Search

We present a systematic assessment of polygenic risk score (PRS) prediction across more than 1,500 traits using genetic and phenotype data in the UK Biobank. We report 813 sparse PRS models with significant (p < 2.5 × 10−5) incremental predictive performance when compared against the covariate-only model that considers age, sex, types of genotyping arrays, and the principal component loadings of genotypes. We report a significant correlation between the number of genetic variants selected in the sparse PRS model and the incremental predictive performance (Spearman’s ρ = 0.61, p = 2.2 × 10−59 for quantitative traits, ρ = 0.21, p = 9.6 × 10−4 for binary traits). The sparse PRS model trained on European individuals showed limited transferability when evaluated on non-European individuals in the UK Biobank. We provide the PRS model weights on the Global Biobank Engine (https://biobankengine.stanford.edu/prs).Author summaryPolygenic risk scores (PRSs), an approach to estimate genetic predisposition on disease liability by aggregating the effects across multiple genetic variants, has attracted increasing research interest. While there have been improvements in the predictive performance of PRS for some traits, the applicability of PRS models across a wide range of human traits has not been clear. Here, applying penalized regression using Batch Screening Iterative Lasso (BASIL) algorithm to more than 269,000 individuals of white British ancestry in UK Biobank, we systematically characterize PRS models across more than 1,500 traits. We report 813 traits with PRS models of statistically significant predictive performance. While the statistical significance does not necessarily directly translate into clinical relevance, we investigate the properties of the 813 significant PRS models and report a significant correlation between predictive performance and estimated SNP-based heritability. We find that the number of genetic variants selected in our sparse PRS model is significantly correlated with the incremental predictive performance in both quantitative and binary traits. Our transferability assessment of PRS models in UK Biobank revealed that the sparse PRS models trained on individuals of European ancestry had a lower predictive performance for individuals of African and Asian ancestry groups.

The SARS-CoV-2 pandemic has differentially impacted populations of varied race, ethnicity and socioeconomic status. Admixture mapping and local ancestry inference represent powerful tools to examine genetic risk within multi-ancestry genomes independent of these confounding social constructs. Here, we leverage a pandemic tracking strategy in which we sequence viral and host genomes and transcriptomes from 1,327 nasopharyngeal swab residuals and integrate them with digital phenotypes from electronic health records. We demonstrate over-representation of individuals possessing Oceanian and Indigenous American ancestry in SARS-CoV-2 positive populations. Genome-wide-association disaggregated by admixture mapping reveals regions of chromosomes 5 and 14 associated with COVID19 severity within African and Oceanic local ancestries, respectively, independent of overall ancestry fraction. Phylodynamic tracking of consensus viral genomes reveals no association with disease severity or inferred ancestry. We further present summary data from a multi-omic investigation of human-leukocyte-antigen (HLA) typing, nasopharyngeal microbiome and human transcriptomics that reveal metagenomic and HLA associations with severe COVID19 infection. This work demonstrates the power of multi-omic pandemic tracking and genomic analyses to reveal distinct epidemiologic, genetic and biological associations for those at the highest risk.

Rare-variant aggregate analysis from exome and whole genome sequencing data typically summarizes with a single statistic the signal for a gene or the unit that is being aggregated. However, when doing so, the effect profile within the unit may not be easily characterized across one or multiple phenotypes. Here, we present an approach we call Multiple Rarevariants and Phenotypes Mixture Model (MRPMM), which clusters rare variants into groups based on their effects on the multivariate phenotype and makes statistical inferences about the properties of the underlying mixture of genetic effects. Using summary statistic data from a meta-analysis of exome sequencing data of 184,698 individuals in the UK Biobank across 6 populations, we demonstrate that our mixture model can identify clusters of variants responsible for significantly disparate effects across a multivariate phenotype; we study three lipid and three renal traits separately. The method is able to estimate (1) the proportion of non-null variants, (2) whether variants with the same predicted consequence in one gene behave similarly, (3) whether variants across genes share effect profiles across the multivariate phenotype, and (4) whether different annotations differ in the magnitude of their effects. As rare-variant data and aggregation techniques become more common, this method can be used to ascribe further meaning to association results.

Genome-wide association studies (GWAS) have identified hundreds of loci associated with Crohns disease (CD), however, as with all complex diseases, deriving pathogenic mechanisms from these non-coding GWAS discoveries has been challenging. To complement GWAS and better define actionable biological targets, we analysed sequenced data from more than 30,000 CD patients and 80,000 population controls. We observe rare coding variants in established CD susceptibility genes as well as ten genes where coding variation directly implicates the gene in disease risk for the first time.

BackgroundThe use of ACEI (Angiotensin-Converting Enzyme Inhibitor) and ARB (Angiotensin II Receptor Blocker) in COVID-19 remains controversial. Our main aim was to describe the effect of ACEI/ARB treatment during COVID-19 hospitalization on mortality and complications.MethodsRetrospective, observational, multicenter study, part of the SEMI-COVID-19 Registry, comparing patients with COVID-19 treated with ACEI/ARB during hospitalization to those not treated. The primary endpoint was incidence of the composite outcome of prognosis (IMV [Invasive Mechanical Ventilation], NIMV [Non-Invasive Mechanical Ventilation], ICU admission [Intensive Care Unit], and/or all-cause mortality). The secondary endpoint was incidence of MACE (Major Adverse Cardiovascular Events). We evaluated both outcomes in patients whose treatment with ACEI/ARB continued or was withdrawn during hospitalization.ResultsBetween February and June 2020, 11,205 patients were included, with mean age 67 years (SD=16.3) and 43.1% female; 2,162 patients received ACEI/ARB treatment. ACEI/ARB treatment showed a protective effect on all-cause mortality (ppConclusionACEI/ARB treatment during COVID-19 hospitalization had a protective effect on mortality. The benefits were greater in hypertensive patients, those who continued treatment during hospitalization, and those taking ARB.SummaryTreatment with ACEI/ARB during COVID-19 hospitalization showed a beneficial effect on mortality in the general population. The benefit was greater in hypertensive patients, in those who maintained treatment during hospitalization and those taking ARB.

