Evaluation of Nyholt’s Procedure for Multiple Testing Correction

Invited review: Good practices in genome-wide association studies to identify candidate sequence variants in dairy cattle

Genotype data from dairy cattle selection programs have greatly facilitated GWAS to identify variants related to economic traits. Results can enhance the accuracy of genomic prediction, analyze more complex models that go beyond additive effects, elucidate the genetic architecture of a trait, and finally, decipher the underlying biology of traits. The entire process, comprising data generation, quality control, statistical analyses, interpretation of association results, and linking results to biology should be designed and executed to minimize the generation of false-positive and false-negative associations and misleading links to biological processes. This review aims to provide general guidelines for data analysis that address data quality control, association tests, adjustment for population stratification, and significance evaluation to improve the reliability of conclusions. We also provide guidance on post-GWAS strategy and the interpretation of results. These guidelines are tailored to dairy cattle, which are characterized by long-range linkage disequilibrium, large half-sib families, and routinely collected phenotypes, requiring different approaches than those applied in human GWAS. We discuss common limitations and challenges that have been overlooked in the analysis and interpretation of GWAS to identify candidate sequence variants in dairy cattle.

Cardiorespiratory fitness, regular physical activity, and autonomic nervous system reactivity to laboratory and daily life stress

The cross-stressor adaptation hypothesis-which posits that adjustment to physical stress as a result of regular physical activity and its effects on fitness crosses over to psychological stress reactivity-has been around for over four decades. However, the literature has been plagued by heterogeneities preventing definitive conclusions. We address these heterogeneity issues in a combined laboratory and daily life study of 116 young adults (M = 22.48 SD = 3.56, 57.76% female). The exposure, i.e., the potential driver of adaptation, was defined in three ways. First, a submaximal test was performed to obtain aerobic fitness measured as the VO₂ max (kg/ml/min). Second, leisure time exercise behavior, and third, overall moderate-to-vigorous physical activity (MVPA), were obtained from a structured interview. Outcomes were autonomic nervous system (ANS) reactivity and affective responsiveness to stressors. ANS activity was measured continuously and expressed as inter-beat-interval (IBI), pre-ejection-period (PEP), respiratory sinus arrythmia (RSA), and non-specific Skin Conductance Responses (ns.SCR). Negative and positive affect were recorded after each experimental condition in the laboratory and hourly in daily life with a nine-item digital questionnaire. Linear regressions were performed between the three exposure measures as predictors and the various laboratory and daily life stress measurements as outcomes. Our results support the resting heart rate reducing effect of aerobic fitness and total MVPA in both the laboratory and daily life. We did not find evidence for the cross-stressor adaptation hypothesis, irrespective of ANS or affective outcome measure or whether the exposure was defined as exercise/MVPA or aerobic fitness.

GWAS significance thresholds for deep phenotyping studies can depend upon minor allele frequencies and sample size

An important issue affecting genome-wide association studies with deep phenotyping (multiple correlated phenotypes) is determining the suitable family-wise significance threshold. Straightforward family-wise correction (Bonferroni) of p < 0.05 for 4.3 million genotypes and 335 phenotypes would give a threshold of p < 3.46E-11. This would be too conservative because it assumes all tests are independent. The effective number of tests, both phenotypic and genotypic, must be adjusted for the correlations between them. Spectral decomposition of the phenotype matrix and LD-based correction of the number of tested SNPs are currently used to determine an effective number of tests. In this paper, we compare these calculated estimates with permutation-determined family-wise significance thresholds. Permutations are performed by shuffling individual IDs of the genotype vector for this dataset, to preserve correlation of phenotypes. Our results demonstrate that the permutation threshold is influenced by minor allele frequency (MAF) of the SNPs, and by the number of individuals tested. For the more common SNPs (MAF > 0.1), the permutation family-wise threshold was in close agreement with spectral decomposition methods. However, for less common SNPs (0.05 < MAF ≤ 0.1), the permutation threshold calculated over all SNPs was off by orders of magnitude. This applies to the number of individuals studied (here 777) but not to very much larger numbers. Based on these findings, we propose that the threshold to find a particular level of family-wise significance may need to be established using separate permutations of the actual data for several MAF bins.

Quick approximation of threshold values for genome-wide association studies

Standard normal statistics, chi-squared statistics, Student's t statistics and F statistics are used to map quantitative trait nucleotides for both small and large sample sizes. In genome-wide association studies (GWASs) of single-nucleotide polymorphisms (SNPs), the statistical distributions depend on both genetic effects and SNPs but are independent of SNPs under the null hypothesis of no genetic effects. Therefore, hypothesis testing when a nuisance parameter is present only under the alternative was introduced to quickly approximate the critical thresholds of these test statistics for GWASs. When only the statistical probabilities are available for high-throughput SNPs, the approximate critical thresholds can be estimated with chi-squared statistics, formulated by statistical probabilities with a degree of freedom of two. High similarities in the critical thresholds between the accurate and approximate estimations were demonstrated by extensive simulations and real data analysis.

