYH7B) and calcium signaling (CAMK2D). Using a transforming growth factor β1-mediated cardiac fibroblast activation assay, we found that 9 of the 11 loci consisted of genes that exhibited temporal changes in expression or open chromatin conformation supporting their biological relevance to myofibroblast cell state acquisition. By harnessing machine learning to perform large-scale quantification of myocardial interstitial fibrosis using cardiac imaging, we validate associations between cardiac fibrosis and disease, and identify new biologically relevant pathways underlying fibrosis.

Estimating heritability explained by local ancestry and evaluating stratification bias in admixture mapping from summary statistics

The heritability explained by local ancestry markers in an admixed population h γ 2 provides crucial insight into the genetic architecture of a complex disease or trait. Estimation of h γ 2 can be susceptible to biases due to population structure in ancestral populations. Here, we present a novel approach, Heritability estimation from Admixture Mapping Summary STAtistics (HAMSTA), which uses summary statistics from admixture mapping to infer heritability explained by local ancestry while adjusting for biases due to ancestral stratification. Through extensive simulations, we demonstrate that HAMSTA h γ 2 estimates are approximately unbiased and are robust to ancestral stratification compared to existing approaches. In the presence of ancestral stratification, we show a HAMSTA-derived sampling scheme provides a calibrated family-wise error rate (FWER) of ~5% for admixture mapping, unlike existing FWER estimation approaches. We apply HAMSTA to 20 quantitative phenotypes of up to 15,988 self-reported African American individuals in the Population Architecture using Genomics and Epidemiology (PAGE) study. We observe h ˆ γ 2 in the 20 phenotypes range from 0.0025 to 0.033 (mean h ˆ γ 2 = 0.012 + / - 9.2 × 10 - 4 ), which translates to h ˆ 2 ranging from 0.062 to 0.85 (mean h ˆ 2 = 0.30 + / - 0.023 ). Across these phenotypes we find little evidence of inflation due to ancestral population stratification in current admixture mapping studies (mean inflation factor of 0.99 +/- 0.001). Overall, HAMSTA provides a fast and powerful approach to estimate genome-wide heritability and evaluate biases in test statistics of admixture mapping studies.

Causal effects on complex traits are similar for common variants across segments of different continental ancestries within admixed individuals

Individuals of admixed ancestries (for example, African Americans) inherit a mosaic of ancestry segments (local ancestry) originating from multiple continental ancestral populations. This offers the unique opportunity of investigating the similarity of genetic effects on traits across ancestries within the same population. Here we introduce an approach to estimate correlation of causal genetic effects (radmix) across local ancestries and analyze 38 complex traits in African-European admixed individuals (N = 53,001) to observe very high correlations (meta-analysis radmix = 0.95, 95% credible interval 0.93-0.97), much higher than correlation of causal effects across continental ancestries. We replicate our results using regression-based methods from marginal genome-wide association study summary statistics. We also report realistic scenarios where regression-based methods yield inflated heterogeneity-by-ancestry due to ancestry-specific tagging of causal effects, and/or polygenicity. Our results motivate genetic analyses that assume minimal heterogeneity in causal effects by ancestry, with implications for the inclusion of ancestry-diverse individuals in studies.

Gene-environment interaction explains a part of missing heritability in human body mass index

Gene-environment (G×E) interaction could partially explain missing heritability in traits; however, the magnitudes of G×E interaction effects remain unclear. Here, we estimate the heritability of G×E interaction for body mass index (BMI) by subjecting genome-wide interaction study data of 331,282 participants in the UK Biobank to linkage disequilibrium score regression (LDSC) and linkage disequilibrium adjusted kinships-software for estimating SNP heritability from summary statistics (LDAK-SumHer) analyses. Among 14 obesity-related lifestyle factors, MET score, pack years of smoking, and alcohol intake frequency significantly interact with genetic factors in both analyses, accounting for the partial variance of BMI. The G×E interaction heritability (%) and standard error of these factors by LDSC and LDAK-SumHer are as follows: MET score, 0.45% (0.12) and 0.65% (0.24); pack years of smoking, 0.52% (0.13) and 0.93% (0.26); and alcohol intake frequency, 0.32% (0.10) and 0.80% (0.17), respectively. Moreover, these three factors are partially validated for their interactions with genetic factors in other obesity-related traits, including waist circumference, hip circumference, waist-to-hip ratio adjusted with BMI, and body fat percentage. Our results suggest that G×E interaction may partly explain the missing heritability in BMI, and two G×E interaction loci identified could help in understanding the genetic architecture of obesity.

