Population Stratification In Association Studies Research Articles

Structure-informed clustering for population stratification in association studies

BackgroundIdentifying variants associated with complex traits is a challenging task in genetic association studies due to linkage disequilibrium (LD) between genetic variants and population stratification, unrelated to the disease risk. Existing methods of population structure correction use principal component analysis or linear mixed models with a random effect when modeling associations between a trait of interest and genetic markers. However, due to stringent significance thresholds and latent interactions between the markers, these methods often fail to detect genuinely associated variants.ResultsTo overcome this, we propose CluStrat, which corrects for complex arbitrarily structured populations while leveraging the linkage disequilibrium induced distances between genetic markers. It performs an agglomerative hierarchical clustering using the Mahalanobis distance covariance matrix of the markers. In simulation studies, we show that our method outperforms existing methods in detecting true causal variants. Applying CluStrat on WTCCC2 and UK Biobank cohorts, we found biologically relevant associations in Schizophrenia and Myocardial Infarction. CluStrat was also able to correct for population structure in polygenic adaptation of height in Europeans.ConclusionsCluStrat highlights the advantages of biologically relevant distance metrics, such as the Mahalanobis distance, which captures the cryptic interactions within populations in the presence of LD better than the Euclidean distance.

Population stratification correction using Bayesian shrinkage priors for genetic association studies.

Population stratification (PS) is a major source of confounding in population-based genetic association studies of quantitative traits. Principal component regression (PCR) and linear mixed model (LMM) are two commonly used approaches to account for PS in association studies. Previous studies have shown that LMM can be interpreted as including all principal components (PCs) as random-effect covariates. However, including all PCs in LMM may dilute the influence of relevant PCs in some scenarios, while including only a few preselected PCs in PCR may fail to fully capture the genetic diversity. To address these shortcomings, we introduce Bayestrat-a method to detect associated variants with PS correction under the Bayesian LASSO framework. To adjust for PS, Bayestrat accommodates a large number of PCs and utilizes appropriate shrinkage priors to shrink the effects of nonassociated PCs. Simulation results show that Bayestrat consistently controls type I error rates and achieves higher power compared to its non-shrinkage counterparts, especially when the number of PCs included in the model is large. As a demonstration of the utility of Bayestrat, we apply it to the Multi-Ethnic Study of Atherosclerosis (MESA). Variants and genes associated with serum triglyceride or HDL cholesterol are identified in our analyses. The automatic and self-selection features of Bayestrat make it particularly suited in situations with complex underlying PS scenarios, where it is unknown a priori which PCs are potential confounders, yet the number that needs to be considered could be large in order to fully account for PS.

Identification of ancestry proportions in admixed groups across the Americas using clinical pharmacogenomic SNP panels

We evaluated the performance of three PGx panels to estimate biogeographical ancestry: the DMET panel, and the VIP and Preemptive PGx panels described in the literature. Our analysis indicate that the three panels capture quite well the individual variation in admixture proportions observed in recently admixed populations throughout the Americas, with the Preemptive PGx and DMET panels performing better than the VIP panel. We show that these panels provide reliable information about biogeographic ancestry and can be used to guide the implementation of PGx clinical decision-support (CDS) tools. We also report that using these panels it is possible to control for the effects of population stratification in association studies in recently admixed populations, as exemplified with a warfarin dosing GWA study in a sample from Brazil.

Comparing the Utility of Mitochondrial and Nuclear DNA to Adjust for Genetic Ancestry in Association Studies.

Mitochondrial genome-wide association studies identify mitochondrial single nucleotide polymorphisms (mtSNPs) that associate with disease or disease-related phenotypes. Most mitochondrial and nuclear genome-wide association studies adjust for genetic ancestry by including principal components derived from nuclear DNA, but not from mitochondrial DNA, as covariates in statistical regression analyses. Furthermore, there is no standard when controlling for genetic ancestry during mitochondrial and nuclear genetic interaction association scans, especially across ethnicities with substantial mitochondrial genetic heterogeneity. The purpose of this study is to (1) compare the degree of ethnic variation captured by principal components calculated from microarray-defined nuclear and mitochondrial DNA and (2) assess the utility of mitochondrial principal components for association studies. Analytic techniques used in this study include a principal component analysis for genetic ancestry, decision-tree classification for self-reported ethnicity, and linear regression for association tests. Data from the Health and Retirement Study, which includes self-reported White, Black, and Hispanic Americans, was used for all analyses. We report that (1) mitochondrial principal component analysis (PCA) captures ethnic variation to a similar or slightly greater degree than nuclear PCA in Blacks and Hispanics, (2) nuclear and mitochondrial DNA classify self-reported ethnicity to a high degree but with a similar level of error, and 3) mitochondrial principal components can be used as covariates to adjust for population stratification in association studies with complex traits, as demonstrated by our analysis of height—a phenotype with a high heritability. Overall, genetic association studies might reveal true and robust mtSNP associations when including mitochondrial principal components as regression covariates.