Association between CUBN gene variants, type 2 diabetes and vitamin D concentrations in an elderly Greek population

Accumulating evidence suggests a potential implication of vitamin D biological network in the pathogenesis of diabetes mellitus. The megalin-cubilin endocytotic system constitutes a key transport structure, with a modulating role in vitamin D metabolism. We aimed to assess the contribution of variants in the CUBN gene to the genetic risk of Type 2 Diabetes Mellitus (T2DM). 95 polymorphisms within CUBN were genotyped in 716 patients with T2DM and 542 controls of Greek origin. Samples were analyzed on Illumina Human PsychArray. Permutation test analysis was implemented to determine statistical significance. Twenty-five-hydroxy-vitamin-D [25(OH)D)] levels were measured in a sub-group of participants (n = 276). Permutation analysis associated rs11254375_G/T (p_emp = 0.00049, OR = 1.482), rs6602175_G/T (p_emp = 0.016, OR = 0.822), rs1801224_G/T (p_emp = 0.025, OR = 0.830), rs4366393_A/G (p_emp = 0.028, OR = 0.829) and rs7071576_A/G (p_emp = 0.04, OR = 1.219) with T2DM. Mean 25(OH)D concentrations were significantly lower in patients with T2DM compared to controls (16.70 ± 6.69 ng/ml vs 18.51 ± 6.71 ng/ml, p < 0.001), although both groups were vitamin D deficient. In a further quantitative analysis, rs41301097 was strongly associated with higher 25(OH)D concentrations (p = 5.233e-6, beta = 15.95). Our results indicate a potential role of CUBN gene in T2DM genetic susceptibility in the Greek population. These findings may also denote an indirect effect of vitamin D metabolism dysregulation on the pathogenesis of T2DM. Further studies are required to replicate our findings and clarify the complex underlying mechanisms.

Genome-wide Significance Thresholds for Admixture Mapping Studies

Admixture mapping studies have become more common in recent years, due in part to technological advances and growing international efforts to increase the diversity of genetic studies. However, many open questions remain about appropriate implementation of admixture mapping studies, including how best to control for multiple testing, particularly in the presence of population structure. In this study, we develop a theoretical framework to characterize the correlation of local ancestry and admixture mapping test statistics in admixed populations with contributions from any number of ancestral populations and arbitrary population structure. Based on this framework, we develop an analytical approach for obtaining genome-wide significance thresholds for admixture mapping studies. We validate our approach via analysis of simulated traits with real genotype data for 8,064 unrelated African American and 3,425 Hispanic/Latina women from the Women's Health Initiative SNP Health Association Resource (WHI SHARe). In an application to these WHI SHARe data, our approach yields genome-wide significant p value thresholds of 2.1 × 10-5 and 4.5 × 10-6 for admixture mapping studies in the African American and Hispanic/Latina cohorts, respectively. Compared to other commonly used multiple testing correction procedures, our method is fast, easy to implement (using our publicly available R package), and controls the family-wise error rate even in structured populations. Importantly, we note that the appropriate admixture mapping significance threshold depends on the number of ancestral populations, generations since admixture, and population structure of the sample; as a result, significance thresholds are not, in general, transferable across studies.

COL4A2 is associated with lacunar ischemic stroke and deep ICH: Meta-analyses among 21,500 cases and 40,600 controls

OBJECTIVE

To determine whether common variants in familial cerebral small vessel disease (SVD) genes confer risk of sporadic cerebral SVD.

METHODS

We meta-analyzed genotype data from individuals of European ancestry to determine associations of common single nucleotide polymorphisms (SNPs) in 6 familial cerebral SVD genes (COL4A1, COL4A2, NOTCH3, HTRA1, TREX1, and CECR1) with intracerebral hemorrhage (ICH) (deep, lobar, all; 1,878 cases, 2,830 controls) and ischemic stroke (IS) (lacunar, cardioembolic, large vessel disease, all; 19,569 cases, 37,853 controls). We applied data quality filters and set statistical significance thresholds accounting for linkage disequilibrium and multiple testing.

RESULTS

A locus in COL4A2 was associated (significance threshold p < 3.5 × 10^-4) with both lacunar IS (lead SNP rs9515201: odds ratio [OR] 1.17, 95% confidence interval [CI] 1.11-1.24, p = 6.62 × 10^-8) and deep ICH (lead SNP rs4771674: OR 1.28, 95% CI 1.13-1.44, p = 5.76 × 10^-5). A SNP in HTRA1 was associated (significance threshold p < 5.5 × 10^-4) with lacunar IS (rs79043147: OR 1.23, 95% CI 1.10-1.37, p = 1.90 × 10^-4) and less robustly with deep ICH. There was no clear evidence for association of common variants in either COL4A2 or HTRA1 with non-SVD strokes or in any of the other genes with any stroke phenotype.

CONCLUSIONS

These results provide evidence of shared genetic determinants and suggest common pathophysiologic mechanisms of distinct ischemic and hemorrhagic cerebral SVD stroke phenotypes, offering new insights into the causal mechanisms of cerebral SVD.

Human Leukocyte Antigen Class I Genes Associated With Stevens-Johnson Syndrome and Severe Ocular Complications Following Use of Cold Medicine in a Brazilian Population

Importance

Describing the association with human leukocyte antigen (HLA) alleles could facilitate the understanding of increased risk factors for development of Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) in patients with severe ocular complications (SOCs).

Objective

To investigate the association between HLA class I genes and cold medicine (CM)-associated SJS/TEN with SOCs.

Design, Setting, and Participants

This case-control study was conducted between February 8, 2013, and August 29, 2014. Thirty-nine Brazilian patients with CM-SJS/TEN of 74 patients with SJS/TEN with SOCs and 133 healthy Brazilian volunteers were enrolled. Human leukocyte antigen class I genes (HLA-A, HLA-B, and HLA-C) were examined to determine whether there was a genetic predisposition for CM-SJS/TEN with SOC. Patients were interviewed to identify possible etiologic factors. Data analysis was performed from April 14, 2013, to August 29, 2014.