SummaryPolygenic risk scores (PRS) aim to quantify the contribution of multiple genetic loci to an individual’s likelihood of a complex trait or disease. However, existing PRS estimate genetic liability using common genetic variants, excluding the impact of rare variants. We identified rare, large-effect variants in individuals with outlier gene expression from the GTEx project and then assessed their impact on PRS predictions in the UK Biobank (UKB). We observed large deviations from the PRS-predicted phenotypes for carriers of multiple outlier rare variants; for example, individuals classified as “low-risk” but in the top 1% of outlier rare variant burden had a 6-fold higher rate of severe obesity. We replicated these findings using data from the NHLBI Trans-Omics for Precision Medicine (TOPMed) biobank and the Million Veteran Program, and demonstrated that PRS across multiple traits will significantly benefit from the inclusion of rare genetic variants.

ObjectivesCurrently available COVID-19 prognostic models have focused on laboratory and radiological data obtained following admission. However, these tests are not available on initial assessment or in resource-limited settings. We aim to develop and validate a prediction model, based on clinical history and examination findings on initial diagnosis of COVID-19, to identify patients at risk of critical outcomes.MethodsWe used data from the SEMI-COVID-19 Registry, a nationwide multicenter cohort of consecutive patients hospitalized for COVID-19 from 132 centers in Spain. Clinical signs and symptoms, demographic variables, and medical history ascertained at hospital admission were screened using Least Absolute Shrinkage and Selection Operator (LASSO) and logistic regression to construct a predictive model. We externally validated the final model in a separate cohort of patients admitted at less-complex hospitals (< 300 beds).We undertook decision curve analysis to assess the clinical usefulness of the predictive model. The primary outcome was a composite of in-hospital death, mechanical ventilation or admission to intensive care unit.ResultsThere were 10,433 patients, 7,850 (primary outcome 25.1%) in the development cohort and 2,583 (primary outcome 27.0%) in the validation cohort. Variables in the final model included: age, cardiovascular disease, chronic kidney disease, dyspnea, tachypnea, confusion, systolic blood pressure, and SpO2≤93% or oxygen requirement.The C-statistic in the development cohort was 0.823 (95% CI,0.813-0.834). On external validation, the C-statistic was 0.792 (95% CI,0.772-0.812). The model showed a positive net benefit for threshold probabilities between 3% and 79%.ConclusionsAmong patients presenting with COVID-19, the model based on easily-obtained clinical information had good discrimination and generalizability for identifying patients at risk of critical outcomes without the need of additional testing. The online calculator provided would facilitate triage of patients during the pandemic. This study could provide a useful tool for decision-making in health systems under pandemic pressure and resource-limited settings.

The current genome-wide association studies (GWASs) do not yet capture sufficient diversity in terms of populations and scope of phenotypes. To address an essential need to expand an atlas of genetic associations in non-European populations, we conducted 220 deep-phenotype GWASs (disease endpoints, biomarkers, and medication usage) in BioBank Japan (n = 179,000), by incorporating past medical history and text-mining results of electronic medical records. Meta-analyses with the harmonized phenotypes in the UK Biobank and FinnGen (ntotal = 628,000) identified over 4,000 novel loci, which substantially deepened the resolution of the genomic map of human traits, benefited from East Asian endemic diseases and East Asian specific variants. This atlas elucidated the globally shared landscape of pleiotropy as represented by the MHC locus, where we conducted fine-mapping by HLA imputation. Finally, to intensify the value of deep-phenotype GWASs, we performed statistical decomposition of matrices of phenome-wide summary statistics, and identified the latent genetic components, which pinpointed the responsible variants and shared biological mechanisms underlying current disease classifications across populations. The decomposed components enabled genetically informed subtyping of similar diseases (e.g., allergic diseases). Our study suggests a potential avenue for hypothesis-free re-investigation of human disease classifications through genetics.

ObjectivesPending for randomized control trials, the use of tocilizumab (TCZ) in COVID-19 remains controversial. We performed a systematic review and meta-analysis to investigate the effect on clinical outcomes of TCZ to treat severe COVID-19.MethodsFrom 1 January to 21 August 2020, we searched PubMed (via MEDLINE), Scopus, and medRxiv repository databases for observational studies in any language reporting efficacy and safety of TCZ use in hospitalized adults with COVID-19. Independent and dually data extraction and quality assessment were performed.ResultsOf 57 eligible studies, 27 controlled and 30 not. The overall included patients were 8,128: 4,021 treated with TCZ, in addition to standard of care (SOC), and 4,107 only receiving SOC. The pooled mortality was lower in the TCZ-group, with a relative risk (RR) of 0.73 (95%CI 0.57-0.93; p=0.010). TCZ-treated patients were transferred to the intensive care unit (ICU) in a higher proportion, but ICU mortality was lower than in the control group. Conversely, a higher proportion of TCZ-treated patients developed secondary infections after TCZ use.ConclusionsTCZ seems beneficial in preventing in-hospital mortality in severe, non-critically ill COVID-19 patients. However, patients receiving TCZ appear to be at higher risk for secondary infections, especially those admitted to ICU.

Highlights • Dexamethasone, or alternative steroids, are recommended in severe COVID-19. • The use of tocilizumab in COVID-19, with or without steroids, is still controversial. • Risk for mortality was assessed in 186 COVID-19 patients receiving tocilizumab. • Mortality was associated with older age, chronic heart failure and liver disease. • In tocilizumab-treated patients, the additional use of steroids was beneficial., Objectives To assess the characteristics and risk factors for mortality in patients with severe COVID-19 treated with tocilizumab (TCZ), alone or in combination with corticosteroids (CS). Methods From March 17 to April 7, 2020, a real-world observational retrospective analysis of consecutive hospitalized adult patients receiving TCZ to treat severe COVID-19 was conducted at our 750-bed university hospital. The main outcome was all-cause in-hospital mortality. Results A total of 1,092 COVID-19 patients were admitted during the study period. Of them, 186 (17%) were treated with TCZ, of which 129 (87.8%) in combination with CS. Of the total 186, 155 (83.3 %) patients were receiving non-invasive ventilation when TCZ was initiated. Mean time from symptoms onset and hospital admission to TCZ use was 12 (± 4.3) and 4.3 days (± 3.4), respectively. Overall, 147 (79%) survived and 39 (21%) died. By multivariate analysis, mortality was associated with older age (HR = 1.09, p