A panel of 32 AIMs suitable for population stratification correction and global ancestry estimation in Mexican mestizos

BackgroundAssociation studies are useful to unravel the genetic basis of common human diseases. However, the presence of undetected population structure can lead to both false positive results and failures to detect genuine associations. Even when most of the approaches to deal with population stratification require genome-wide data, the use of a well-selected panel of ancestry informative markers (AIMs) may appropriately correct for population stratification. Few panels of AIMs have been developed for Latino populations and most contain a high number of markers (> 100 AIMs). For some association studies such as candidate gene approaches, it may be unfeasible to genotype a numerous set of markers to avoid false positive results. In such cases, methods that use fewer AIMs may be appropriate.ResultsWe validated an accurate and cost-effective panel of AIMs, for use in population stratification correction of association studies and global ancestry estimation in Mexicans, as well as in populations having large proportions of both European and Native American ancestries.Based on genome-wide data from 1953 Mexican individuals, we performed a PCA and SNP weights were calculated to select subsets of unlinked AIMs within percentiles 0.10 and 0.90, ensuring that all chromosomes were represented. Correlations between PC1 calculated using genome-wide data versus each subset of AIMs (16, 32, 48 and 64) were r2 = 0.923, 0.959, 0.972 and 0.978, respectively. When evaluating PCs performance as population stratification adjustment covariates, no correlation was found between P values obtained from uncorrected and genome-wide corrected association analyses (r2 = 0.141), highlighting that population stratification correction is compulsory for association analyses in admixed populations. In contrast, high correlations were found when adjusting for both PC1 and PC2 for either subset of AIMs (r2 > 0.900). After multiple validations, including an independent sample, we selected a minimal panel of 32 AIMs, which are highly informative of the major ancestral components of Mexican mestizos, namely European and Native American ancestries. Finally, the correlation between the global ancestry proportions calculated using genome-wide data and our panel of 32 AIMs was r2 = 0.972.ConclusionsOur panel of 32 AIMs accurately estimated global ancestry and corrected for population stratification in association studies in Mexican individuals.

Cuba: exploring the history of admixture and the genetic basis of pigmentation using autosomal and uniparental markers.

We carried out an admixture analysis of a sample comprising 1,019 individuals from all the provinces of Cuba. We used a panel of 128 autosomal Ancestry Informative Markers (AIMs) to estimate the admixture proportions. We also characterized a number of haplogroup diagnostic markers in the mtDNA and Y-chromosome in order to evaluate admixture using uniparental markers. Finally, we analyzed the association of 16 single nucleotide polymorphisms (SNPs) with quantitative estimates of skin pigmentation. In the total sample, the average European, African and Native American contributions as estimated from autosomal AIMs were 72%, 20% and 8%, respectively. The Eastern provinces of Cuba showed relatively higher African and Native American contributions than the Western provinces. In particular, the highest proportion of African ancestry was observed in the provinces of Guantánamo (40%) and Santiago de Cuba (39%), and the highest proportion of Native American ancestry in Granma (15%), Holguín (12%) and Las Tunas (12%). We found evidence of substantial population stratification in the current Cuban population, emphasizing the need to control for the effects of population stratification in association studies including individuals from Cuba. The results of the analyses of uniparental markers were concordant with those observed in the autosomes. These geographic patterns in admixture proportions are fully consistent with historical and archaeological information. Additionally, we identified a sex-biased pattern in the process of gene flow, with a substantially higher European contribution from the paternal side, and higher Native American and African contributions from the maternal side. This sex-biased contribution was particularly evident for Native American ancestry. Finally, we observed that SNPs located in the genes SLC24A5 and SLC45A2 are strongly associated with melanin levels in the sample.