Main Outcomes and Measures

Genetic predisposition for CM-SJS/TEN with SOCs by analysis of HLA class I genes.

Results

Of 74 patients included in the analysis, 32 (43%) were male; mean (SD) age was 36.01 [15.42] years. HLA-A*66:01 (odds ratio [OR], 24.0; 95% CI, 2.79-206.0; P < .001), HLA-B*44:03 (OR, 2.71; 95% CI, 1.11-6.65; P = .04), and HLA-C*12:03 (OR, 5.6; 95% CI, 1.67-18.80; P = .006) were associated with Brazilian CM-SJS/TEN with SOCs, and HLA-A*11:01 (OR, 0.074; 95% CI, 0.004-1.26; P = .008), HLA-B*08:01 (OR, 0.15; 95% CI, 0.02-1.15; P = .048), and HLA-B*51:01 (OR, 0.23; 95% CI, 0.05-1.03; P = .045) were inversely associated with Brazilian CM-SJS/TEN with SOCs (39 cases: 19 Pardo and 16 European ancestry; 14 males and 25 females; age, 35.2 [14.4] years; and 133 controls: 66 Pardo and 61 European ancestry; 55 males and 78 females; age, 41.2 [12.9] years). When multiple test correction within the HLA locus, HLA-A*66:01 and HLA-C*12:03 demonstrated associations. When participants were segregated into Pardo and locus is considered, HLA-A*66:01 was associated with CM-SJS/TEN with SOC among individuals of both ethnic groups (Pardo: OR, 12.2; 95% CI, 1.19-125.0; P = .03; and European: OR, 21.2; 95% CI, 0.97-465.0; P = .04). An association was observed only in the European cohort for HLA-B*44:03 (OR, 5.50; 95% CI, 1.47-20.50; P = .01) and HLA-C*12:03 (OR, 8.79; 95% CI, 1.83-42.20; P = .008).

Conclusions and Relevance

This study suggests that HLA-A*66:01 might be a marker for CM-SJS/TEN with SOCs in Brazilian individuals of Pardo and European ancestry and that HLA-B*44:03 and HLA-C*12:03 might be markers only in those of European ancestry. Moreover, HLA-A*11:01 might be a marker of resistance to CM-SJS/TEN with SOCs.

An Efficient Multiple-Testing Adjustment for eQTL Studies that Accounts for Linkage Disequilibrium between Variants

Methods for multiple-testing correction in local expression quantitative trait locus (cis-eQTL) studies are a trade-off between statistical power and computational efficiency. Bonferroni correction, though computationally trivial, is overly conservative and fails to account for linkage disequilibrium between variants. Permutation-based methods are more powerful, though computationally far more intensive. We present an alternative correction method called eigenMT, which runs over 500 times faster than permutations and has adjusted p values that closely approximate empirical ones. To achieve this speed while also maintaining the accuracy of permutation-based methods, we estimate the effective number of independent variants tested for association with a particular gene, termed Meff, by using the eigenvalue decomposition of the genotype correlation matrix. We employ a regularized estimator of the correlation matrix to ensure Meff is robust and yields adjusted p values that closely approximate p values from permutations. Finally, using a common genotype matrix, we show that eigenMT can be applied with even greater efficiency to studies across tissues or conditions. Our method provides a simpler, more efficient approach to multiple-testing correction than existing methods and fits within existing pipelines for eQTL discovery.

Common variation in COL4A1/COL4A2 is associated with sporadic cerebral small vessel disease

OBJECTIVES

We hypothesized that common variants in the collagen genes COL4A1/COL4A2 are associated with sporadic forms of cerebral small vessel disease.

METHODS

We conducted meta-analyses of existing genotype data among individuals of European ancestry to determine associations of 1,070 common single nucleotide polymorphisms (SNPs) in the COL4A1/COL4A2 genomic region with the following: intracerebral hemorrhage and its subtypes (deep, lobar) (1,545 cases, 1,485 controls); ischemic stroke and its subtypes (cardioembolic, large vessel disease, lacunar) (12,389 cases, 62,004 controls); and white matter hyperintensities (2,733 individuals with ischemic stroke and 9,361 from population-based cohorts with brain MRI data). We calculated a statistical significance threshold that accounted for multiple testing and linkage disequilibrium between SNPs (p < 0.000084).

RESULTS

Three intronic SNPs in COL4A2 were significantly associated with deep intracerebral hemorrhage (lead SNP odds ratio [OR] 1.29, 95% confidence interval [CI] 1.14-1.46, p = 0.00003; r(2) > 0.9 between SNPs). Although SNPs associated with deep intracerebral hemorrhage did not reach our significance threshold for association with lacunar ischemic stroke (lead SNP OR 1.10, 95% CI 1.03-1.18, p = 0.0073), and with white matter hyperintensity volume in symptomatic ischemic stroke patients (lead SNP OR 1.07, 95% CI 1.01-1.13, p = 0.016), the direction of association was the same. There was no convincing evidence of association with white matter hyperintensities in population-based studies or with non-small vessel disease cerebrovascular phenotypes.

CONCLUSIONS

Our results indicate an association between common variation in the COL4A2 gene and symptomatic small vessel disease, particularly deep intracerebral hemorrhage. These findings merit replication studies, including in ethnic groups of non-European ancestry.

