scholarly journals Improved genetic prediction of complex traits from individual-level data or summary statistics

2021 ◽  
Vol 12 (1) ◽  
Author(s):  
Qianqian Zhang ◽  
Florian Privé ◽  
Bjarni Vilhjálmsson ◽  
Doug Speed
Keyword(s):  
Complex Traits ◽  
Prediction Models ◽  
Phenotypic Variance ◽  
Uk Biobank ◽  
Genetic Prediction ◽  
Individual Level ◽  
Prediction Tools ◽  
Data Prediction ◽  
Level Data ◽  
Variance Explained

AbstractMost existing tools for constructing genetic prediction models begin with the assumption that all genetic variants contribute equally towards the phenotype. However, this represents a suboptimal model for how heritability is distributed across the genome. Therefore, we develop prediction tools that allow the user to specify the heritability model. We compare individual-level data prediction tools using 14 UK Biobank phenotypes; our new tool LDAK-Bolt-Predict outperforms the existing tools Lasso, BLUP, Bolt-LMM and BayesR for all 14 phenotypes. We compare summary statistic prediction tools using 225 UK Biobank phenotypes; our new tool LDAK-BayesR-SS outperforms the existing tools lassosum, sBLUP, LDpred and SBayesR for 223 of the 225 phenotypes. When we improve the heritability model, the proportion of phenotypic variance explained increases by on average 14%, which is equivalent to increasing the sample size by a quarter.

scholarly journals Improved genetic prediction of complex traits from individual-level data or summary statistics

2020 ◽  
Author(s):  
Qianqian Zhang ◽  
Florian Privé ◽  
Bjarni Vilhjálmsson ◽  
Doug Speed
Keyword(s):  
Ridge Regression ◽  
Complex Traits ◽  
Prediction Models ◽  
Realistic Model ◽  
Phenotypic Variance ◽  
Summary Statistics ◽  
Genetic Prediction ◽  
Individual Level ◽  
Level Data ◽  
The Uk

At present, most tools for constructing genetic prediction models begin with the assumption that all genetic variants contribute equally towards the phenotype. However, this represents a sub-optimal model for how heritability is distributed across the genome. Here we construct prediction models for 14 phenotypes from the UK Biobank (200,000 individuals per phenotype) using four of the most popular prediction tools: lasso, ridge regression, Bolt-LMM and BayesR. When we improve the assumed heritability model, prediction accuracy always improves (i.e., for all four tools and for all 14 phenotypes). When we construct prediction models using individual-level data, the best-performing tool is Bolt-LMM; if we replace its default heritability model with the most realistic model currently available, the average proportion of phenotypic variance explained increases by 19% (s.d. 2), equivalent to increasing the sample size by about a quarter. When we construct prediction models using summary statistics, the best tool depends on the phenotype. Therefore, we develop MegaPRS, a summary statistic prediction tool for constructing lasso, ridge regression, Bolt-LMM and BayesR prediction models, that allows the user to specify the heritability model.

scholarly journals GxEsum: a novel approach to estimate the phenotypic variance explained by genome-wide GxE interaction based on GWAS summary statistics for biobank-scale data

2021 ◽  
Vol 22 (1) ◽  
Author(s):  
Jisu Shin ◽  
Sang Hong Lee
Keyword(s):  
Complex Traits ◽  
Error Rates ◽  
Type I ◽  
Phenotypic Variance ◽  
Environment Interaction ◽  
Summary Statistics ◽  
Gxe Interaction ◽  
Genome Wide ◽  
Scale Data ◽  
Variance Explained

AbstractGenetic variation in response to the environment, that is, genotype-by-environment interaction (GxE), is fundamental in the biology of complex traits and diseases. However, existing methods are computationally demanding and infeasible to handle biobank-scale data. Here, we introduce GxEsum, a method for estimating the phenotypic variance explained by genome-wide GxE based on GWAS summary statistics. Through comprehensive simulations and analysis of UK Biobank with 288,837 individuals, we show that GxEsum can handle a large-scale biobank dataset with controlled type I error rates and unbiased GxE estimates, and its computational efficiency can be hundreds of times higher than existing GxE methods.

scholarly journals Analysis of genetic dominance in the UK Biobank

2021 ◽  
Author(s):  
Duncan S Palmer ◽  
Wei Zhou ◽  
Liam Abbott ◽  
Nik Baya ◽  
Claire Churchhouse ◽  
...  
Keyword(s):  
Complex Traits ◽  
Multiple Testing ◽  
Model Organisms ◽  
Systematic Evaluation ◽  
Hair Color ◽  
Phenotypic Variance ◽  
Additive Effects ◽  
Uk Biobank ◽  
Genome Wide ◽  
The Uk