Accurate and Efficient Estimation of Local Heritability using Summary Statistics and LD Matrix

Existing SNP-heritability estimation methods that leverage GWAS summary statistics produce estimators that are less efficient than the restricted maximum likelihood (REML) estimator using individual-level data under linear mixed models (LMMs). Increasing the precision of a heritability estimator is particularly important for regional analyses, as local genetic variances tend to be small. We introduce a new estimator for local heritability, "HEELS", which attains comparable statistical efficiency as REML (\emph{i.e.} relative efficiency greater than 92%) but only requires summary-level statistics -- Z-scores from the marginal association tests plus the empirical LD matrix. HEELS significantly improves the statistical efficiency of the existing summary-statistics-based heritability estimators-- for instance, HEELS produces heritability estimates that are more than 3-fold and 7-times less variable than GRE and LDSC, respectively. Moreover, we introduce a unified framework to evaluate and compare the performance of different LD approximation strategies. We propose representing the empirical LD as the sum of a low-rank matrix and a banded matrix. This approximation not only reduces the storage and memory cost of using the LD matrix, but also improves the computational efficiency of the HEELS estimation. We demonstrate the statistical efficiency of HEELS and the advantages of our proposed LD approximation strategies both in simulations and through empirical analyses of the UK Biobank data.

Genome-wide meta-analysis identifies 93 risk loci and enables risk prediction equivalent to monogenic forms of venous thromboembolism

We report a genome-wide association study of venous thromboembolism (VTE) incorporating 81,190 cases and 1,419,671 controls sampled from six cohorts. We identify 93 risk loci, of which 62 are previously unreported. Many of the identified risk loci are at genes encoding proteins with functions converging on the coagulation cascade or platelet function. A VTE polygenic risk score (PRS) enabled effective identification of both high- and low-risk individuals. Individuals within the top 0.1% of PRS distribution had a VTE risk similar to homozygous or compound heterozygous carriers of the variants G20210A (c.*97 G > A) in F2 and p.R534Q in F5. We also document that F2 and F5 mutation carriers in the bottom 10% of the PRS distribution had a risk similar to that of the general population. We further show that PRS improved individual risk prediction beyond that of genetic and clinical risk factors. We investigated the extent to which venous and arterial thrombosis share clinical risk factors using Mendelian randomization, finding that some risk factors for arterial thrombosis were directionally concordant with VTE risk (for example, body mass index and smoking) whereas others were discordant (for example, systolic blood pressure and triglyceride levels).

Genome-Wide Analysis of Left Ventricular Maximum Wall Thickness in the UK Biobank Cohort Reveals a Shared Genetic Background With Hypertrophic Cardiomyopathy

BACKGROUND

Left ventricular maximum wall thickness (LVMWT) is an important biomarker of left ventricular hypertrophy and provides diagnostic and prognostic information in hypertrophic cardiomyopathy (HCM). Limited information is available on the genetic determinants of LVMWT.

METHODS

We performed a genome-wide association study of LVMWT measured from the cardiovascular magnetic resonance examinations of 42 176 European individuals. We evaluated the genetic relationship between LVMWT and HCM by performing pairwise analysis using the data from the Hypertrophic Cardiomyopathy Registry in which the controls were randomly selected from UK Biobank individuals not included in the cardiovascular magnetic resonance sub-study.

RESULTS

Twenty-one genetic loci were discovered at P<5×10-8. Several novel candidate genes were identified including PROX1, PXN, and PTK2, with known functional roles in myocardial growth and sarcomere organization. The LVMWT genetic risk score is predictive of HCM in the Hypertrophic Cardiomyopathy Registry (odds ratio per SD: 1.18 [95% CI, 1.13-1.23]) with pairwise analyses demonstrating a moderate genetic correlation (rg=0.53) and substantial loci overlap (19/21).

CONCLUSIONS

Our findings provide novel insights into the genetic underpinning of LVMWT and highlight its shared genetic background with HCM, supporting future endeavours to elucidate the genetic etiology of HCM.

Machine Learning to Understand Genetic and Clinical Factors Associated With the Pulse Waveform Dicrotic Notch

BACKGROUND

Absence of a dicrotic notch on finger photoplethysmography is an easily ascertainable and inexpensive trait that has been associated with age and prevalent cardiovascular disease. However, the trait exists along a continuum, and little is known about its genetic underpinnings or prognostic value for incident cardiovascular disease.

METHODS

In 169 787 participants in the UK Biobank, we identified absent dicrotic notch on photoplethysmography and created a novel continuous trait reflecting notch smoothness using machine learning. Next, we determined the heritability, genetic basis, polygenic risk, and clinical relations for the binary absent notch trait and the newly derived continuous notch smoothness trait.

