Effect of non-normality and low count variants on cross-phenotype association tests in GWAS

A novel application of data-consistent inversion to overcome spurious inference in genome-wide association studies

The genome-wide association studies (GWAS) typically use linear or logistic regression models to identify associations between phenotypes (traits) and genotypes (genetic variants) of interest. However, the use of regression with the additive assumption has potential limitations. First, the normality assumption of residuals is the one that is rarely seen in practice, and deviation from normality increases the Type-I error rate. Second, building a model based on such an assumption ignores genetic structures, like, dominant, recessive, and protective-risk cases. Ignoring genetic variants may result in spurious conclusions about the associations between a variant and a trait. We propose an assumption-free model built upon data-consistent inversion (DCI), which is a recently developed measure-theoretic framework utilized for uncertainty quantification. This proposed DCI-derived model builds a nonparametric distribution on model inputs that propagates to the distribution of observed data without the required normality assumption of residuals in the regression model. This characteristic enables the proposed DCI-derived model to cover all genetic variants without emphasizing on additivity of the classic-GWAS model. Simulations and a replication GWAS with data from the COPDGene demonstrate the ability of this model to control the Type-I error rate at least as well as the classic-GWAS (additive linear model) approach while having similar or greater power to discover variants in different genetic modes of transmission.

Characterizing Common and Rare Variations in Nontraditional Glycemic Biomarkers Using Multivariate Approaches on Multiancestry ARIC Study

Genetic studies of nontraditional glycemic biomarkers, glycated albumin and fructosamine, can shed light on unknown aspects of type 2 diabetes genetics and biology. We performed a multiphenotype genome-wide association study of glycated albumin and fructosamine from 7,395 White and 2,016 Black participants in the Atherosclerosis Risk in Communities (ARIC) study on common variants from genotyped/imputed data. We discovered two genome-wide significant loci, one mapping to a known type 2 diabetes gene (ARAP1/STARD10) and another mapping to a novel region (UGT1A complex of genes), using multiomics gene-mapping strategies in diabetes-relevant tissues. We identified additional loci that were ancestry- and sex-specific (e.g., PRKCA in African ancestry, FCGRT in European ancestry, TEX29 in males). Further, we implemented multiphenotype gene-burden tests on whole-exome sequence data from 6,590 White and 2,309 Black ARIC participants. Ten variant sets annotated to genes across different variant aggregation strategies were exome-wide significant only in multiancestry analysis, of which CD1D, EGFL7/AGPAT2, and MIR126 had notable enrichment of rare predicted loss of function variants in African ancestry despite smaller sample sizes. Overall, 8 of 14 discovered loci and genes were implicated to influence these biomarkers via glycemic pathways, and most of them were not previously implicated in studies of type 2 diabetes. This study illustrates improved locus discovery and potential effector gene discovery by leveraging joint patterns of related biomarkers across the entire allele frequency spectrum in multiancestry analysis. Future investigation of the loci and genes potentially acting through glycemic pathways may help us better understand the risk of developing type 2 diabetes.

ARTICLE HIGHLIGHTS

A fast non-parametric test of association for multiple traits

The increasing availability of multidimensional phenotypic data in large cohorts of genotyped individuals requires efficient methods to identify genetic effects on multiple traits. Permutational multivariate analysis of variance (PERMANOVA) offers a powerful non-parametric approach. However, it relies on permutations to assess significance, which hinders the analysis of large datasets. Here, we derive the limiting null distribution of the PERMANOVA test statistic, providing a framework for the fast computation of asymptotic p values. Our asymptotic test presents controlled type I error and high power, often outperforming parametric approaches. We illustrate its applicability in the context of QTL mapping and GWAS.

Characterizing common and rare variations in non-traditional glycemic biomarkers using multivariate approaches on multi-ancestry ARIC study

Glycated hemoglobin, fasting glucose, glycated albumin, and fructosamine are biomarkers that reflect different aspects of the glycemic process. Genetic studies of these glycemic biomarkers can shed light on unknown aspects of type 2 diabetes genetics and biology. While there exists several GWAS of glycated hemoglobin and fasting glucose, very few GWAS have focused on glycated albumin or fructosamine. We performed a multi-phenotype GWAS of glycated albumin and fructosamine from 7,395 White and 2,016 Black participants in the Atherosclerosis Risk in Communities (ARIC) study on the common variants from genotyped/imputed data. We found 2 genome-wide significant loci, one mapping to known type 2 diabetes gene (ARAP1/STARD10, p = 2.8 × 10-8) and another mapping to a novel gene (UGT1A, p = 1.4 × 10-8) using multi-omics gene mapping strategies in diabetes-relevant tissues. We identified additional loci that were ancestry-specific (e.g., PRKCA from African ancestry individuals, p = 1.7 × 10-8) and sex-specific (TEX29 locus in males only, p = 3.0 × 10-8). Further, we implemented multi-phenotype gene-burden tests on whole-exome sequence data from 6,590 White and 2,309 Black ARIC participants. Eleven genes across different rare variant aggregation strategies were exome-wide significant only in multi-ancestry analysis. Four out of 11 genes had notable enrichment of rare predicted loss of function variants in African ancestry participants despite smaller sample size. Overall, 8 out of 15 loci/genes were implicated to influence these biomarkers via glycemic pathways. This study illustrates improved locus discovery and potential effector gene discovery by leveraging joint patterns of related biomarkers across entire allele frequency spectrum in multi-ancestry analyses. Most of the loci/genes we identified have not been previously implicated in studies of type 2 diabetes, and future investigation of the loci/genes potentially acting through glycemic pathways may help us better understand risk of developing type 2 diabetes.