Correction for multiple testing in a gene region

Several methods to correct for multiple testing within a gene region have been proposed. These methods are useful for candidate gene studies, and to fine map gene-regions from GWAs. The Bonferroni correction and permutation are common adjustments, but are overly conservative and computationally intensive, respectively. Other options include calculating the effective number of independent single-nucleotide polymorphisms (SNPs) or using theoretical approximations. Here, we compare a theoretical approximation based on extreme tail theory with four methods for calculating the effective number of independent SNPs. We evaluate the type-I error rates of these methods using single SNP association tests over 10 gene regions simulated using 1000 Genomes data. Overall, we find that the effective number of independent SNP method by Gao et al, as well as extreme tail theory produce type-I error rates at the or close to the chosen significance level. The type-I error rates for the other effective number of independent SNP methods vary by gene region characteristics. We find Gao et al and extreme tail theory to be efficient alternatives to more computationally intensive approaches to control for multiple testing in gene regions.

Rapid and robust resampling-based multiple-testing correction with application in a genome-wide expression quantitative trait loci study

Genome-wide expression quantitative trait loci (eQTL) studies have emerged as a powerful tool to understand the genetic basis of gene expression and complex traits. In a typical eQTL study, the huge number of genetic markers and expression traits and their complicated correlations present a challenging multiple-testing correction problem. The resampling-based test using permutation or bootstrap procedures is a standard approach to address the multiple-testing problem in eQTL studies. A brute force application of the resampling-based test to large-scale eQTL data sets is often computationally infeasible. Several computationally efficient methods have been proposed to calculate approximate resampling-based P-values. However, these methods rely on certain assumptions about the correlation structure of the genetic markers, which may not be valid for certain studies. We propose a novel algorithm, rapid and exact multiple testing correction by resampling (REM), to address this challenge. REM calculates the exact resampling-based P-values in a computationally efficient manner. The computational advantage of REM lies in its strategy of pruning the search space by skipping genetic markers whose upper bounds on test statistics are small. REM does not rely on any assumption about the correlation structure of the genetic markers. It can be applied to a variety of resampling-based multiple-testing correction methods including permutation and bootstrap methods. We evaluate REM on three eQTL data sets (yeast, inbred mouse, and human rare variants) and show that it achieves accurate resampling-based P-value estimation with much less computational cost than existing methods. The software is available at http://csbio.unc.edu/eQTL.

Association of sirtuin 1 (SIRT1) gene SNPs and transcript expression levels with severe obesity

Recent studies have reported associations of sirtuin 1 (SIRT1) single nucleotide polymorphisms (SNPs) to both obesity and BMI. This study was designed to investigate association between SIRT1 SNPs, SIRT1 gene expression and obesity. Case-control analyses were performed using 1,533 obese subjects (896 adults, BMI >40 kg/m(2) and 637 children, BMI >97th percentile for age and sex) and 1,237 nonobese controls, all French Caucasians. Two SNPs (in high linkage disequilibrium (LD), r(2) = 0.96) were significantly associated with adult obesity, rs33957861 (P value = 0.003, odds ratio (OR) = 0.75, confidence interval (CI) = 0.61-0.92) and rs11599176 (P value: 0.006, OR = 0.74, CI = 0.61-0.90). Expression of SIRT1 mRNA was measured in BMI-discordant siblings from 154 Swedish families. Transcript expression was significantly correlated to BMI in the lean siblings (r(2) = 0.13, P value = 3.36 × 10(-7)) and lower SIRT1 expression was associated with obesity (P value = 1.56 × 10(-35)). There was also an association between four SNPs (rs11599176, rs12413112, rs33957861, and rs35689145) and BMI (P values: 4 × 10(-4), 6 × 10(-4), 4 × 10(-4), and 2 × 10(-3)) with the rare allele associated with a lower BMI. However, no SNP was associated with SIRT1 transcript expression level. In summary, both SNPs and SIRT1 gene expression are associated with severe obesity.

Evaluating the effective numbers of independent tests and significant p-value thresholds in commercial genotyping arrays and public imputation reference datasets

Current genome-wide association studies (GWAS) use commercial genotyping microarrays that can assay over a million single nucleotide polymorphisms (SNPs). The number of SNPs is further boosted by advanced statistical genotype-imputation algorithms and large SNP databases for reference human populations. The testing of a huge number of SNPs needs to be taken into account in the interpretation of statistical significance in such genome-wide studies, but this is complicated by the non-independence of SNPs because of linkage disequilibrium (LD). Several previous groups have proposed the use of the effective number of independent markers (M_e) for the adjustment of multiple testing, but current methods of calculation for M_e are limited in accuracy or computational speed. Here, we report a more robust and fast method to calculate M_e. Applying this efficient method [implemented in a free software tool named Genetic type 1 error calculator (GEC)], we systematically examined the M_e, and the corresponding p-value thresholds required to control the genome-wide type 1 error rate at 0.05, for 13 Illumina or Affymetrix genotyping arrays, as well as for HapMap Project and 1000 Genomes Project datasets which are widely used in genotype imputation as reference panels. Our results suggested the use of a p-value threshold of ~10⁻⁷ as the criterion for genome-wide significance for early commercial genotyping arrays, but slightly more stringent p-value thresholds ~5 × 10⁻⁸ for current or merged commercial genotyping arrays, ~10⁻⁸ for all common SNPs in the 1000 Genomes Project dataset and ~5 × 10⁻⁸ for the common SNPs only within genes.