In classical statistical genetic theory, a dominance effect is defined as the deviation from a purely additive genetic effect for a biallelic variant. Dominance effects are well documented in model organisms. However, evidence in humans is limited to a handful of traits, particularly those with strong single locus effects such as hair color. We carried out the largest systematic evaluation of dominance effects on phenotypic variance in the UK Biobank. We curated and tested over 1,000 phenotypes for dominance effects through GWAS scans, identifying 175 loci at genome-wide significance correcting for multiple testing (P < 4.7 × 10-11). Power to detect non-additive loci is much lower than power to detect additive effects for complex traits: based on the relative effect sizes at genome-wide significant additive loci, we estimate a factor of 20-30 increase in sample size will be necessary to capture clear evidence of dominance similar to those currently observed for additive effects. However, these localised dominance hits do not extend to a significant aggregate contribution to phenotypic variance genome-wide. By deriving a version of LD-score regression to detect dominance effects tagged by common variation genome-wide (minor allele frequency > 0.05), we found no strong evidence of a contribution to phenotypic variance when accounting for multiple testing. Across the 267 continuous and 793 binary traits the median contribution was 5.73 × 10-4, with unbiased point estimates ranging from -0.261 to 0.131. Finally, we introduce dominance fine-mapping to explore whether the more rapid decay of dominance LD can be leveraged to find causal variants. These results provide the most comprehensive assessment of dominance trait variation in humans to date.

scholarly journals A method to estimate the contribution of regional genetic associations to complex traits from summary association statistics

2015 ◽  
Author(s):  
Guillaume Pare ◽  
Shihong Mao ◽  
Wei Q. Deng
Keyword(s):  
Complex Traits ◽  
Regression Models ◽  
Phenotypic Variance ◽  
Genetic Associations ◽  
Multiple Regression Models ◽  
Novel Method ◽  
Combine Information ◽  
Regional Associations ◽  
Retirement Study ◽  
Variance Explained

AbstractDespite considerable efforts, known genetic associations only explain a small fraction of predicted heritability. Regional associations combine information from multiple contiguous genetic variants and can improve variance explained at established association loci. However, regional associations are not easily amenable to estimation using summary association statistics because of sensitivity to linkage disequilibrium (LD). We now propose a novel method to estimate phenotypic variance explained by regional associations using summary statistics while accounting for LD. Our method is asymptotically equivalent to multiple regression models when no interaction or haplotype effects are present. It has multiple applications, such as ranking of genetic regions according to variance explained or comparison of variance explained by to or more regions. Using height and BMI data from the Health Retirement Study (N=7,776), we show that most genetic variance lies in a small proportion of the genome and that previously identified linkage peaks have higher than expected regional variance.

scholarly journals Mendelian randomization while jointly modeling cis genetics identifies causal relationships between gene expression and lipids

2020 ◽  
Vol 11 (1) ◽  
Author(s):  
Adriaan van der Graaf ◽  
◽  
Annique Claringbould ◽  
Antoine Rimbert ◽  
Harm-Jan Westra ◽  
...  
Keyword(s):  
Gene Expression ◽  
Complex Traits ◽  
Mendelian Randomization ◽  
Low Density Lipoprotein ◽  
Density Lipoprotein ◽  
Summary Statistics ◽  
Low Density Lipoprotein Cholesterol ◽  
Causal Inferences ◽  
Individual Level ◽  
Level Data

Abstract Inference of causality between gene expression and complex traits using Mendelian randomization (MR) is confounded by pleiotropy and linkage disequilibrium (LD) of gene-expression quantitative trait loci (eQTL). Here, we propose an MR method, MR-link, that accounts for unobserved pleiotropy and LD by leveraging information from individual-level data, even when only one eQTL variant is present. In simulations, MR-link shows false-positive rates close to expectation (median 0.05) and high power (up to 0.89), outperforming all other tested MR methods and coloc. Application of MR-link to low-density lipoprotein cholesterol (LDL-C) measurements in 12,449 individuals with expression and protein QTL summary statistics from blood and liver identifies 25 genes causally linked to LDL-C. These include the known SORT1 and ApoE genes as well as PVRL2, located in the APOE locus, for which a causal role in liver was not known. Our results showcase the strength of MR-link for transcriptome-wide causal inferences.

