Search

Studying gene-environment (G × E) interactions is important, as they extend our knowledge of the genetic architecture of complex traits and may help to identify novel variants not detected via analysis of main effects alone. The main statistical framework for studying G × E interactions uses a single regression model that includes both the genetic main and G × E interaction effects (the "joint" framework). The alternative "stratified" framework combines results from genetic main-effect analyses carried out separately within the exposed and unexposed groups. Although there have been several investigations using theory and simulation, an empirical comparison of the two frameworks is lacking. Here, we compare the two frameworks using results from genome-wide association studies of systolic blood pressure for 3.2 million low frequency and 6.5 million common variants across 20 cohorts of European ancestry, comprising 79,731 individuals. Our cohorts have sample sizes ranging from 456 to 22,983 and include both family-based and population-based samples. In cohort-specific analyses, the two frameworks provided similar inference for population-based cohorts. The agreement was reduced for family-based cohorts. In meta-analyses, agreement between the two frameworks was less than that observed in cohort-specific analyses, despite the increased sample size. In meta-analyses, agreement depended on (1) the minor allele frequency, (2) inclusion of family-based cohorts in meta-analysis, and (3) filtering scheme. The stratified framework appears to approximate the joint framework well only for common variants in population-based cohorts. We conclude that the joint framework is the preferred approach and should be used to control false positives when dealing with low-frequency variants and/or family-based cohorts., (© 2016 WILEY PERIODICALS, INC.)

For complex traits, most associated single nucleotide variants (SNV) discovered to date have a small effect, and detection of association is only possible with large sample sizes. Because of patient confidentiality concerns, it is often not possible to pool genetic data from multiple cohorts, and meta-analysis has emerged as the method of choice to combine results from multiple studies. Many meta-analysis methods are available for single SNV analyses. As new approaches allow the capture of low frequency and rare genetic variation, it is of interest to jointly consider multiple variants to improve power. However, for the analysis of haplotypes formed by multiple SNVs, meta-analysis remains a challenge, because different haplotypes may be observed across studies. We propose a two-stage meta-analysis approach to combine haplotype analysis results. In the first stage, each cohort estimate haplotype effect sizes in a regression framework, accounting for relatedness among observations if appropriate. For the second stage, we use a multivariate generalized least square meta-analysis approach to combine haplotype effect estimates from multiple cohorts. Haplotype-specific association tests and a global test of independence between haplotypes and traits are obtained within our framework. We demonstrate through simulation studies that we control the type-I error rate, and our approach is more powerful than inverse variance weighted meta-analysis of single SNV analysis when haplotype effects are present. We replicate a published haplotype association between fasting glucose-associated locus (G6PC2) and fasting glucose in seven studies from the Cohorts for Heart and Aging Research in Genomic Epidemiology Consortium and we provide more precise haplotype effect estimates., (© 2016 WILEY PERIODICALS, INC.)

Along with the accumulated data of genetic variants and biomedical phenotypes in the genome era, statistical identification of pleiotropy is of growing interest for dissecting and understanding genetic correlations between complex traits. We proposed a novel method for estimating and testing pleiotropic effect of a genetic variant on two quantitative traits. Based on a covariance decomposition and estimation, our method quantifies pleiotropy as the portion of between-trait correlation explained by the same genetic variant. Unlike most multiple-trait methods that assess potential pleiotropy (i.e., whether a variant contributes to at least one trait), our method formulates a statistic that tests exact pleiotropy (i.e., whether a variant contributes to both of two traits). We developed two approaches (a regression approach and a bootstrapping approach) for such test and investigated their statistical properties, in comparison with other potential pleiotropy test methods. Our simulation shows that the regression approach produces correct P-values under both the complete null (i.e., a variant has no effect on both two traits) and the incomplete null (i.e., a variant has effect on only one of two traits), but requires large sample sizes to achieve a good power, when the bootstrapping approach has a better power and produces conservative P-values under the complete null. We demonstrate our method for detecting exact pleiotropy using a real GWAS dataset. Our method provides an easy-to-implement tool for measuring, testing, and understanding the pleiotropic effect of a single variant on the correlation architecture of two complex traits., (© 2014 WILEY PERIODICALS, INC.)