A new approach to account for the correlations among single nucleotide polymorphisms in genome: wide association studies

In genetic association studies, such as genome-wide association studies (GWAS), the number of single nucleotide polymorphisms (SNPs) can be as large as hundreds of thousands. Due to linkage disequilibrium, many SNPs are highly correlated; assuming they are independent is not valid. The commonly used multiple comparison methods, such as Bonferroni correction, are not appropriate and are too conservative when applied to GWAS. To overcome these limitations, many approaches have been proposed to estimate the so-called effective number of independent tests to account for the correlations among SNPs. However, many current effective number estimation methods are based on eigenvalues of the correlation matrix. When the dimension of the matrix is large, the numeric results may be unreliable or even unobtainable. To circumvent this obstacle and provide better estimates, we propose a new effective number estimation approach which is not based on the eigenvalues. We compare the new method with others through simulated and real data. The comparison results show that the proposed method has very good performance.

Effective sample size: Quick estimation of the effect of related samples in genetic case-control association analyses

Affected relatives are essential for pedigree linkage analysis, however, they cause a violation of the independent sample assumption in case-control association studies. To avoid the correlation between samples, a common practice is to take only one affected sample per pedigree in association analysis. Although several methods exist in handling correlated samples, they are still not widely used in part because these are not easily implemented, or because they are not widely known. We advocate the effective sample size method as a simple and accessible approach for case-control association analysis with correlated samples. This method modifies the chi-square test statistic, p-value, and 95% confidence interval of the odds-ratio by replacing the apparent number of allele or genotype counts with the effective ones in the standard formula, without the need for specialized computer programs. We present a simple formula for calculating effective sample size for many types of relative pairs and relative sets. For allele frequency estimation, the effective sample size method captures the variance inflation exactly. For genotype frequency, simulations showed that effective sample size provides a satisfactory approximation. A gene which is previously identified as a type 1 diabetes susceptibility locus, the interferon-induced helicase gene (IFIH1), is shown to be significantly associated with rheumatoid arthritis when the effective sample size method is applied. This significant association is not established if only one affected sib per pedigree were used in the association analysis. Relationship between the effective sample size method and other methods - the generalized estimation equation, variance of eigenvalues for correlation matrices, and genomic controls - are discussed.

Multiple testing and power calculations in genetic association studies

Modern genetic association studies typically involve multiple single-nucleotide polymorphisms (SNPs) and/or multiple genes. With the development of high-throughput genotyping technologies and the reduction in genotyping cost, investigators can now assay up to a million SNPs for direct or indirect association with disease phenotypes. In addition, some studies involve multiple disease or related phenotypes and use multiple methods of statistical analysis. The combination of multiple genetic loci, multiple phenotypes, and multiple methods of evaluating associations between genotype and phenotype means that modern genetic studies often involve the testing of an enormous number of hypotheses. When multiple hypothesis tests are performed in a study, there is a risk of inflation of the type I error rate (i.e., the chance of falsely claiming an association when there is none). Several methods for multiple-testing correction are in popular use, and they all have strengths and weaknesses. Because no single method is universally adopted or always appropriate, it is important to understand the principles, strengths, and weaknesses of the methods so that they can be applied appropriately in practice. In this article, we review the three principle methods for multiple-testing correction and provide guidance for calculating statistical power.

DNA sequence variants of the FKBP5 gene are associated with unipolar depression

FKBP5 is a glucocorticoid receptor-regulating co-chaperone of hsp-90 and, therefore, is suggested to play a role in the regulation of the hypothalamic-pituitary-adrenocortical system and the pathophysiology of depression. Previously, three studies identified single nucleotide polymorphisms (SNPs) in the FKBP5 gene associated with response to antidepressants, and one study found an association with diagnosis of depression. We selected five markers from the region of interest. A case-control sample comprising 268 German in-patients with recurrent unipolar depression, and 284 German controls recruited from the general population were available. Association of the selected FKBP5 sequence variants with clinical depression were analysed. In addition, we explored association with treatment response by (a) the Hamilton Depression Rating Scale and (b) the dexamethasone/corticotrophin-releasing hormone (Dex/CRH) test, as well as association with hippocampal volumes in a subpopulation of 110 patients. For three of the five investigated SNPs we were able to show association with the diagnosis of depression. In the subpopulation of 110 patients, diagnosis-related alleles were also associated with the reduction of cortisol secretion in the Dex/CRH test during a 4-wk treatment period, while psychopathological changes were not associated. Furthermore, diagnosis-related alleles were associated with reduction of the hippocampal volume. This study extends the replicated association of a promoter SNP with antidepressant response on a biological level by demonstrating normalization of the cortisol response under Dex/CRH stimulation during treatment. Furthermore, several of the investigated SNPs were associated with the disease status and the intermediate phenotype of hippocampal volume.

Avoiding the high Bonferroni penalty in genome-wide association studies

A major challenge in genome-wide association studies (GWASs) is to derive the multiple testing threshold when hypothesis tests are conducted using a large number of single nucleotide polymorphisms. Permutation tests are considered the gold standard in multiple testing adjustment in genetic association studies. However, it is computationally intensive, especially for GWASs, and can be impractical if a large number of random shuffles are used to ensure accuracy. Many researchers have developed approximation algorithms to relieve the computing burden imposed by permutation. One particularly attractive alternative to permutation is to calculate the effective number of independent tests, M(eff), which has been shown to be promising in genetic association studies. In this study, we compare recently developed M(eff) methods and validate them by the permutation test with 10,000 random shuffles using two real GWAS data sets: an Illumina 1M BeadChip and an Affymetrix GeneChip Human Mapping 500K Array Set. Our results show that the simpleM method produces the best approximation of the permutation threshold, and it does so in the shortest amount of time. We also show that M(eff) is indeed valid on a genome-wide scale in these data sets based on statistical theory and significance tests. The significance thresholds derived can provide practical guidelines for other studies using similar population samples and genotyping platforms.