Correcting for differential genotyping error in genetic association analysis

Differential genotype error in case-control association studies occurs when cases and controls are genotyped under different conditions. Existence of differential errors can considerably bias the association test, resulting in inflation of type I error and spurious significance. With the availability of high-throughput genotyping technologies such as the SNPchip, null markers that are unlinked with the disease can be used to correct for the bias caused by differential errors. A similar method, known as the genomic control, had been used to correct for population stratification in association studies. In this paper, we show that the same idea can be used to correct for the bias caused by differential errors, under the assumption that the null markers and the candidate marker are subject to the same or similar genotyping error model. The variance inflation is shown to be minor and the bias in the association test is the major source of type I error inflation in the presence of differential errors. Our method centralizes the test statistic by deducting the bias estimated from null markers through a quadratic regression method, which adjusts for the variability of null marker allele frequencies. Simulation results show that the proposed method performs very well in correcting for the type I error inflations.

Improved ancestry inference using weights from external reference panels

Inference of ancestry using genetic data is motivated by applications in genetic association studies, population genetics and personal genomics. Here, we provide methods and software for improved ancestry inference using genome-wide single nucleotide polymorphism (SNP) weights from external reference panels. This approach makes it possible to leverage the rich ancestry information that is available from large external reference panels, without the administrative and computational complexities of re-analyzing the raw genotype data from the reference panel in subsequent studies. We extensively validate our approach in multiple African American, Latino American and European American datasets, making use of genome-wide SNP weights derived from large reference panels, including HapMap 3 populations and 6546 European Americans from the Framingham Heart Study. We show empirically that our approach provides much greater accuracy than either the prevailing ancestry-informative marker (AIM) approach or the analysis of genome-wide target genotypes without a reference panel. For example, in an independent set of 1636 European American genome-wide association study samples, we attained prediction accuracy (R(2)) of 1.000 and 0.994 for the first two principal components using our method, compared with 0.418 and 0.407 using 150 published AIMs or 0.955 and 0.003 by applying principal component analysis directly to the target samples. We finally show that the higher accuracy in inferring ancestry using our method leads to more effective correction for population stratification in association studies. The SNPweights software is available online at http://www.hsph.harvard.edu/faculty/alkes-price/software/. Supplementary data are available at Bioinformatics online.

SHIPS: Spectral Hierarchical Clustering for the Inference of Population Structure in Genetic Studies

Inferring the structure of populations has many applications for genetic research. In addition to providing information for evolutionary studies, it can be used to account for the bias induced by population stratification in association studies. To this end, many algorithms have been proposed to cluster individuals into genetically homogeneous sub-populations. The parametric algorithms, such as Structure, are very popular but their underlying complexity and their high computational cost led to the development of faster parametric alternatives such as Admixture. Alternatives to these methods are the non-parametric approaches. Among this category, AWclust has proven efficient but fails to properly identify population structure for complex datasets. We present in this article a new clustering algorithm called Spectral Hierarchical clustering for the Inference of Population Structure (SHIPS), based on a divisive hierarchical clustering strategy, allowing a progressive investigation of population structure. This method takes genetic data as input to cluster individuals into homogeneous sub-populations and with the use of the gap statistic estimates the optimal number of such sub-populations. SHIPS was applied to a set of simulated discrete and admixed datasets and to real SNP datasets, that are data from the HapMap and Pan-Asian SNP consortium. The programs Structure, Admixture, AWclust and PCAclust were also investigated in a comparison study. SHIPS and the parametric approach Structure were the most accurate when applied to simulated datasets both in terms of individual assignments and estimation of the correct number of clusters. The analysis of the results on the real datasets highlighted that the clusterings of SHIPS were the more consistent with the population labels or those produced by the Admixture program. The performances of SHIPS when applied to SNP data, along with its relatively low computational cost and its ease of use make this method a promising solution to infer fine-scale genetic patterns.

Assessing the impact of global versus local ancestry in association studies.

To account for population stratification in association studies, principal-components analysis is often performed on single-nucleotide polymorphisms (SNPs) across the genome. Here, we use Framingham Heart Study (FHS) Genetic Analysis Workshop 16 data to compare the performance of local ancestry adjustment for population stratification based on principal components (PCs) estimated from SNPs in a local chromosomal region with global ancestry adjustment based on PCs estimated from genome-wide SNPs. Standardized height residuals from unrelated adults from the FHS Offspring Cohort were averaged from longitudinal data. PCs of SNP genotype data were calculated to represent individual's ancestry either 1) globally using all SNPs across the genome or 2) locally using SNPs in adjacent 20-Mbp regions within each chromosome. We assessed the extent to which there were differences in association studies of height depending on whether PCs for global, local, or both global and local ancestry were included as covariates. The correlations between local and global PCs were low (r < 0.12), suggesting variability between local and global ancestry estimates. Genome-wide association tests without any ancestry adjustment demonstrated an inflated type I error rate that decreased with adjustment for local ancestry, global ancestry, or both. A known spurious association was replicated for SNPs within the lactase gene, and this false-positive association was abolished by adjustment with local or global ancestry PCs. Population stratification is a potential source of bias in this seemingly homogenous FHS population. However, local and global PCs derived from SNPs appear to provide adequate information about ancestry.