Genetics plays a key role in drug response, affecting efficacy and toxicity. Pharmacogenomics aims to understand how genetic variation influences drug response and develop clinical guidelines to aid clinicians in personalized treatment decisions informed by genetics. Although pharmacogenomics has not been broadly adopted into clinical practice, genetics influences treatment decisions regardless. Physicians adjust patient care based on observed response to medication, which may occur as a result of genetic variants harbored by the patient. Here we seek to understand the genetics of drug selection in statin therapy, a class of drugs widely used for high cholesterol treatment. Genetics are known to play an important role in statin efficacy and toxicity, leading to significant changes in patient outcome. We performed genome-wide association studies (GWAS) on statin selection among 59,198 participants in the UK Biobank and found that variants known to influence statin efficacy are significantly associated with statin selection. Specifically, we find that carriers of variants inAPOEandLPAthat are known to decrease efficacy of treatment are more likely to be on atorvastatin, a stronger statin. Additionally, carriers of theAPOEandLPAvariants are more likely to be on a higher intensity dose (a dose that reduces low-density lipoprotein cholesterol by greater than 40%) of atorvastatin than non-carriers (APOE:p(high intensity)= 0.16, OR = 1.7,P= 1.64 × 10−4,LPA:p(high intensity)= 0.17, OR = 1.4,P= 1.14 × 10−2). These findings represent the largest genetic association study of statin selection and statin dose association to date and provide evidence for the role ofLPAandAPOEin statin response, furthering the possibility of personalized statin therapy.

During COVID19 and other viral pandemics, rapid generation of host and pathogen genomic data is critical to tracking infection and informing therapies. There is an urgent need for efficient approaches to this data generation at scale. We have developed a scalable, high throughput approach to generate high fidelity low pass whole genome and HLA sequencing, viral genomes, and representation of human transcriptome from single nasopharyngeal swabs of COVID19 patients.

BackgroundAtrial fibrillation (AF) is associated with a five-fold increased risk of ischemic stroke. A portion of this risk is heritable, however current risk stratification tools (CHA2DS2-VASc) don’t include family history or genetic risk. We hypothesized that we could improve ischemic stroke prediction in patients with AF by incorporating polygenic risk scores (PRS).ObjectivesTo construct and test a PRS to predict ischemic stroke in patients with AF, both independently and integrated with clinical risk factors.MethodsUsing data from the largest available GWAS in Europeans, we combined over half a million genetic variants to construct a PRS to predict ischemic stroke in patients with AF. We externally validated this PRS in independent data from the UK Biobank (UK Biobank), both independently and integrated with clinical risk factors.ResultsThe integrated PRS and clinical risk factors risk tool had the greatest predictive ability. Compared with the currently recommended risk tool (CHA2DS2-VASc), the integrated tool significantly improved net reclassification (NRI: 2.3% (95%CI: 1.3% to 3.0%)), and fit (χ2 P =0.002). Using this improved tool, >115,000 people with AF would have improved risk classification in the US. Independently, PRS was a significant predictor of ischemic stroke in patients with AF prospectively (Hazard Ratio: 1.13 per 1 SD (95%CI: 1.06 to 1.23))). Lastly, polygenic risk scores were uncorrelated with clinical risk factors (Pearson’s correlation coefficient: −0.018).ConclusionsIn patients with AF, there appears to be a significant association between PRS and risk of ischemic stroke. The greatest predictive ability was found with the integration of PRS and clinical risk factors, however the prediction of stroke remains challenging.

Enlargement of the aorta is an important risk factor for aortic aneurysm and dissection, a leading cause of morbidity in the developed world. While Mendelian genetics account for a portion of thoracic aortic disease, the contribution of common variation is not known. Using standard techniques in computer vision, we performed automated extraction of Ascending Aortic Diameter (AsAoD) from cardiac MRI of 36,021 individuals from the UK Biobank. A multi-ethnic genome wide association study and trans-ethnic meta-analysis identified 99 lead variants across 71 loci including genes related to cardiovascular development (HAND2, TBX20) and Mendelian forms of thoracic aortic disease (ELN, FBN1). A polygenic risk score predicted prevalent risk of thoracic aortic aneurysm within the UK Biobank (OR 1.50 per standard deviation (SD) polygenic risk score (PRS), p=6.30x10-03) which was validated across three additional biobanks including FinnGen, the Penn Medicine Biobank, and the Million Veterans Program (MVP) in individuals of European descent (OR 1.37 [1.31 - 1.43] per SD PRS), individuals of Hispanic descent (OR 1.40 [1.16 - 1.69] per SD PRS, p=5.6x10-04), and individuals of African American descent (OR 1.08 [1.00 - 1.18] per SD PRS, p=0.05). Within individuals of European descent who carried a diagnosis of thoracic aneurysm, the PRS was specifically predictive of the need for surgical intervention (OR 1.57 [1.15 - 2.15] per SD PRS, p=4.45x10-03). Using Mendelian Randomization our data highlight the primary causal role of blood pressure in reducing dilation of the thoracic aorta. Overall our findings link normal anatomic variation to extremes observed in Mendelian syndromes and provide a roadmap for the use of genetic determinants of human anatomy in both understanding cardiovascular development while simultaneously improving prediction and prevention of human disease.