scholarly journals Phenome-wide Heritability Analysis of the UK Biobank

2016 ◽  
Author(s):  
Tian Ge ◽  
Chia-Yen Chen ◽  
Benjamin M. Neale ◽  
Mert R. Sabuncu ◽  
Jordan W. Smoller
Keyword(s):  
Complex Traits ◽  
Large Scale ◽  
Prediction Models ◽  
Population Characteristics ◽  
Uk Biobank ◽  
Multiple Traits ◽  
Common Genetic Variants ◽  
Heritability Estimation ◽  
Heritability Analysis ◽  
The Uk

Heritability estimation provides important information about the relative contribution of genetic and environmental factors to phenotypic variation, and provides an upper bound for the utility of genetic risk prediction models. Recent technological and statistical advances have enabled the estimation of additive heritability attributable to common genetic variants (SNP heritability) across a broad phenotypic spectrum. However, assessing the comparative heritability of multiple traits estimated in different cohorts may be misleading due to the population-specific nature of heritability. Here we report the SNP heritability for 551 complex traits derived from the large-scale, population-based UK Biobank, comprising both quantitative phenotypes and disease codes, and examine the moderating effect of three major demographic variables (age, sex and socioeconomic status) on the heritability estimates. Our study represents the first comprehensive phenome-wide heritability analysis in the UK Biobank, and underscores the importance of considering population characteristics in comparing and interpreting heritability.

scholarly journals Exploiting collider bias to apply two-sample summary data Mendelian randomization methods to one-sample individual level data

2020 ◽  
Author(s):  
Ciarrah Barry ◽  
Junxi Liu ◽  
Rebecca Richmond ◽  
Martin K Rutter ◽  
Deborah A Lawlor ◽  
...  
Keyword(s):  
Mendelian Randomization ◽  
Association Studies ◽  
General Procedure ◽  
Meta Analysis ◽  
Genome Wide Association Studies ◽  
Uk Biobank ◽  
Individual Level ◽  
Level Data ◽  
Summary Data ◽  
Collider Bias

AbstractOver the last decade the availability of SNP-trait associations from genome-wide association studies data has led to an array of methods for performing Mendelian randomization studies using only summary statistics. A common feature of these methods, besides their intuitive simplicity, is the ability to combine data from several sources, incorporate multiple variants and account for biases due to weak instruments and pleiotropy. With the advent of large and accessible fully-genotyped cohorts such as UK Biobank, there is now increasing interest in understanding how best to apply these well developed summary data methods to individual level data, and to explore the use of more sophisticated causal methods allowing for non-linearity and effect modification.In this paper we describe a general procedure for optimally applying any two sample summary data method using one sample data. Our procedure first performs a meta-analysis of summary data estimates that are intentionally contaminated by collider bias between the genetic instruments and unmeasured confounders, due to conditioning on the observed exposure. A weighted sum of these estimates is then used to correct the standard observational association between an exposure and outcome. Simulations are conducted to demonstrate the method’s performance against naive applications of two sample summary data MR. We apply the approach to the UK Biobank cohort to investigate the causal role of sleep disturbance on HbA1c levels, an important determinant of diabetes.Our approach is closely related to the work of Dudbridge et al. (Nat. Comm. 10: 1561), who developed a technique to adjust for index event bias when uncovering genetic predictors of disease progression based on case-only data. Our paper serves to clarify that in any one sample MR analysis, it can be advantageous to estimate causal relationships by artificially inducing and then correcting for collider bias.

scholarly journals Machine learning approach to dynamic risk modeling of mortality in COVID-19: a UK Biobank study

2021 ◽  
Author(s):  
Mohammad A. Dabbah ◽  
Angus B. Reed ◽  
Adam T.C. Booth ◽  
Arrash Yassaee ◽  
Alex Despotovic ◽  
...  
Keyword(s):  
High Risk ◽  
Mortality Risk ◽  
Prediction Models ◽  
Vital Signs ◽  
Classification Model ◽  
Risk Modeling ◽  
Uk Biobank ◽  
Individual Level ◽  
Random Forest Classification ◽  
The Uk

Abstract The COVID-19 pandemic has created an urgent need for robust, scalable monitoring tools supporting stratification of high-risk patients. This research aims to develop and validate prediction models, using the UK Biobank, to estimate COVID-19 mortality risk in confirmed cases. From the 11,245 participants testing positive for COVID-19, we develop a data-driven random forest classification model with excellent performance (AUC: 0.91), using baseline characteristics, pre-existing conditions, symptoms, and vital signs, such that the score could dynamically assess mortality risk with disease deterioration. We also identify several significant novel predictors of COVID-19 mortality with equivalent or greater predictive value than established high-risk comorbidities, such as detailed anthropometrics and prior acute kidney failure, urinary tract infection, and pneumonias. The model design and feature selection enables utility in outpatient settings. Possible applications include supporting individual-level risk profiling and monitoring disease progression across patients with COVID-19 at-scale, especially in hospital-at-home settings.