Recent advances of genetic ancestry testing in biomedical research and direct to consumer testing

In the post-Human Genome Project era, the debate on the concept of race/ethnicity and its implications for biomedical research are dependent on two critical issues: whether and how to classify individuals and whether biological factors play a role in health disparities. The advent of reliable estimates of genetic (or biogeographic) ancestry has provided this debate with a quantitative and more objective tool. The estimation of genetic ancestry allows investigators to control for population stratification in association studies and helps to detect biological causation behind population-specific differences in disease and drug response. New techniques such as admixture mapping can specifically detect population-specific risk alleles for a disease in admixed populations. However, researchers have to be mindful of the correlation between genetic ancestry and socioeconomic and environmental factors that could underlie these differences. More importantly, researchers must avoid the stigmatization of individuals based on perceived or real genetic risks. The latter point will become increasingly sensitive as several 'for profit companies' are offering ancestry and genetic testing directly to consumers and the consequences of the spread of the services of these companies are still unforeseeable.

Fast model-based estimation of ancestry in unrelated individuals

Population stratification has long been recognized as a confounding factor in genetic association studies. Estimated ancestries, derived from multi-locus genotype data, can be used to perform a statistical correction for population stratification. One popular technique for estimation of ancestry is the model-based approach embodied by the widely applied program structure. Another approach, implemented in the program EIGENSTRAT, relies on Principal Component Analysis rather than model-based estimation and does not directly deliver admixture fractions. EIGENSTRAT has gained in popularity in part owing to its remarkable speed in comparison to structure. We present a new algorithm and a program, ADMIXTURE, for model-based estimation of ancestry in unrelated individuals. ADMIXTURE adopts the likelihood model embedded in structure. However, ADMIXTURE runs considerably faster, solving problems in minutes that take structure hours. In many of our experiments, we have found that ADMIXTURE is almost as fast as EIGENSTRAT. The runtime improvements of ADMIXTURE rely on a fast block relaxation scheme using sequential quadratic programming for block updates, coupled with a novel quasi-Newton acceleration of convergence. Our algorithm also runs faster and with greater accuracy than the implementation of an Expectation-Maximization (EM) algorithm incorporated in the program FRAPPE. Our simulations show that ADMIXTURE's maximum likelihood estimates of the underlying admixture coefficients and ancestral allele frequencies are as accurate as structure's Bayesian estimates. On real-world data sets, ADMIXTURE's estimates are directly comparable to those from structure and EIGENSTRAT. Taken together, our results show that ADMIXTURE's computational speed opens up the possibility of using a much larger set of markers in model-based ancestry estimation and that its estimates are suitable for use in correcting for population stratification in association studies.

Inference of Population Structure Using Genetic Markers and a Bayesian Model Averaging Approach for Clustering

The analysis of the structure of populations on the basis of genetic data is essential in population genetics. It is used, for instance, to study the evolution of species or to correct for population stratification in association studies. These genetic data, normally based on DNA polymorphisms, may contain irrelevant information that biases the inference of population structure. In this paper we adapt a recently proposed algorithm, named multistart EMA, to be used in the inference of population structure. This algorithm is able to deal with irrelevant information when obtaining the (probabilistic) population partition. Additionally, we present a maker selection test able to obtain the most relevant markers to retrieve that population partition. The proposed algorithm is compared with the widely used STRUCTURE software on the basis of the F(ST) metric and the log-likelihood score. It is shown that the proposed algorithm improves the obtention of the population structure. Moreover, information about relevant markers obtained by the multi-start EMA can be used to improve the results obtained by other methods, correct for population stratification or even also reduce the economical cost of sequencing new samples. The software presented in this paper is available online at http://www.sc.ehu.es/ccwbayes/members/guzman.