Recent advances and challenges of rare variant association analysis in the biobank sequencing era

Causal variants for rare genetic diseases are often rare in the general population. Rare variants may also contribute to common complex traits and can have much larger per-allele effect sizes than common variants, although power to detect these associations can be limited. Sequencing costs have steadily declined with technological advancements, making it feasible to adopt whole-exome and whole-genome profiling for large biobank-scale sample sizes. These large amounts of sequencing data provide both opportunities and challenges for rare-variant association analysis. Herein, we review the basic concepts of rare-variant analysis methods, the current state-of-the-art methods in utilizing variant annotations or external controls to improve the statistical power, and particular challenges facing rare variant analysis such as accounting for population structure, extremely unbalanced case-control design. We also review recent advances and challenges in rare variant analysis for familial sequencing data and for more complex phenotypes such as survival data. Finally, we discuss other potential directions for further methodology investigation.

Revisiting the Wald Test in Small Case-Control Studies With a Skewed Covariate

The Wald test is routinely used in case-control studies to test for association between a covariate and disease. However, when the evidence for association is high, the Wald test tends to inflate small P values as a result of the Hauck-Donner effect (HDE). Here, we investigate the HDE in the context of genetic burden, both with and without additional covariates. First, we examine the burden-based P values in the absence of association using whole-exome sequence data from 1000 Genomes Project reference samples (n = 54) and selected preterm infants with neonatal complications (n = 74). Our careful analysis of the burden-based P values shows that the HDE is present and that the cause of the HDE in this setting is likely a natural extension of the well-known cause of the HDE in 2 × 2 contingency tables. Second, in a reanalysis of real data, we find that the permutation test provides increased power over the Wald, Firth, and likelihood ratio tests, which agrees with our intuition since the permutation test is valid for any sample size and since it does not suffer from the HDE. Therefore, we propose a powerful and computationally efficient permutation-based approach for the analysis and reanalysis of small case-control association studies.

Recent innovations and in-depth aspects of post-genome wide association study (Post-GWAS) to understand the genetic basis of complex phenotypes

In the past decade, the high throughput and low cost of sequencing/genotyping approaches have led to the accumulation of a large amount of data from genome-wide association studies (GWASs). The first aim of this review is to highlight how post-GWAS analysis can be used make sense of the obtained associations. Novel directions for integrating GWAS results with other resources, such as somatic mutation, metabolite-transcript, and transcriptomic data, are also discussed; these approaches can help us move beyond each individual data point and provide valuable information about complex trait genetics. In addition, cross-phenotype association tests, when the loci detected by GWASs have significant associations with multiple traits, are reviewed to provide biologically informative results for use in real-time applications. This review also discusses the challenges of identifying interactions between genetic mutations (epistasis) and mutations of loci affecting more than one trait (pleiotropy) as underlying causes of cross-phenotype associations; these challenges can be overcome using post-GWAS analysis. Genetic similarities between phenotypes that can be revealed using post-GWAS analysis are also discussed. In summary, different methodologies of post-GWAS analysis are now available, enhancing the value of information obtained from GWAS results, and facilitating application in both humans and nonhuman species. However, precise methods still need to be developed to overcome challenges in the field and uncover the genetic underpinnings of complex traits.

Comparison of adaptive multiple phenotype association tests using summary statistics in genome-wide association studies

Genome-wide association studies have been successful mapping loci for individual phenotypes, but few studies have comprehensively interrogated evidence of shared genetic effects across multiple phenotypes simultaneously. Statistical methods have been proposed for analyzing multiple phenotypes using summary statistics, which enables studies of shared genetic effects while avoiding challenges associated with individual-level data sharing. Adaptive tests have been developed to maintain power against multiple alternative hypotheses because the most powerful single-alternative test depends on the underlying structure of the associations between the multiple phenotypes and a single nucleotide polymorphism (SNP). Here we compare the performance of six such adaptive tests: two adaptive sum of powered scores (aSPU) tests, the unified score association test (metaUSAT), the adaptive test in a mixed-models framework (mixAda) and two principal-component-based adaptive tests (PCAQ and PCO). Our simulations highlight practical challenges that arise when multivariate distributions of phenotypes do not satisfy assumptions of multivariate normality. Previous reports in this context focus on low minor allele count (MAC) and omit the aSPU test, which relies less than other methods on asymptotic and distributional assumptions. When these assumptions are not satisfied, particularly when MAC is low and/or phenotype covariance matrices are singular or nearly singular, aSPU better preserves type I error, sometimes at the cost of decreased power. We illustrate this trade-off with multiple phenotype analyses of six quantitative electrocardiogram traits in the Population Architecture using Genomics and Epidemiology (PAGE) study.