A new measure of the effective number of tests, a practical tool for comparing families of non-independent significance tests

p-Values from tests of significance can be combined using the Sidák correction (or the closely related Bonferroni correction) or Fisher's method, but both these methods require that the p-values combined be independent when all null hypotheses tested are true. In this paper adjustments to these methods are proposed, using a new eigenvalue-based measure of the effective number of independent tests to which the actual tests performed are equivalent, and are compared with adjustments proposed by previous authors. The adjusted methods are evaluated using a sample of 726 Alzheimer's disease (AD) cases and 707 group-matched controls, genotyped at 84,975 single-nucleotide polymorphism loci in 2,000 randomly chosen genes. The tests for genetic association with AD at loci within each gene are combined. The number of loci tested per gene varies from 2 to 994. The adjusted combined p-values agree well with the significance of the combined p-values determined empirically by random permutation of the data (Sidák correction: r=0.990; Fisher's method: r=0.994). This indicates that the combined p-values can be used to assess the relative strength of evidence for association of these genes with AD. The adjustment proposed here is a refinement of that of Nyholt ([2004] Am. J. Hum. Genet. 74:765-769), giving improved agreement with the results of random permutation. The improvement obtained is similar to that given by the refinement proposed by Li and Ji ([2005] Heredity 95:221-227). It is concluded that the concept of an effective number of tests is a valid approximation that allows p-values to be combined in a highly informative way.

Evidence that TGFA influences risk to cleft lip with/without cleft palate through unconventional genetic mechanisms

This study examined the association between markers in transforming growth factor alpha (TGFA) and isolated, non-syndromic cleft lip with/without palate (CL/P) using a case-parent trio design, considering parent-of-origin effects. We also tested for gene-environmental interaction with common maternal exposures, and for gene-gene interaction using markers in TGFA and another recognized causal gene, IRF6. CL/P case-parent trios from four populations (76 from Maryland, 146 from Taiwan, 35 from Singapore, and 40 from Korea) were genotyped for 17 single nucleotide polymorphisms (SNPs) in TGFA. The transmission disequilibrium test was used to test individual SNPs, and the parent-of-origin likelihood ratio test (PO-LRT) was used to assess parent-of-origin effects. We also screened for possible gene-environment interaction using PBAT, and tested for gene-gene interaction using conditional logistic regression models. When all trios were combined, four SNPs showed significant excess maternal transmission, two of which gave significant PO-LRT values [rs3821261: P = 0.004 and OR(imprinting) = 4.17; and rs3771475: P = 0.027 and OR(imprinting) = 2.44]. Haplotype analysis of these two SNPS also supported excess maternal transmission. We saw intriguing but suggestive evidence of G x E interaction for several SNPs in TGFA when either individual SNPs or haplotypes of adjacent SNPs were considered. Thus, TGFA appears to influence risk of CL/P through unconventional means with an apparent parent-of-origin effect (excess maternal transmission) and possible interaction with maternal exposures.

Maternal transmission effects of the PAX genes among cleft case-parent trios from four populations

Isolated cleft lip with or without cleft palate (CL/P) is among the most common human birth defects, with a prevalence of 1 in 700 live births. The paired box (PAX) genes have been suggested as candidate genes for CL/P based largely on mouse models; however, few human studies have focused on this gene family. This study tests for association between markers in four PAX genes and CL/P using a case-parent trio design considering parent-of-origin effects. Trios from four populations (76 from Maryland, 146 from Taiwan, 35 from Singapore, and 40 from Korea) were genotyped for 34 single nucleotide polymorphisms (SNPs) in the PAX3, PAX6, PAX7, and PAX9 genes. We performed the transmission disequilibrium test (TDT) on individual SNPs. Parent-of-origin effects were assessed using the transmission asymmetry test (TAT) and the parent-of-origin likelihood ratio test (PO-LRT). TDT analysis showed one SNP (rs766325) in PAX7 yielding evidence of linkage and association when parent-of-origin was not considered, with an OR(transmission)=1.62 (P=0.003), and five SNPs in PAX6 (including two pairs in near perfect linkage disequilibrium). TAT analysis of all trios revealed two SNPs in PAX7 and four SNPs in PAX3 showing significant excess maternal transmission. For these six SNPs, the maternal OR(transmission) ranged between 1.74 and 2.40, and PO-LRT was also significant (P-values=0.035-0.012). When this analysis was limited to trios with male cases, SNPs in PAX7 showed higher maternal OR(transmission) and greater significance. PAX genes may influence the risk of CL/P through maternal effects, possibly imprinting, which seems to be stronger among male cases.

Genome-wide significance for dense SNP and resequencing data

The problem of multiple testing is an important aspect of genome-wide association studies, and will become more important as marker densities increase. The problem has been tackled with permutation and false discovery rate procedures and with Bayes factors, but each approach faces difficulties that we briefly review. In the current context of multiple studies on different genotyping platforms, we argue for the use of truly genome-wide significance thresholds, based on all polymorphisms whether or not typed in the study. We approximate genome-wide significance thresholds in contemporary West African, East Asian and European populations by simulating sequence data, based on all polymorphisms as well as for a range of single nucleotide polymorphism (SNP) selection criteria. Overall we find that significance thresholds vary by a factor of >20 over the SNP selection criteria and statistical tests that we consider and can be highly dependent on sample size. We compare our results for sequence data to those derived by the HapMap Consortium and find notable differences which may be due to the small sample sizes used in the HapMap estimate.