Introduction: Genetic discoveries are validated through the meta-analysis of genome-wide association scans in large international consortia. Because environmental variables may interact with genetic factors, investigation of differing genetic effects for distinct levels of an environmental exposure in these large consortia may yield additional susceptibility loci undetected by main effects analysis. We describe a method of joint meta-analysis (JMA) of SNP and SNP by Environment (SNP × E) regression coefficients for use in gene-environment interaction studies., Methods: In testing SNP × E interactions, one approach uses a two degree of freedom test to identify genetic variants that influence the trait of interest. This approach detects both main and interaction effects between the trait and the SNP. We propose a method to jointly meta-analyze the SNP and SNP × E coefficients using multivariate generalized least squares. This approach provides confidence intervals of the two estimates, a joint significance test for SNP and SNP × E terms, and a test of homogeneity across samples., Results: We present a simulation study comparing this method to four other methods of meta-analysis and demonstrate that the JMA performs better than the others when both main and interaction effects are present. Additionally, we implemented our methods in a meta-analysis of the association between SNPs from the type 2 diabetes-associated gene PPARG and log-transformed fasting insulin levels and interaction by body mass index in a combined sample of 19,466 individuals from five cohorts., (© 2010 Wiley-Liss, Inc.)

A general-purpose modeling framework for performing path and segregation analysis jointly, called SEGPATH (Province and Rao [1995] Stat. Med. 7:185-198), has been extended to cover "model-free" robust, variance-components linkage analysis, based on identity-by-descent (IBD) sharing. These extended models can be used to analyze linkage to a single marker or to perform multipoint linkage analysis, with a single phenotype or multivariate vector of phenotypes, in pedigrees. Within a single, consistent approach, SEGPATH models can perform segregation analysis, path analysis, linkage analysis, or combinations thereof. SEGPATH models can incorporate environmental or other measured covariate fixed effects (including measured genotypes), genotype-specific covariate effects, population heterogeneity models, repeated-measures models, longitudinal models, autoregressive models, developmental models, gene-by-environment interaction models, etc., with or without linkage components. The data analyzed can have any missing value structure (assumed missing at random), with entire individuals missing, or missing on one or more measurements. Corrections for ascertainment can be made on a vector of phenotypes and/or other measures. Because of the flexibility of the class of models, the SEGPATH approach can also be used in nongenetic applications where there is a hierarchical structure, such as longitudinal, repeated-measures, time series, or nested models. A variety of specific models are provided, as well as some comparisons with other linkage analysis models. Particular applications demonstrate the importance of correctly accounting for the extraneous sources of familial resemblance, as can be done easily with these SEGPATH models, so as to give added power to detect linkage as well as to protect against spuriously inferring linkage., (Copyright 2003 Wiley-Liss, Inc.)

The simulated data problem was designed via an interactive process by the Simulation Problem Organizing Committee and the selected data simulators. Based on discussions at the previous Genetic Analysis Workshop, many of the features of previous simulation problems, such as a complex disease, genome scan, and replication, were retained and in addition, a population genetics model was used to generate the simulated genes. We describe the process that was used to structure the problem and summarize the discussions about many of the scientific issues that were considered.

We discuss the Genetic Analysis Workshop 11 analyses of data from the Collaborative Study on the Genetics of Alcoholism from a methodological perspective, concentrating on approaches and issues relevant to linkage and association studies of complex human phenotypes. Genome screening by parametric linkage, nonparametric linkage, association, and combined linkage/association methods are discussed. Issues particular to complex disease include etiologic heterogeneity, multivariate phenotype modeling, and parent-of-origin effects. Other methodological topics discussed are new and enhanced methods, ascertainment, weighting of nonindependent sib pairs, and data cleaning and validation.