Pleiotropy method reveals genetic overlap between orofacial clefts at multiple novel loci from GWAS of multi-ethnic trios

Based on epidemiologic and embryologic patterns, nonsyndromic orofacial clefts- the most common craniofacial birth defects in humans- are commonly categorized into cleft lip with or without cleft palate (CL/P) and cleft palate alone (CP), which are traditionally considered to be etiologically distinct. However, some evidence of shared genetic risk in IRF6, GRHL3 and ARHGAP29 regions exists; only FOXE1 has been recognized as significantly associated with both CL/P and CP in genome-wide association studies (GWAS). We used a new statistical approach, PLACO (pleiotropic analysis under composite null), on a combined multi-ethnic GWAS of 2,771 CL/P and 611 CP case-parent trios. At the genome-wide significance threshold of 5 × 10-8, PLACO identified 1 locus in 1q32.2 (IRF6) that appears to increase risk for one OFC subgroup but decrease risk for the other. At a suggestive significance threshold of 10-6, we found 5 more loci with compelling candidate genes having opposite effects on CL/P and CP: 1p36.13 (PAX7), 3q29 (DLG1), 4p13 (LIMCH1), 4q21.1 (SHROOM3) and 17q22 (NOG). Additionally, we replicated the recognized shared locus 9q22.33 (FOXE1), and identified 2 loci in 19p13.12 (RAB8A) and 20q12 (MAFB) that appear to influence risk of both CL/P and CP in the same direction. We found locus-specific effects may vary by racial/ethnic group at these regions of genetic overlap, and failed to find evidence of sex-specific differences. We confirmed shared etiology of the two OFC subtypes comprising CL/P, and additionally found suggestive evidence of differences in their pathogenesis at 2 loci of genetic overlap. Our novel findings include 6 new loci of genetic overlap between CL/P and CP; 3 new loci between pairwise OFC subtypes; and 4 loci not previously implicated in OFCs. Our in-silico validation showed PLACO is robust to subtype-specific effects, and can achieve massive power gains over existing approaches for identifying genetic overlap between disease subtypes. In summary, we found suggestive evidence for new genetic regions and confirmed some recognized OFC genes either exerting shared risk or with opposite effects on risk to OFC subtypes.

A powerful method for pleiotropic analysis under composite null hypothesis identifies novel shared loci between Type 2 Diabetes and Prostate Cancer

There is increasing evidence that pleiotropy, the association of multiple traits with the same genetic variants/loci, is a very common phenomenon. Cross-phenotype association tests are often used to jointly analyze multiple traits from a genome-wide association study (GWAS). The underlying methods, however, are often designed to test the global null hypothesis that there is no association of a genetic variant with any of the traits, the rejection of which does not implicate pleiotropy. In this article, we propose a new statistical approach, PLACO, for specifically detecting pleiotropic loci between two traits by considering an underlying composite null hypothesis that a variant is associated with none or only one of the traits. We propose testing the null hypothesis based on the product of the Z-statistics of the genetic variants across two studies and derive a null distribution of the test statistic in the form of a mixture distribution that allows for fractions of variants to be associated with none or only one of the traits. We borrow approaches from the statistical literature on mediation analysis that allow asymptotic approximation of the null distribution avoiding estimation of nuisance parameters related to mixture proportions and variance components. Simulation studies demonstrate that the proposed method can maintain type I error and can achieve major power gain over alternative simpler methods that are typically used for testing pleiotropy. PLACO allows correlation in summary statistics between studies that may arise due to sharing of controls between disease traits. Application of PLACO to publicly available summary data from two large case-control GWAS of Type 2 Diabetes and of Prostate Cancer implicated a number of novel shared genetic regions: 3q23 (ZBTB38), 6q25.3 (RGS17), 9p22.1 (HAUS6), 9p13.3 (UBAP2), 11p11.2 (RAPSN), 14q12 (AKAP6), 15q15 (KNL1) and 18q23 (ZNF236).

We propose a new approach PLACO that uses aggregate-level genotype-phenotype association statistics—commonly referred to as GWAS summary statistics—to identify genetic variants that influence risk of two traits or diseases. It allows correlation in summary statistics between studies that may arise due to sharing of controls between disease traits. We demonstrate that PLACO can achieve major power gain over alternative methods that are typically used. We applied PLACO to Type 2 Diabetes and Prostate Cancer summary data from two large case-control studies. Many previous studies have reported an inverse association of these two chronic diseases suggesting shared risk factors; however, shared genetic mechanisms underlying this association is poorly understood. PLACO identified a number of novel shared genetic regions that are not detected by individual trait analysis. Many of the loci implicated by PLACO increase risk for one disease while decreasing risk for the other. PLACO can similarly be used on other traits to shed light on shared genetic risk factors.