Estimating Local Ancestry in Admixed Populations

Large-scale genotyping of SNPs has shown a great promise in identifying markers that could be linked to diseases. One of the major obstacles involved in performing these studies is that the underlying population substructure could produce spurious associations. Population substructure can be caused by the presence of two distinct subpopulations or a single pool of admixed individuals. In this work, we focus on the latter, which is significantly harder to detect in practice. New advances in this research direction are expected to play a key role in identifying loci that are different among different populations and are still associated with a disease. We evaluated current methods for inference of population substructure in such cases and show that they might be quite inaccurate even in relatively simple scenarios. We therefore introduce a new method, LAMP (Local Ancestry in adMixed Populations), which infers the ancestry of each individual at every single-nucleotide polymorphism (SNP). LAMP computes the ancestry structure for overlapping windows of contiguous SNPs and combines the results with a majority vote. Our empirical results show that LAMP is significantly more accurate and more efficient than existing methods for inferring locus-specific ancestries, enabling it to handle large-scale datasets. We further show that LAMP can be used to estimate the individual admixture of each individual. Our experimental evaluation indicates that this extension yields a considerably more accurate estimate of individual admixture than state-of-the-art methods such as STRUCTURE or EIGENSTRAT, which are frequently used for the correction of population stratification in association studies.

Detection and controlling for population stratification in association studies of human complex disease

Case-control studies, which serve as standard design for genetic association analysis, can be the most practical and powerful approach to detect genetic polymorphisms contributing to susceptibility to complex human diseases. However, considerable concern has been expressed that this approach is prone to population stratification, which can lead to biased or spurious results. We review several methods to detect and account for population stratification; these methods include nuclear family-based transmission/disequilibrium test (TDT), and genomic control (GC) and structured association (SA) based on unlinked genetic markers.

Use of Unlinked Genetic Markers to Detect Population Stratification in Association Studies

We examine the issue of population stratification in association-mapping studies. In case-control studies of association, population subdivision or recent admixture of populations can lead to spurious associations between a phenotype and unlinked candidate loci. Using a model of sampling from a structured population, we show that if population stratification exists, it can be detected by use of unlinked marker loci. We show that the case-control-study design, using unrelated control individuals, is a valid approach for association mapping, provided that marker loci unlinked to the candidate locus are included in the study, to test for stratification. We suggest guidelines as to the number of unlinked marker loci to use.

Population studies of polymorphisms of the serotonin transporter protein gene

The range of allele frequency variation in humans for any locus that may have functionally important genetic variation needs to be documented. Therefore, we tested two polymorphisms at the serotonin transporter protein locus (SLC6A4) in samples from seven specific populations from five continental regions. We studied the promoter polymorphism which is reported to have functional significance and to be associated with anxiety- and depression-related phenotypes [Lesch et al., 1996: Science 274:1527-1531], and the intron 2 VNTR polymorphism [Lesch et al., 1994: J Neural Transm 95:157-162]. Allele frequencies for both systems show significant global variation, and consequently so do haplotype frequencies. Linkage disequilibrium varied among the populations, being absent in some and highly significant in others. These differences further document a large potential for population stratification in association studies of either of these SLC6A4 polymorphisms.

Serotonin transporter protein (SLC6A4) allele and haplotype frequencies and linkage disequilibria in African- and European-American and Japanese populations and in alcohol-dependent subjects.

The SLC6A4 locus encodes the serotonin transporter, which in turn mediates the synaptic inactivation of the neurotransmitter serotonin. Two PCR-formatted polymorphisms at this locus have been described, the first of which is a variable number tandem repeat located in exon 2, and the second a repeat sequence polymorphism located in the promoter region. The latter polymorphism alters transcriptional activity of SLC6A4, and has been reported to be associated with anxiety and depression-related traits. We studied allele frequencies, and computed haplotype frequencies and linkage disequilibrium measures, for these two polymorphisms in European-American, African-American, and Japanese populations, and in a set of alcohol-dependent European-American subjects. Allele frequencies for both systems showed variation, with significant differences overall for each system, and significant differences between each pair of populations for both systems. Linkage disequilibrium also varied among the populations. There were no significant differences in allele or haplotype frequencies between the European-American population samples and alcohol-dependent subjects. The population differences demonstrate a potential for population stratification in association studies of either of these SLC6A4 polymorphisms. If genetic variation at this locus really is associated with behavioral variation, these results could reflect either different behavioral adaptations in different populations, or random genetic drift of a behaviorally important but selectively neutral polymorphism.