In cases where sibship data are collected for a quantitative trait locus (QTL) linkage study without access to parental genotypes, the proportion of genes shared identical by descent must be estimated using the marker allele frequencies. No systematic study has been conducted to date to evaluate the effect of misspecification of these frequencies on a test of quantitative trait linkage. Analysis of both simulated and actual data on quantitative traits was carried out under various sets of allele frequency estimates. While correctly specifying the allele frequency distribution led to a slightly more powerful test and higher lod scores, the differences were small and would not likely alter the conclusion of a study. These results suggest that, at least for QTL analysis, there is a great deal of tolerance for misspecifying marker allele frequencies with little, if any, appreciable effect on the linkage test. However, the observed variations may be sufficiently large to alter the priority on might give to a positive finding for follow up.

We investigated the power of the affected sibpair method for detecting a disease locus when the disease is inherited through two bi-allelic loci. The power was computed for all possible values of the gene frequencies and penetrances that lead to a given population prevalence and a given sibling relative risk. A method to generate rapidly all possible models that give a specific population prevalence and relative risk is provided. We applied it to the case of a two-locus disease with a prevalence of 10% and a low sibling relative risk of 1.5. For this particular example, regardless of the true underlying model, a sample size (N = 450 for alpha = 0.05, N = 1,500 for alpha = 0.0001) may be determined such that one would expect enough power (0.80) to detect at least one of the two disease genes. In addition to the general case, we examined a special class of models in which the marginal penetrances at each locus are either recessive or dominant. In this instance, the gene frequencies were excellent predictors of the power afforded by a particular sample size. These methods have been implemented in a C program called SIBPOWER which is freely available from the first author. With this program, investigators can perform their own power calculations for any two-locus model of their choice thus avoiding the need to use single-locus approximations that may grossly underestimate the necessary sample size.

An assumption-free algorithm has been applied to compute the probabilities of all possible identity-by-descent (IBD) configurations when the parents have not been genotyped. These probabilities can be used to determine the amount of information that a particular sibship will bring to a nonparametric linkage analysis. It is shown that the number of possible configurations can be extremely large even when the markers are closely spaced and are rather polymorphic. Further, it is not always possible to reduce this number to a manageable one simply by consideration of the relative probabilities of the configurations.

We compare the results of genomic scans conducted with the Haseman-Elston sib-pair method using either (1) the average marker information from several adjacent loci or (2) each marker individually. Under smoothing, the squared sib-pair trait difference is regressed on the average number of alleles shared identical by descent averaged at several adjacent loci. This results in a significant decrease in the number of false-positives when compared to the individual marker approach. Linkage of Q4 to MG4 was found only with smoothing but not the individual marker approach. Overall, smoothing resulted in the loss of two true linkages.

Fulker and Cardon's interval mapping extension of Haseman and Elston's sib-pair linkage method was used to map loci affecting the quantitative phenotypes presented as part of Problem 2, both adjusted and not adjusted for covariates: Q1 was adjusted for age and the environmental factor (EF); Q2 and Q3 for EF; and Q4 for age, sex, and EF. Adjusted Q2 and Q4 were also log-transformed. The effect of candidate locus C5 (D5G28) on Q1 was detected by a test of association--apparently, allele 1 of C5 is protective (leading to lower values of Q1), allele 2 has no effect, and allele 3 contributes to elevated levels of Q1. C5 accounted for 5.2% of the variation in Q1; it was included as an additional concomitant in the adjustment procedure. Analysis of the correlational structure among the variables revealed that Q4 was not associated with either affected status or Q1 after controlling for the effects of age, and we concluded that Q4 probably does not itself play a role in the etiology of the disease. Mapping studies using a significance level of 0.05 lead to the detection of all the genes, but also resulted in a high frequency of false positive results. On the other hand, using a 0.0005 significance level resulted in the detection of D2G10-11 for both Q1 and Q3, and D1G2 was detected for Q2. One false positive was detected using this significance level and the effects of D1G2 on Q1 and D5G22-23 on Q4 were missed. There was no systematic effect of adjustment for covariates on the detection of loci, although in general, analysis of adjusted phenotypes yielded substantially higher rates of false positives. Finally, this mapping approach correctly located D1G2 and D2G10-11 for Q1 using the nonadjusted phenotype, and D1G10-11 for Q3 using the adjusted phenotype. The maximum difference between the estimated map location from the true location was 1.5 cM. It would be important to estimate the error interval around these inferred locations in order to assess the utility of this method for fine mapping over small (e.g,. 2 cM) intervals.