In high-dimensional regression problems, often a relatively small subset of the features are relevant for predicting the outcome, and methods that impose sparsity on the solution are popular. When multiple correlated outcomes are available (multitask), reduced rank regression is an effective way to borrow strength and capture latent structures that underlie the data. Our proposal is motivated by the UK Biobank population-based cohort study, where we are faced with large-scale, ultrahigh-dimensional features, and have access to a large number of outcomes (phenotypes): lifestyle measures, biomarkers, and disease outcomes. We are hence led to fit sparse reduced-rank regression models, using computational strategies that allow us to scale to problems of this size. We use an iterative algorithm that alternates between solving the sparse regression problem and solving the reduced rank decomposition. For the sparse regression component, we propose a scalable iterative algorithm based on adaptive screening that leverages the sparsity assumption and enables us to focus on solving much smaller sub-problems. The full solution is reconstructed and tested via an optimality condition to make sure it is a valid solution for the original problem. We further extend the method to cope with practical issues such as the inclusion of confounding variables and imputation of missing values among the phenotypes. Experiments on both synthetic data and the UK Biobank data demonstrate the effectiveness of the method and the algorithm. We presentmultiSnpnetpackage, available athttp://github.com/junyangq/multiSnpnetthat works on top of PLINK2 files, which we anticipate to be a valuable tool for generating polygenic risk scores from human genetic studies.

The human leukocyte antigen (HLA) region is one of the most disease-associated regions of the human genome, yet even well-studied alleles in the HLA region have unknown impact on disease. Here, we study the effect of 156 HLA alleles on 677 binary phenotypes for 337,138 individuals in the UK Biobank. We assess HLA allele associations and subsequently use Bayesian Model Averaging for conditional analysis, a) replicating 88 known associations between HLA alleles and binary disease phenotypes such as cancer, and b) discovering 90 novel associations to phenotypes such as skin and reproductive tract cancers and to other phenotypes not previously associated with the HLA region (e.g. anemias and acne). We find several non-additive effects, suggesting a more complex landscape of disease-modifying effects throughout the region. Finally, we discover associations between homozygous HLA allele burden and several cancer and other phenotypes, suggesting that peptide presentation spectra as coded for by the HLA region are important in determining disease risk. Our results demonstrate the HLA region’s complexity and richness while underscoring its clinical relevance.

BackgroundSpain has been one of the countries most affected by the COVID-19 pandemic.ObjectiveTo create a registry of patients with COVID-19 hospitalized in Spain in order to improve our knowledge of the clinical, diagnostic, therapeutic, and prognostic aspects of this disease.MethodsA multicentre retrospective cohort study, including consecutive patients hospitalized with confirmed COVID-19 throughout Spain. Epidemiological and clinical data, additional tests at admission and at seven days, treatments administered, and progress at 30 days of hospitalization were collected from electronic medical records.ResultsUp to April 30th 2020, 6,424 patients from 109 hospitals were included. Their median age was 69.1 years (range: 18-102 years) and 56.9% were male. Prevalences of hypertension, dyslipidemia, and diabetes mellitus were 50.2%, 39.7%, and 18.7%, respectively. The most frequent symptoms were fever (86.2%) and cough (76.5%). High values of ferritin (72.4%), lactate dehydrogenase (70.2%), and D-dimer (61.5%), as well as lymphopenia (52.6%), were frequent. The most used antiviral drugs were hydroxychloroquine (85.7%) and lopinavir/ritonavir (62.4%). 31.5% developed respiratory distress. Overall mortality rate was 21.1%, with a marked increase with age (50-59 years: 4.2%, 60-69 years: 9.1%, 70-79 years: 21.4%, 80-89 years: 42.5%, ≥ 90 years: 51.1%).ConclusionsThe SEMI-COVID-19 Network provides data on the clinical characteristics of patients with COVID-19 hospitalized in Spain. Patients with COVID-19 hospitalized in Spain are mostly severe cases, as one in three patients developed respiratory distress and one in five patients died. These findings confirm a close relationship between advanced age and mortality.

Parkinson’s disease is characterized by dopaminergic neurodegeneration in the substantia nigra. Although dopaminergic drugs are the mainstay of Parkinson’s treatment, their putative disease-modifying properties remain controversial. We explored whether prescription of dopaminergic stimulants for attention-deficit hyperactivity disorder (ADHD) might affect Parkinson’s incidence. We performed Cox survival analyses for outpatient Parkinson’s diagnosis among ADHD-diagnosed seniors in the Optum Clinformatics™ Data Mart de-identified administrative claims database, correcting for diverse demographic and socio-economic status covariates. We compared 5,683 sustained users (≥ 90 days) of dopaminergic stimulants to 252 sustained users of atomoxetine, a noradrenergic first-line ADHD medication. Parkinson’s incidence was reduced among sustained dopaminergic stimulant users compared to atomoxetine users (adjusted hazard ratio [HR] 0.15, 95% confidence interval [CI] 0.04-0.56, p = 0.005). Effect sizes were comparable between derivatives of amphetamine (adjusted HR 0.12, 95% CI 0.03-0.48, p = 0.003) and methylphenidate (adjusted HR 0.27, 95% CI 0.04-1.76, p = 0.2). In sensitivity analyses, similar trends were observed when other psychotropics (SSRIs, gabapentin) were used as comparators instead of atomoxetine, or when the threshold for sustained use was defined as 45, 180 or 360 days instead of 90. Thus, sustained dopaminergic stimulant use was associated with lower Parkinson’s incidence among seniors with ADHD. Our results are consistent with a protective effect of dopaminergic stimulants on the development of Parkinson’s, and support a re-examination of certain dopaminergics, particularly rasagiline and other selective monoamine oxidase B inhibitors, as potential disease-modifying agents.

The aortic valve is an important determinant of cardiovascular physiology and anatomic location of common human diseases. From a sample of 26,142 European-ancestry participants, we estimated functional aortic valve area by planimetry from prospectively obtained cardiac MRI sequences of the aortic valve. A genome-wide association study of aortic valve area in these UK Biobank participants showed two significant associations indexed by rs71190365 (chr13:50764607,DLEU1, p=1.8×10−9) and rs35991305 (chr12:94191968,CRADD, p=3.4×10−8). From the GWAS findings we constructed a polygenic risk score for aortic valve area, which in a separate cohort of 311,728 individuals without imaging demonstrated that smaller aortic valve area is predictive of increased risk for aortic valve disease (Odds Ratio 0.88,p=2.3×10−6). After excluding subjects with a medical diagnosis of aortic valve stenosis (remaining n=310,546 individuals), phenome-wide association of >10,000 traits showed multiple links between the polygenic score for aortic valve disease and key health-related comorbidities involving the cardiovascular system and autoimmune disease. Genetic correlation analysis supports a shared genetic etiology with between aortic valve size and birthweight along with other cardiovascular conditions. These results illustrate the use of automated phenotyping of cardiac imaging data from the general population to investigate the genetic etiology of aortic valve disease, perform clinical prediction, and uncover new clinical and genetic correlates of cardiac anatomy.