So many correlated tests, so little time! Rapid adjustment of P values for multiple correlated tests

Contemporary genetic association studies may test hundreds of thousands of genetic variants for association, often with multiple binary and continuous traits or under more than one model of inheritance. Many of these association tests may be correlated with one another because of linkage disequilibrium between nearby markers and correlation between traits and models. Permutation tests and simulation-based methods are often employed to adjust groups of correlated tests for multiple testing, since conventional methods such as Bonferroni correction are overly conservative when tests are correlated. We present here a method of computing P values adjusted for correlated tests (P(ACT)) that attains the accuracy of permutation or simulation-based tests in much less computation time, and we show that our method applies to many common association tests that are based on multiple traits, markers, and genetic models. Simulation demonstrates that P(ACT) attains the power of permutation testing and provides a valid adjustment for hundreds of correlated association tests. In data analyzed as part of the Finland-United States Investigation of NIDDM Genetics (FUSION) study, we observe a near one-to-one relationship (r(2)>.999) between P(ACT) and the corresponding permutation-based P values, achieving the same precision as permutation testing but thousands of times faster.

T cell-specific T-box transcription factor haplotype is associated with allergic asthma in children

BACKGROUND

T cell-specific T-box transcription factor (T-bet) is a member of the T-box family of transcription factors regulating lineage commitment of T(H) lymphocytes toward a predominant T(H)1 phenotype. Asthma and allergy are common complex diseases characterized by T(H)2-mediated inflammation.

OBJECTIVE

We aimed to assess possible relationships between the T-bet gene (TBX21) and asthma and allergy in children.

METHOD

Twelve single nucleotide polymorphisms (SNPs) in the TBX21 region were genotyped in 948 children from the Environment and Childhood Asthma study. Allele and haplotype frequencies were compared in children with and without asthma (by 10 years) and allergy (> or =1 positive skin prick test response), as well as for the quantitative traits bronchial hyperresponsiveness determined by means of methacholine bronchial challenge testing, lung function determined by means of forced flow volume loops, fractional exhaled nitric oxide measurement, eosinophil count, and serum total IgE level.

RESULTS

Allergic asthma was significantly associated with 2 of the tested SNPs (rs11650354 and rs16947078) and further associated with the particular haplotype including these SNPs, with homozygote status resulting in an odds ratio of 8.3 (95% CI, 2.5-26.9) for allergic asthma. Neither nonallergic asthma or "allergy alone" nor the remaining quantitative variables were associated with TBX21 SNPs or haplotypes.

CONCLUSION

An association between a specific TBX21 haplotype and allergic asthma in children is demonstrated for the first time and might explain previously detected associations between SNPs within TBX21 and asthma and bronchial hyperresponsiveness.

Investigation of the estrogen receptor-alpha gene with type 2 diabetes and/or nephropathy in African-American and European-American populations

The estrogen receptor-alpha gene (ESR1) was selected as a positional candidate under a type 2 diabetes linkage peak at 6q24-27. A total of 42 ESR1 single nucleotide polymorphisms (SNPs) were genotyped in 380 African-American type 2 diabetic case subjects with end-stage renal disease (ESRD) and 276 African-American control subjects. A total of 22 ancestry informative markers were also genotyped, and the program Admixmap was used to adjust allelic and haplotypic association tests for individual estimates of admixture. The most significant association with type 2 diabetes-ESRD was with rs1033182 in intron 2 (P = 0.013, admixture-adjusted P(a) = 0.021). Genotyping 17 SNPs across a region of ESR1 intron 1-intron 2 in an expanded population of 851 case and 635 control subjects supported association with rs1033182 (P = 0.004, P(a) = 0.027) and with an independent six-SNP haplotype of high linkage disequilibrium spanning 6.4 kb (P < 0.0001, P(a) < 0.0001). The same 17 ESR1 SNPs were genotyped in 300 European-American type 2 diabetes-ESRD case subjects and 310 European-American control subjects. Two intron 2 SNPs, rs2431260 (P = 0.015) and rs1709183 (P = 0.019), and a four-SNP haplotype containing these SNPs (P = 0.033) were associated with type 2 diabetes and/or ESRD. Results suggest that intron 1 and intron 2 of the ESR1 gene may contain functionally important regions related to type 2 diabetes or ESRD risk.