RESULTS

Heritability of the continuous notch smoothness trait was 7.5%, compared with 5.6% for the binary absent notch trait. A genome-wide association study of notch smoothness identified 15 significant loci, implicating genes including NT5C2 (P=1.2×10-26), IGFBP3 (P=4.8×10-18), and PHACTR1 (P=1.4×10-13), compared with 6 loci for the binary absent notch trait. Notch smoothness stratified risk of incident myocardial infarction or coronary artery disease, stroke, heart failure, and aortic stenosis. A polygenic risk score for notch smoothness was associated with incident cardiovascular disease and all-cause death in UK Biobank participants without available photoplethysmography data.

CONCLUSIONS

We found that a machine learning derived continuous trait reflecting dicrotic notch smoothness on photoplethysmography was heritable and associated with genes involved in vascular stiffness. Greater notch smoothness was associated with greater risk of incident cardiovascular disease. Raw digital phenotyping may identify individuals at risk for disease via specific genetic pathways.

Polygenic architecture of rare coding variation across 394,783 exomes

Both common and rare genetic variants influence complex traits and common diseases. Genome-wide association studies have identified thousands of common-variant associations, and more recently, large-scale exome sequencing studies have identified rare-variant associations in hundreds of genes1-3. However, rare-variant genetic architecture is not well characterized, and the relationship between common-variant and rare-variant architecture is unclear4. Here we quantify the heritability explained by the gene-wise burden of rare coding variants across 22 common traits and diseases in 394,783 UK Biobank exomes5. Rare coding variants (allele frequency < 1 × 10-3) explain 1.3% (s.e. = 0.03%) of phenotypic variance on average-much less than common variants-and most burden heritability is explained by ultrarare loss-of-function variants (allele frequency < 1 × 10-5). Common and rare variants implicate the same cell types, with similar enrichments, and they have pleiotropic effects on the same pairs of traits, with similar genetic correlations. They partially colocalize at individual genes and loci, but not to the same extent: burden heritability is strongly concentrated in significant genes, while common-variant heritability is more polygenic, and burden heritability is also more strongly concentrated in constrained genes. Finally, we find that burden heritability for schizophrenia and bipolar disorder6,7 is approximately 2%. Our results indicate that rare coding variants will implicate a tractable number of large-effect genes, that common and rare associations are mechanistically convergent, and that rare coding variants will contribute only modestly to missing heritability and population risk stratification.

Population analyses of mosaic X chromosome loss identify genetic drivers and widespread signatures of cellular selection

Mosaic loss of the X chromosome (mLOX) is the most commonly occurring clonal somatic alteration detected in the leukocytes of women, yet little is known about its genetic determinants or phenotypic consequences. To address this, we estimated mLOX in >900,000 women across eight biobanks, identifying 10% of women with detectable X loss in approximately 2% of their leukocytes. Out of 1,253 diseases examined, women with mLOX had an elevated risk of myeloid and lymphoid leukemias and pneumonia. Genetic analyses identified 49 common variants influencing mLOX, implicating genes with established roles in chromosomal missegregation, cancer predisposition, and autoimmune diseases. Complementary exome-sequence analyses identified rare missense variants in FBXO10 which confer a two-fold increased risk of mLOX. A small fraction of these associations were shared with mosaic Y chromosome loss in men, suggesting different biological processes drive the formation and clonal expansion of sex chromosome missegregation events. Allelic shift analyses identified alleles on the X chromosome which are preferentially retained, demonstrating that variation at many loci across the X chromosome is under cellular selection. A novel polygenic score including 44 independent X chromosome allelic shift loci correctly inferred the retained X chromosomes in 80.7% of mLOX cases in the top decile. Collectively our results support a model where germline variants predispose women to acquiring mLOX, with the allelic content of the X chromosome possibly shaping the magnitude of subsequent clonal expansion.

Polygenic adaptation dynamics in large, finite populations

Although many phenotypic traits are determined by a large number of genetic variants, how a polygenic trait adapts in response to a change in the environment is not completely understood. In the framework of diffusion theory, we study the steady state and the adaptation dynamics of a large but finite population evolving under stabilizing selection and symmetric mutations when selection and mutation are moderately large. We find that in the stationary state, the allele frequency distribution at a locus is unimodal if its effect size is below a threshold effect and bimodal otherwise; these results are the stochastic analog of the deterministic ones where the stable allele frequency becomes bistable when the effect size exceeds a threshold. It is known that following a sudden shift in the phenotypic optimum, in an infinitely large population, selective sweeps at a large-effect locus are prevented and adaptation proceeds exclusively via subtle changes in the allele frequency; in contrast, we find that the chance of sweep is substantially enhanced in large, finite populations and the allele frequency at a large-effect locus can reach a high frequency at short times even for small shifts in the phenotypic optimum.