Summary We develop a scalable and highly efficient algorithm to fit a Cox proportional hazard model by maximizing the $L^1$-regularized (Lasso) partial likelihood function, based on the Batch Screening Iterative Lasso (BASIL) method developed in Qian and others (2019). Our algorithm is particularly suitable for large-scale and high-dimensional data that do not fit in the memory. The output of our algorithm is the full Lasso path, the parameter estimates at all predefined regularization parameters, as well as their validation accuracy measured using the concordance index (C-index) or the validation deviance. To demonstrate the effectiveness of our algorithm, we analyze a large genotype-survival time dataset across 306 disease outcomes from the UK Biobank (Sudlow and others, 2015). We provide a publicly available implementation of the proposed approach for genetics data on top of the PLINK2 package and name it snpnet-Cox.

G protein-coupled receptors (GPCRs) drive an array of important physiological functions and are the targets of nearly one-third of all FDA approved drugs. Large scale genomic initiatives are mapping the genetic diversity in GPCRs, however, a map of which GPCR genetic variants are associated with phenotypic variation and disease is lacking. Furthermore, the mechanistic basis of how the individual GPCR genetic variants regulate molecular function is also largely unknown. We performed a phenome-wide association analysis for 269 common protein-altering variants in 156 GPCRs and 275 phenotypes using data from 337,205 unrelated white British UK Biobank participants and identified 138 associations at a false discovery rate of 5%. We found a novel association between rs12295710 in MRGPRE , a member of the Mas-related receptor family involved in nociception, and migraine risk. We also identified an association between rs3732378, a missense mutation in the binding pocket of CX3CR1, and hypothyroidism. Five orphan GPCRs had eight genetic associations, highlighting novel biology for these receptors of unknown function. We found several associations between GPCR variants and food intake phenotypes, including an association between the variants in TAS2R38 known to affect the ability to taste phenylthiocarbamide and tea intake as well as a non-additive associations between variants in TAS2R19 and TAS2R31 and coffee and tea intake. Finally, we tested whether genetic variants in ADRB2 associated with immune cell amounts and pulmonary function affect downstream signaling pathways and found that two ADRB2 haplotypes are associated with differential signaling relative to the most common haplotype. Overall, this study provides a map of genetic associations for GPCR coding variants across a wide variety of phenotypes that can inform future drug discovery efforts targeting GPCRs.

A physically active lifestyle is essential for maintaining health, and is a powerful way to prevent chronic disease. However, the molecular mechanisms that drive exercise adaptation and transduce its beneficial effects, are incompletely understood. Here, we combined data from 49 studies that measured the whole transcriptome in humans before and after exercise to provide the power to draw novel observations not seen in any individual study alone. The resulting curated and standardized resource includes samples from skeletal muscle (n=1,260) and blood (n=726) in response to endurance or resistance exercise and training. Using a linear mixed effects meta-regression model selection strategy, we detected specific time patterns and novel regulatory modulators of the acute exercise response. Acute and long term responses to exercise were transcriptionally distinct. Exercise induced a more pronounced inflammatory response in skeletal muscle of older individuals. We identified multiple sex-specific response genes, where MTMR3 is a novel exercise-regulated gene. These results deepen our understanding of the transcriptional responses to exercise and provide a powerful resource for future research efforts in exercise physiology and medicine.

The clinical evaluation of a genetic syndrome relies upon recognition of a characteristic pattern of signs or symptoms to guide targeted genetic testing for confirmation of the diagnosis. However, individuals displaying a few phenotypes of a complex syndrome may not meet criteria for clinical diagnosis or genetic testing. Here, we present a phenome-wide association study (PheWAS) approach to systematically explore pleiotropy of common and rare alleles in genes associated with four well-described syndromic diseases (Alagille (AS), Marfan (MS), DiGeorge (DS), and Noonan (NS) syndromes) in the general population.Using human phenotype ontology (HPO) terms, we systematically mapped 60 phenotypes related to AS, MS, DS and NS in 337,198 unrelated white British from the UK Biobank (UKBB) based on their hospital admission records, self-administrated questionnaires, and physiological measurements. We performed logistic regression adjusting for age, sex, and the first 5 genetic principal components, for each phenotype and each variant in the target genes (JAG1, TBX1, FBN1, PTPN11, NOTCH2, and MAP2K1) and performed a gene burden testing.Overall, we observed multiple phenotype-genotype correlations, such as the association between variation in JAG1, FBN1, PTPN11 and SOS2 with diastolic and systolic blood pressure; and pleiotropy among multiple variants in syndromic genes. For example, rs11066309 in PTPN11 was significantly associated with a lower body mass index, an increased risk of hypothyroidism and a smaller size for gestational age, all in concordance with NS-related phenotypes. Similarly, rs589668 in FBN1 was associated with an increase in body height and blood pressure, and a reduced body fat percentage as observed in Marfan syndrome.Our findings suggest that the spectrum of associations of common and rare variants in genes involved in syndromic diseases can be extended to individual phenotypes within the general population.Author SummaryStandard medical evaluation of genetic syndromes relies upon recognizing a characteristic pattern of signs or symptoms to guide targeted genetic testing for confirmation of the diagnosis. This may lead to missing diagnoses in patients with silent or a low expressed form of the syndrome. Here we take advantage of a rich electronic health record, various phenotypic measurements, and genetic information in 337,198 unrelated white British from the UKBB, to study the relation between single syndromic disease phenotypes and genes related to syndromic disease. We show multiple phenotype-genotypes associations in concordance with phenotypes variations found in syndromic diseases. For example, we show that mutation in FBN1 was associated with high standing/sitting height ratio and reduced body fat percentage as observed in individuals with Marfan syndrome. Our findings suggest that common and rare alleles in SD genes are causative of individual component phenotypes present in a general population; further research is needed to characterize the pleiotropic effect of alleles in syndromic genes in persons without the syndromic disease.