The power to detect major gene effects by rejection of the "no major gene" null hypothesis against a mixed model alternative was determined in random samples of nuclear families over a variety of conditions. Benchmarks have been developed using a varying number of families whose structure includes both parents and three children. Phenotypes were simulated assuming a Mendelian major gene under either recessive or dominant inheritance, with 0-30% residual polygenic heritability. Three trait prevalences--5, 10, and 20%--were considered in combination with increasing displacement between homozygous means, spanning a range of 14 to 36% of the phenotypic variance attributable to the major gene effect. All other assumptions of the traditional mixed model were adopted in the generating models. Segregation analysis was carried out on the simulated data sets and the proportion of samples out of 200 replications in which the null hypothesis q = 0 was rejected is reported as the power. Thus, failure to detect a major gene effect in this context is solely due to sampling variation, since no other perturbations were introduced. In general, there appears to be greater power to detect dominant major gene effects as opposed to recessive ones using otherwise comparable parameter values, and the effect of varying sibship size under dominant models appears to be greater as well. The use of joint vs. conditional likelihood calculations also was evaluated: substantial drops in power were observed when using conditional likelihoods under recessive inheritance, while the differences in power appeared to be nominal under dominant inheritance. The results of this investigation are offered as a guide to assist in the design of family studies whose aim is to detect major gene effects.

A single 51-member kindred, ascertained on the basis of a normotriglyceridemic proband with depressed high-density lipoprotein cholesterol (HDL-C) and myocardial infarctions at ages 40 and 42, was studied with respect to quantitative variation in HDL-C and apolipoprotein (apo) AI and AII levels. The results of bivariate segregation analysis suggest that the etiology of depressed HDL-C involves one or possibly two major loci: one has a pleiotropic effect on apo AI and apo AII levels and, possibly another one that affects apo AI levels. Both the major loci were characterized as having a dominant allele leading to depression of the respective trait(s). In addition, analysis of the cosegregation of HDL-C and apo AI levels gave evidence of residual nonfamilial factors common to both traits, leading to a positive covariance between them. This could reflect the role of apo AI in the transformation of nascent HDL-C particles into mature ones via its cofactor activity to lecithin cholesterol acyltransferase. The proposed two-locus model represents one possible etiology for the heterogeneous disorder of hypoalphalipoproteinemia. This analysis of a single pedigree does not completely define the genetic mechanism, but it does illustrate a useful new analytic approach.

Complex segregation analysis under the unified mixed model of inheritance (major gene and multifactorial) is performed on families ascertained through 23 probands with hypoalphalipoproteinemia (depressed HDL-cholesterol, denoted HDL-c). Evidence for segregation of a recessive major gene for depressed HDL-c with frequency q = 0.116, in addition to multifactorial transmission (H = 0.572), is found in these families. Reanalysis of a subset of families with severely depressed HDL-c confirms the conclusions based on the original analysis, except that different definitions of "affection" give rise to different estimates of gene frequency. Our finding of a recessive mode of inheritance differs from previous claims for a dominant gene because previous analyses did not use a mixed model for segregation analysis of hypoalphalipoproteinemia. When the significant multifactorial background is neglected, we also find evidence for the invalid claim of a dominant gene. This demonstrates the necessity of using mixed models for determining the mode of inheritance of a given phenotype.

Atopic disease is generally recognized to be familial, although specific genetic components have yet to be identified. High levels of a unique class of immunoglobulins, immunoglobulin E (IgE), have been shown to be associated with allergies. Several investigators have reported evidence indicating a recessive regulatory locus where an individual with the homozygous recessive genotype has persistently elevated levels of IgE. Willcox and Marsh [1978] have proposed a hypothesis relating IgE production and liability to become allergic. A test of this hypothesis was carried out in the present study. Bivariate segregation analysis of IgE levels and allergy was performed on 173 nuclear families, and the results indicate that an IgE regulatory locus contributes to the familial transmission of allergy. The results are further discussed in the context of the Willcox and Marsh hypothesis.