scholarly journals Exploiting collider bias to apply two-sample summary data Mendelian randomization methods to one-sample individual level data

PLoS Genetics ◽  
2021 ◽  
Vol 17 (8) ◽  
pp. e1009703
Author(s):  
Ciarrah Barry ◽  
Junxi Liu ◽  
Rebecca Richmond ◽  
Martin K. Rutter ◽  
Deborah A. Lawlor ◽  
...  
Keyword(s):  
Mendelian Randomization ◽  
Association Studies ◽  
General Procedure ◽  
Meta Analysis ◽  
Genome Wide Association Studies ◽  
Uk Biobank ◽  
Individual Level ◽  
Level Data ◽  
Summary Data ◽  
Collider Bias

Over the last decade the availability of SNP-trait associations from genome-wide association studies has led to an array of methods for performing Mendelian randomization studies using only summary statistics. A common feature of these methods, besides their intuitive simplicity, is the ability to combine data from several sources, incorporate multiple variants and account for biases due to weak instruments and pleiotropy. With the advent of large and accessible fully-genotyped cohorts such as UK Biobank, there is now increasing interest in understanding how best to apply these well developed summary data methods to individual level data, and to explore the use of more sophisticated causal methods allowing for non-linearity and effect modification. In this paper we describe a general procedure for optimally applying any two sample summary data method using one sample data. Our procedure first performs a meta-analysis of summary data estimates that are intentionally contaminated by collider bias between the genetic instruments and unmeasured confounders, due to conditioning on the observed exposure. These estimates are then used to correct the standard observational association between an exposure and outcome. Simulations are conducted to demonstrate the method’s performance against naive applications of two sample summary data MR. We apply the approach to the UK Biobank cohort to investigate the causal role of sleep disturbance on HbA1c levels, an important determinant of diabetes. Our approach can be viewed as a generalization of Dudbridge et al. (Nat. Comm. 10: 1561), who developed a technique to adjust for index event bias when uncovering genetic predictors of disease progression based on case-only data. Our work serves to clarify that in any one sample MR analysis, it can be advantageous to estimate causal relationships by artificially inducing and then correcting for collider bias.

scholarly journals Development and Validation of a Multivariable Prediction Model for Missed HIV Healthcare Provider Visits in a Large United States Clinical Cohort

2021 ◽  
Author(s):  
April C Pettit ◽  
Aihua Bian ◽  
Cassandra O Schember ◽  
Peter F Rebeiro ◽  
Jeanne C Keruly ◽  
...  
Keyword(s):  
Prediction Models ◽  
Community Level ◽  
Research Network ◽  
Hiv Care ◽  
People Living With Hiv ◽  
Individual Level ◽  
Visit Adherence ◽  
Level Data ◽  
Missed Visits ◽  
The Individual

Abstract Background Identifying individuals at high risk of missing HIV care provider visits could support proactive intervention. Previous prediction models for missed visits have not incorporated data beyond the individual-level. Methods We developed prediction models for missed visits among people living with HIV (PLWH) with ≥1 follow-up visit in the Center for AIDS Research Network of Integrated Clinical Systems from 2010-2016. Individual-level (medical record data and patient-reported outcomes), community-level (American Community Survey), HIV care site-level (standardized clinic leadership survey), and structural-level (HIV criminalization laws, Medicaid expansion, and state AIDS Drug Assistance Program budget) predictors were included. Models were developed using random forests with 10-fold cross-validation; candidate models with highest area under the curve (AUC) were identified. Results Data from 382,432 visits among 20,807 PLWH followed for a median of 3.8 years were included; median age was 44 years, 81% were male, 37% were Black, 15% reported injection drug use, and 57% reported male-to-male sexual contact. The highest AUC was 0.76 and strongest predictors were at the individual-level (prior visit adherence, age, CD4+ count) and community-level (proportion living in poverty, unemployed, and of Black race). A simplified model, including readily accessible variables available in a web-based calculator, had a slightly lower AUC of 0.700. Conclusions Prediction models validated using multi-level data had a similar AUC to previous models developed using only individual-level data. Strongest predictors were individual-level variables, particularly prior visit adherence, though community-level variables were also predictive. Absent additional data, PLWH with previous missed visits should be prioritized by interventions to improve visit adherence.

Sign in / Sign up

Export Citation Format

Share Document