Polygenic risk models have led to significant advances in understanding complex diseases and their clinical presentation. While traditional models of genetic risk like polygenic risk scores (PRS) can effectively predict outcomes, they do not generally account for disease subtypes or pathways which underlie within-trait diversity. Here, we introduce a latent factor model of genetic risk based on components from Decomposition of Genetic Associations (DeGAs), which we call the DeGAs polygenic risk score (dPRS). We compute DeGAs on associations from 1,905 traits in the UK Biobank and find that dPRS performs comparably to standard PRS while offering greater interpretability. We highlight results for body mass index (BMI), myocardial infarction (heart attack), and gout in 337,151 white British individuals (spilt 70/10/20 for training, validation, and testing), with replication in a further set of 25,486 non-British whites from the Biobank. We show how to decompose an individual’s genetic risk for a trait across these latent components. For example, we find that BMI polygenic risk factorizes into distinct components relating to fat-free mass, fat mass, and overall health indicators like sleep duration and alcohol and water intake. Most individuals with high dPRS for BMI have strong contributions from both a fat mass component and a fat-free mass component, whereas a few ‘outlier’ individuals have strong contributions from only one of the two components. Our methods enable fine-scale interpretation of the drivers of genetic risk for complex traits.

Protein-altering variants that are protective against human disease provide in vivo validation of therapeutic targets. Here we use genotyping data in UK Biobank and FinnGen to conduct a search for protein-altering variants conferring lower intraocular pressure (IOP) and protection against glaucoma. Through protein-altering variant association analysis we find a missense variant in UK Biobank (rs28991009 (MAF=0.8%) genotyped in 81,527 individuals with measured IOP and an independent set of 4,269 glaucoma patients and 251,355 controls) that significantly lowers IOP (β = −0.73 mmHg for heterozygotes, −2.96 mmHg for homozygotes, P = 1 × 10−13) and is associated with 34% reduced risk of glaucoma (P = 0.005). In FinnGen, we identify an ANGPTL7 missense variant at a greater than 50-fold increased frequency in Finland compared with other populations (rs147660927, p.Arg220Cys, MAF Finland = 4.1%), which was genotyped in 5,177 glaucoma patients and 130,461 controls and is associated with 30% lower glaucoma risk (P = 1 × 10−9). We further find three rarer variants in UK Biobank, including a protein-truncating variant, which confer a strong composite lowering of IOP (P = 0.002), suggesting the protective mechanism likely resides in the loss of an interaction or function. Our results support inhibition or down-regulation of ANGPTL7 as a therapeutic strategy for glaucoma

Availability of large genetic databases has led to the development of powerful causal inference methods that use genetic variables as instruments to estimate causal effects. Such methods typically make many assumptions about the underlying causal graphical model, are limited in the patterns they search for in the data, and there is no guide for systematic analysis of a large database. Here, we present cGAUGE, a new pipeline for causal Graphical Analysis Using GEnetics that utilizes large changes in the significance of local conditional independencies between the genetic instruments and the phenotypes. We detect cases where causal inference can be performed with minimal risk of horizontal pleiotropy. Moreover, we search for new graphical patterns to reveal novel information about the underlying causal diagram that is not covered by extant methods, including new direct links, colliders, and evidence for confounding. We present theoretical justification, simulations, and apply our pipeline to 70 complex phenotypes from 337,198 subjects from the UK Biobank. Our results cover 102 detected causal relationships, of which some are new and many are expected. For example, we detect a direct causal link from high cholesterol to angina and a feedback loop between angina and myocardial infarction. We also corroborate a recent observational link between asthma and Crohn’s disease. Finally, we detect important features of the causal network structure including several causal hubs such as intelligence and waist circumference.

Copy number variations (CNV) represent a significant proportion of the genetic differences between individuals and many CNVs associate causally with syndromic disease and clinical outcomes. Here, we characterize the landscape of copy number variation and their phenome-wide effects in a sample of 472,228 array-genotyped individuals from the UK Biobank. In addition to population-level selection effects against genic loci conferring high-mortality, we describe genetic burden from syndromic and previously uncharacterized CNV loci across nearly 2,000 quantitative and dichotomous traits, with separate analyses for common and rare classes of variation. Specifically, we highlight the effects of CNVs at two well-known syndromic loci 16p11.2 and 22q11.2, as well as novel associations at 9p23, in the context of acute coronary artery disease and high body mass index. Our data constitute a deeply contextualized portrait of population-wide burden of copy number variation, as well as a series of known and novel dosage-mediated genic associations across the medical phenome.

Population-based biobanks with genomic and dense phenotype data provide opportunities for generating effective therapeutic hypotheses and understanding the genomic role in disease predisposition. To characterize latent components of genetic associations, we applied truncated singular value decomposition (DeGAs) to matrices of summary statistics derived from genome-wide association analyses across 2,138 phenotypes measured in 337,199 White British individuals in the UK Biobank study. We systematically identified key components of genetic associations and the contributions of variants, genes, and phenotypes to each component. As an illustration of the utility of the approach to inform downstream experiments, we report putative loss of function variants, rs114285050 (GPR151) and rs150090666 (PDE3B), that substantially contribute to obesity-related traits, and experimentally demonstrate the role of these genes in adipocyte biology. Our approach to dissect components of genetic associations across the human phenome will accelerate biomedical hypothesis generation by providing insights on previously unexplored latent structures.