Linkage and association analysis of CACNG3 in childhood absence epilepsy

Childhood absence epilepsy (CAE) is an idiopathic generalised epilepsy characterised by absence seizures manifested by transitory loss of awareness with 2.5-4 Hz spike-wave complexes on ictal EEG. A genetic component to aetiology is established but the mechanism of inheritance and the genes involved are not fully defined. Available evidence suggests that genes encoding brain expressed voltage-gated calcium channels, including CACNG3 on chromosome 16p12-p13.1, may represent susceptibility loci for CAE. The aim of this work was to further evaluate CACNG3 as a susceptibility locus by linkage and association analysis. Assuming locus heterogeneity, a significant HLOD score (HLOD = 3.54, alpha = 0.62) was obtained for markers encompassing CACNG3 in 65 nuclear families with a proband with CAE. The maximum non-parametric linkage score was 2.87 (P < 0.002). Re-sequencing of the coding exons in 59 patients did not identify any putative causal variants. A linkage disequilibrium (LD) map of CACNG3 was constructed using 23 single nucleotide polymorphisms (SNPs). Transmission disequilibrium was sought using individual SNPs and SNP-based haplotypes with the pedigree disequilibrium test in 217 CAE trios and the 65 nuclear pedigrees. Evidence for transmission disequilibrium (P < or = 0.01) was found for SNPs within a approximately 35 kb region of high LD encompassing the 5'UTR, exon 1 and part of intron 1 of CACNG3. Re-sequencing of this interval was undertaken in 24 affected individuals. Seventy-two variants were identified: 45 upstream; two 5'UTR; and 25 intronic SNPs. No coding sequence variants were identified, although four variants are predicted to affect exonic splicing. This evidence supports CACNG3 as a susceptibility locus in a subset of CAE patients.

Analysis of candidate genes on chromosome 2 in oral cleft case-parent trios from three populations

Isolated oral clefts, including cleft lip with/without cleft palate (CL/P) and cleft palate (CP), have a complex and heterogeneous etiology. Case-parent trios from three populations were used to study genes spanning chromosome 2, where single nucleotide polymorphic (SNP) markers were analyzed individually and as haplotypes. Case-parent trios from three populations (74 from Maryland, 64 from Singapore and 95 from Taiwan) were genotyped for 962 SNPs in 104 genes on chromosome 2, including two well-recognized candidate genes: TGFA and SATB2. Individual SNPs and haplotypes (in sliding windows of 2-5 SNPs) were used to test for linkage and disequilibrium separately in CL/P and CP trios. A novel candidate gene (ZNF533) showed consistent evidence of linkage and disequilibrium in all three populations for both CL/P and CP. SNPs in key regions of ZNF533 showed considerable variability in estimated genotypic odds ratios and their significance, suggesting allelic heterogeneity. Haplotype frequencies for regions of ZNF533 were estimated and used to partition genetic variance into among-and within-population components. Wright's fixation index, a measure of genetic diversity, showed little difference between Singapore and Taiwan compared with Maryland. The tensin-1 gene (TNS1) also showed evidence of linkage and disequilibrium among both CL/P and CP trios in all three populations, albeit at a lower level of significance. Additional genes (VAX2, GLI2, ZHFX1B on 2p; WNT6-WNT10A and COL4A3-COL4A4 on 2q) showed consistent evidence of linkage and disequilibrium only among CL/P trios in all three populations, and TGFA showed significant evidence in two of three populations.

Convergent evidence that oligodendrocyte lineage transcription factor 2 (OLIG2) and interacting genes influence susceptibility to schizophrenia

Abnormal oligodendrocyte function has been postulated as a primary etiological event in schizophrenia. Oligodendrocyte lineage transcription factor 2 (OLIG2) encodes a transcription factor central to oligodendrocyte development. Analysis of OLIG2 in a case-control sample (n = approximately 1,400) in the U.K. revealed several SNPs to be associated with schizophrenia (minimum P = 0.0001, gene-wide P = 0.0009). To obtain independent support for this association, we sought evidence for genetic interaction between OLIG2 and three genes of relevance to oligodendrocyte function for which we have reported evidence for association with schizophrenia: CNP, NRG1, and ERBB4. We found interaction effects on disease risk between OLIG2 and CNP (minimum P = 0.0001, corrected P = 0.008) for interaction with ERBB4 (minimum P = 0.002, corrected P = 0.04) but no evidence for interaction with NRG1. To investigate the biological plausibility of the interactions, we sought correlations between the expression of the genes. The results were similar to those of the genetic interaction analysis. OLIG2 expression significantly correlated in cerebral cortex with CNP (P < 10(-7)) and ERBB4 (P = 0.002, corrected P = 0.038) but not NRG1. In mouse striatum, Olig2 and Cnp expression also was correlated, and linkage analysis for trans-effects on gene expression suggests that each locus regulates the other's expression. Our data provide strong convergent evidence that variation in OLIG2 confers susceptibility to schizophrenia alone and as part of a network of genes implicated in oligodendrocyte function.

Detecting multiple associations in genome-wide studies

Recent developments in the statistical analysis of genome-wide studies are reviewed. Genome-wide analyses are becoming increasingly common in areas such as scans for disease-associated markers and gene expression profiling. The data generated by these studies present new problems for statistical analysis, owing to the large number of hypothesis tests, comparatively small sample size and modest number of true gene effects. In this review, strategies are described for optimising the genotyping cost by discarding promising genes at an earlier stage, saving resources for the genes that show a trend of association. In addition, there is a review of new methods of analysis that combine evidence across genes to increase sensitivity to multiple true associations in the presence of many non-associated genes. Some methods achieve this by including only the most significant results, whereas others model the overall distribution of results as a mixture of distributions from true and null effects. Because genes are correlated even when having no effect, permutation testing is often necessary to estimate the overall significance, but this can be very time consuming. Efficiency can be improved by fitting a parametric distribution to permutation replicates, which can be re-used in subsequent analyses. Methods are also available to generate random draws from the permutation distribution. The review also includes discussion of new error measures that give a more reasonable interpretation of genome-wide studies, together with improved sensitivity. The false discovery rate allows a controlled proportion of positive results to be false, while detecting more true positives; and the local false discovery rate and false-positive report probability give clarity on whether or not a statistically significant test represents a real discovery.