Ethical layering in AI-driven polygenic risk scores-New complexities, new challenges

Researchers aim to develop polygenic risk scores as a tool to prevent and more effectively treat serious diseases, disorders and conditions such as breast cancer, type 2 diabetes mellitus and coronary heart disease. Recently, machine learning techniques, in particular deep neural networks, have been increasingly developed to create polygenic risk scores using electronic health records as well as genomic and other health data. While the use of artificial intelligence for polygenic risk scores may enable greater accuracy, performance and prediction, it also presents a range of increasingly complex ethical challenges. The ethical and social issues of many polygenic risk score applications in medicine have been widely discussed. However, in the literature and in practice, the ethical implications of their confluence with the use of artificial intelligence have not yet been sufficiently considered. Based on a comprehensive review of the existing literature, we argue that this stands in need of urgent consideration for research and subsequent translation into the clinical setting. Considering the many ethical layers involved, we will first give a brief overview of the development of artificial intelligence-driven polygenic risk scores, associated ethical and social implications, challenges in artificial intelligence ethics, and finally, explore potential complexities of polygenic risk scores driven by artificial intelligence. We point out emerging complexity regarding fairness, challenges in building trust, explaining and understanding artificial intelligence and polygenic risk scores as well as regulatory uncertainties and further challenges. We strongly advocate taking a proactive approach to embedding ethics in research and implementation processes for polygenic risk scores driven by artificial intelligence.

Intermediate molecular phenotypes to identify genetic markers of anthracycline-induced cardiotoxicity risk

Cardiotoxicity due to anthracyclines (CDA) affects cancer patients, but we cannot predict who may suffer from this complication. CDA is a complex disease whose polygenic component is mainly unidentified. We propose that levels of intermediate molecular phenotypes in the myocardium associated with histopathological damage could explain CDA susceptibility; so that variants of genes encoding these intermediate molecular phenotypes could identify patients susceptible to this complication. A genetically heterogeneous cohort of mice generated by backcrossing (N = 165) was treated with doxorubicin and docetaxel. Cardiac histopathological damage was measured by fibrosis and cardiomyocyte size by an Ariol slide scanner. We determine intramyocardial levels of intermediate molecular phenotypes of CDA associated with histopathological damage and quantitative trait loci (ipQTLs) linked to them. These ipQTLs seem to contribute to the missing heritability of CDA because they improve the heritability explained by QTL directly linked to CDA (cda-QTLs) through genetic models. Genes encoding these molecular subphenotypes were evaluated as genetic markers of CDA in three cancer patient cohorts (N = 517) whose cardiac damage was quantified by echocardiography or Cardiac Magnetic Resonance. Many SNPs associated with CDA were found using genetic models. LASSO multivariate regression identified two risk score models, one for pediatric cancer patients and the other for women with breast cancer. Molecular intermediate phenotypes associated with heart damage can identify genetic markers of CDA risk, thereby allowing a more personalized patient management. A similar strategy could be applied to identify genetic markers of other complex trait diseases.

Decomposing heritability and genetic covariance by direct and indirect effect paths

Estimation of heritability and genetic covariance is crucial for quantifying and understanding complex trait genetic architecture and is employed in almost all recent genome-wide association studies (GWAS). However, many existing approaches for heritability estimation and almost all methods for estimating genetic correlation ignore the presence of indirect genetic effects, i.e., genotype-phenotype associations confounded by the parental genome and family environment, and may thus lead to incorrect interpretation especially for human sociobehavioral phenotypes. In this work, we introduce a statistical framework to decompose heritability and genetic covariance into multiple components representing direct and indirect effect paths. Applied to five traits in UK Biobank, we found substantial involvement of indirect genetic components in shared genetic architecture across traits. These results demonstrate the effectiveness of our approach and highlight the importance of accounting for indirect effects in variance component analysis of complex traits.

Meta-analysis fine-mapping is often miscalibrated at single-variant resolution

Meta-analysis is pervasively used to combine multiple genome-wide association studies (GWASs). Fine-mapping of meta-analysis studies is typically performed as in a single-cohort study. Here, we first demonstrate that heterogeneity (e.g., of sample size, phenotyping, imputation) hurts calibration of meta-analysis fine-mapping. We propose a summary statistics-based quality-control (QC) method, suspicious loci analysis of meta-analysis summary statistics (SLALOM), that identifies suspicious loci for meta-analysis fine-mapping by detecting outliers in association statistics. We validate SLALOM in simulations and the GWAS Catalog. Applying SLALOM to 14 meta-analyses from the Global Biobank Meta-analysis Initiative (GBMI), we find that 67% of loci show suspicious patterns that call into question fine-mapping accuracy. These predicted suspicious loci are significantly depleted for having nonsynonymous variants as lead variant (2.7×; Fisher's exact p = 7.3 × 10^-4). We find limited evidence of fine-mapping improvement in the GBMI meta-analyses compared with individual biobanks. We urge extreme caution when interpreting fine-mapping results from meta-analysis of heterogeneous cohorts.