ObjectiveCurrently, dedicated tagging staff spend considerable effort assigning clinical codes to patient summaries for public health purposes, and machine-learning automated tagging is bottlenecked by availability of electronic medical records. Veterinary medical records, a largely untapped data source that could benefit both human and non-human patients, could fill the gap.Materials and MethodsIn this retrospective study, we trained long short-term memory (LSTM) recurrent neural networks (RNNs) on 52,722 human and 89,591 veterinary records. We established relevant baselines by training Decision Trees (DT) and Random Forests (RF) on the same data. We finally investigated the effect of merging data across clinical settings and probed model portability.ResultsWe show that the LSTM-RNNs accurately classify veterinary/human text narratives into top-level categories with an average weighted macro F1, score of 0.735/0.675 respectively. The evaluation metric for the LSTM was 7 and 8% higher than that of the DT and RF models respectively. We generally did not find evidence of model portability albeit moderate performance increases in select categories.DiscussionWe see a strong positive correlation between number of training samples and classification performance, which is promising for future efforts. The use of LSTM-RNN models represents a scalable structure that could prove useful in cohort selection, which could in turn better address emerging public health concerns.ConclusionDigitization of human and veterinary health information will continue to be a reality. Our approach is a step forward for these two domains to learn from, and inform, one another.

Whole genome sequencing studies applied to large populations or biobanks with extensive phenotyping raise new analytic challenges. The need to consider many variants at a locus or group of genes simultaneously and the potential to study many correlated phenotypes with shared genetic architecture provide opportunities for discovery and inference that are not addressed by the traditional one variant, one phenotype association study. Here, we introduce a Bayesian model comparison approach that we refer to as MRP (Multiple Rare-variants and Phenotypes) for rare-variant association studies that considers correlation, scale, and direction of genetic effects across a group of genetic variants, phenotypes, and studies. The approach requires only summary statistic data. To demonstrate the efficacy of MRP, we apply our method to exome sequencing data (N = 184,698) across 2,019 traits from the UK Biobank, aggregating signals in genes. MRP demonstrates an ability to recover previously-verified signals such as associations between PCSK9 and LDL cholesterol levels. We additionally find MRP effective in conducting meta-analyses in exome data. Notable non-biomarker findings include associations between MC1R and red hair color and skin color, IL17RA and monocyte count, IQGAP2 and mean platelet volume, and JAK2 and platelet count and crit (mass). Finally, we apply MRP in a multi-phenotype setting; after clustering the 35 biomarker phenotypes based on genetic correlation estimates into four clusters, we find that joint analysis of these phenotypes results in substantial power gains for gene-trait associations, such as in TNFRSF13B in one of the clusters containing diabetes and lipid-related traits. Overall, we show that the MRP model comparison approach is able to improve upon useful features from widely-used meta-analysis approaches for rare variant association analyses and prioritize protective modifiers of disease risk.

PCSK9 inhibitors are a potent new therapy for hypercholesterolemia and have been shown to decrease risk of coronary heart disease. Although short-term clinical trial results have not demonstrated major adverse effects, long-term data will not be available for some time. Genetic studies in large well-phenotyped biobanks offer a unique opportunity to predict drug effects and provide context for the evaluation of future clinical trial outcomes. We tested association of the PCSK9 loss-of-function variant rsll591147 (R46L) in a hypothesis-driven 11 phenotype set and a hypothesis-generating 278 phenotype set in 337,536 individuals of British ancestry in the United Kingdom Biobank (UKB), with independent discovery (n = 225K) and replication (n = 112K). In addition to the known association with lipid levels (OR 0.63) and coronary heart disease (OR 0.73), the T allele of rs11591147 showed a protective effect on ischemic stroke (OR 0.61, p = 0.002) but not hemorrhagic stroke in the hypothesis-driven screen. We did not observe an association with type 2 diabetes, cataracts, heart failure, atrial fibrillation, and cognitive dysfunction. In the phenome-wide screen, the variant was associated with a reduction in metabolic disorders, ischemic heart disease, coronary artery bypass graft operations, percutaneous coronary interventions and history of angina. A single variant analysis of UKB data using TreeWAS, a Bayesian analysis framework to study genetic associations leveraging phenotype correlations, also showed evidence of association with cerebral infarction and vascular occlusion. This result represents the first genetic evidence in a large cohort for the protective effect of PCSK9 inhibition on ischemic stroke, and corroborates exploratory evidence from clinical trials. PCSK9 inhibition was not associated with variables other than those related to low density lipoprotein cholesterol and atherosclerosis, suggesting that other effects are either small or absent.

Transcriptome-wide association studies (TWAS) integrate GWAS and gene expression datasets to find gene-trait associations. In this Perspective, we explore properties of TWAS as a potential approach to prioritize causal genes, using simulations and case studies of literature-curated candidate causal genes for schizophrenia, LDL cholesterol and Crohn’s disease. We explore risk loci where TWAS accurately prioritizes the likely causal gene, as well as loci where TWAS prioritizes multiple genes, some of which are unlikely to be causal, because they share the same variants as eQTLs. We illustrate that TWAS is especially prone to spurious prioritization when using expression data from tissues or cell types that are less related to the trait, due to substantial variation in both expression levels and eQTL strengths across cell types. Nonetheless, TWAS prioritizes candidate causal genes at GWAS loci more accurately than simple baselines based on proximity to lead GWAS variant and expression in trait-related tissue. We discuss current strategies and future opportunities for improving the performance of TWAS for causal gene prioritization. Our results showcase the strengths and limitations of using expression variation across individuals to determine causal genes at GWAS loci and provide guidelines and best practices when using TWAS to prioritize candidate causal genes.

Genomic imprinting is an important regulatory mechanism that silences one of the parental copies of a gene. To systematically characterize this phenomenon, we analyze tissue specificity of imprinting from allelic expression data in 1582 primary tissue samples from 178 individuals from the Genotype-Tissue Expression (GTEx) project. We characterize imprinting in 42 genes, including both novel and previously identified genes. Tissue specificity of imprinting is widespread, and gender-specific effects are revealed in a small number of genes in muscle with stronger imprinting in males. IGF2 shows maternal expression in the brain instead of the canonical paternal expression elsewhere. Imprinting appears to have only a subtle impact on tissue-specific expression levels, with genes lacking a systematic expression difference between tissues with imprinted and biallelic expression. In summary, our systematic characterization of imprinting in adult tissues highlights variation in imprinting between genes, individuals, and tissues.

Protein-truncating variants can have profound effects on gene function and are critical for clinical genome interpretation and generating therapeutic hypotheses, but their relevance to medical phenotypes has not been systematically assessed. We characterized the effect of 18,228 protein-truncating variants across 135 phenotypes from the UK Biobank and found 27 associations between medical phenotypes and protein-truncating variants in genes outside the major histocompatibility complex. We performed phenome-wide analyses and directly measured the effect of homozygous carriers, commonly referred to as “human knockouts,” across medical phenotypes for genes implicated to be protective against disease or associated with at least one phenotype in our study and found several genes with strong pleiotropic or non-additive effects. Our results illustrate the importance of protein-truncating variants in a variety of diseases.

As part of a broader collaborative network of exome sequencing studies, we developed a jointly called data set of 5,685 Ashkenazi Jewish exomes. We make publicly available a resource of site and allele frequencies, which should serve as a reference for medical genetics in the Ashkenazim. We estimate that 30% of protein-coding alleles present in the Ashkenazi Jewish population at frequencies greater than 0.2% are significantly more frequent (mean 7.6-fold) than their maximum frequency observed in other reference populations. Arising via a well-described founder effect, this catalog of enriched alleles can contribute to differences in genetic risk and overall prevalence of diseases between populations. As validation we document 151 AJ enriched protein-altering alleles that overlap with “pathogenic” ClinVar alleles, including those that account for 10-100 fold differences in prevalence between AJ and non-AJ populations of some rare diseases including Gaucher disease (GBA, p.Asn409Ser, 8-fold enrichment); Canavan disease (ASPA, p.Glu285Ala, 12-fold enrichment); and Tay-Sachs disease (HEXA, c.1421+1G>C, 27-fold enrichment; p.Tyr427IlefsTer5, 12-fold enrichment). We next sought to use this catalog, of well-established relevance to Mendelian disease, to explore Crohn’s disease, a common disease with an estimated two to four-fold excess prevalence in AJ. We specifically evaluate whether strong acting rare alleles, enriched by the same founder-effect, contribute excess genetic risk to Crohn’s disease in AJ, and find that ten rare genetic risk factors in NOD2 and LRRK2 are strongly enriched in AJ, including several novel contributing alleles, show evidence of association to CD. Independently, we find that genomewide common variant risk defined by GWAS shows a strong difference between AJ and non-AJ European control population samples (0.97 s.d. higher, p−16). Taken together, the results suggest coordinated selection in AJ population for higher CD risk alleles in general. The results and approach illustrate the value of exome sequencing data in case-control studies along with reference data sets like ExAC to pinpoint genetic variation that contributes to variable disease predisposition across populations.

Recent understanding of the causal role of blood-detectable somatic protein-truncating DNA variants in leukemia prompts questions about the generalizability of such observations across cancer types. We used the cancer genome atlas exome sequencing (~8000 samples) to compare 22 different cancer phenotypes with more than 6000 controls using a case–control study design and demonstrate that mosaic protein truncating variants in these genes are also associated with solid-tumor cancers. The absence of these cancer-associated mosaic variants from the tumors themselves suggest these are not themselves tumor drivers. Through analysis of different cancer phenotypes we observe gene-specificity for mosaic mutations. We confirm a specific link between PPM1D and ovarian cancer, consistent with previous reports linking PPM1D to breast and ovarian cancer. Additionally, glioblastoma, melanoma and lung cancers show gene specific burdens of mosaic protein truncating mutations. Taken together, these results extend existing observations and broadly link solid-tumor cancers to somatic blood DNA changes., Cancer: Blood mutations linked to solid tumor risk Increased risk for solid tumors linked to mutations detected in blood samples that shorten the coding sequence of their genes. Mark Daly and colleagues from the Broad Institute in Cambridge, Massachusetts, USA, used large genomic databases to test whether having blood cells both with and without genetic variants predicted to shorten the encoded protein—a phenomenon known as mosaic protein-truncating variants (PTVs)—was associated with developing a range of solid-tumor cancers. They studied DNA from around 8,000 people with cancer and 6,000 healthy controls. Other studies have shown that mosaic PTVs precede and predict the development of leukemia. In this study, the scientists not only confirmed previous reports linking these variants to breast and ovarian cancer but also extended the association to include tumors of the brain, skin and lungs. These results broadly connect cancer to blood DNA changes.

We conducted a search for protein truncating variants conferring protection against inflammatory bowel disease exploiting knowledge of common variants associated with the same disease. We found that a protein truncating variant (rs36095412, p.R179X, genotyped in 11,148 ulcerative colitis patients and 295,446 controls, MAF = up to 0.78%) in RNF186, a single-exon ring finger E3 ligase with strong colonic expression, protects against ulcerative colitis (overall P = 6.89×10-7, odds ratio (OR) = 0.30). We further demonstrate that the truncated protein is expressed, suggesting the protective mechanism may reside in the loss of an interaction or function via mislocalization or loss of an essential protein element.

Motivation: RNA sequencing enables allele specific expression (ASE) studies that complement standard genotype expression studies for common variants and, importantly, also allow measuring the regulatory impact of rare variants. The Genotype-Tissue Expression project (GTEx) is collecting RNA-seq data on multiple tissues of a same set of individuals and novel methods are required for the analysis of these data. Results: We present a statistical method to compare different patterns of ASE across tissues and to classify genetic variants according to their impact on the tissue-wide expression profile. We focus on strong ASE effects that we are expecting to see for protein-truncating variants, but our method can also be adjusted for other types of ASE effects. We illustrate the method with a real data example on a tissue-wide expression profile of a variant causal for lipoid proteinosis, and with a simulation study to assess our method more generally. Availability: MAMBA software: http://birch.well.ox.ac.uk/~rivas/mamba/ R source code and data examples: http://www.iki.fi/mpirinen/ Contact: matti.pirinen@helsinki.fi rivas@well.ox.ac.uk