Efficiency robust statistics for genetic linkage and association studies under genetic model uncertainty

Genome-wide association testing in malaria studies in the presence of overdominance

BACKGROUND

In human genetics, heterozygote advantage (heterosis) has been detected in studies that focused on specific genes but not in genome-wide association studies (GWAS). For example, heterosis is believed to confer resistance to certain strains of malaria in patients heterozygous for the sickle-cell gene, haemoglobin S (HbS). Yet the power of allelic tests can be substantially diminished by heterosis. Since GWAS (and haplotype-associations) also utilize allelic tests, it is unclear to what degree GWAS could underachieve because heterosis is ignored.

METHODS

In this study, a two-step approach to genetic association testing in malaria studies in a GWAS setting that may enhance the power of the tests was proposed, by identifying the underlying genetic model first before applying the association tests. Generalized linear models for dominant, recessive, additive, and heterotic effects were fitted and model selection was performed. This was achieved via tests of significance using the MAX and allelic tests, noting the minimum p-values across all the models and the proportion of tests that a given genetic model was deemed the best. An example dataset, based on 17 SNPs, from a robust genetic association study and simulated genotype datasets, were used to illustrate the method. Case-control genotype data on malaria from Kenya and Gambia were used for validation.

RESULTS AND CONCLUSION

Results showed that the allelic test returned some false negatives under the heterosis model, suggesting reduced power in testing genetic association. Disparities were observed for some chromosomes in the Kenyan and Gambian datasets, including the sex chromosomes. Thus, GWAS and haplotype associations should be treated with caution, unless the underlying genetic model had been determined.

The association of lncRNA SNPs and SNPs-environment interactions based on GWAS with HBV-related HCC risk and progression

Background

Long non‐coding RNA (lncRNA) plays an essential role in hepatitis B virus‐related hepatocellular carcinoma (HBV‐related HCC) occurrence and development. Single nucleotide polymorphism (SNP) may affect HBV‐related HCC susceptibility by altering the function of lncRNA. However, the relationship between lncRNA SNPs and HBV‐related HCC occurrence and development is still unclear.

Methods

In the present study, based on HBV‐related HCC genome‐wide association studies, eight potentially functional SNPs from two lncRNAs were predicted using a set of bioinformatics strategies. In 643 HBV‐related HCC patients, 549 CHB carriers, and 553 HBV natural clearance subjects from Southern Chinese, we evaluated associations between SNPs and HBV‐related HCC occurrence or development with odds ratio (OR) and 95% confidence interval (CI) under credible genetic models.

Results

In HBV‐related HCC patients, rs9908998 was found to significantly increase the risk of lymphatic metastasis under recessive model (Adjusted OR = 1.95, 95% CI = 1.20–3.17). Lnc‐RP11‐150O12.3 rs2275959, rs1008547, and rs11776545 with cancer family history may show significant multiplicative and additive interactions on HBV‐related HCC susceptibility (all p_Adjusted < .05). The associations of rs2275959, rs1008547, and rs11776545 with distant metastasis of HBV‐related HCC patients were observed in additive model (Adjusted OR = 1.45, 95% CI = 1.06–1.97 for rs2275959; Adjusted OR = 1.45, 95% CI = 1.06–1.98 for rs1008547; Adjusted OR = 1.40, 95% CI = 1.03–1.91 for rs11776545).

Conclusion

Taken together, lnc‐ACACA‐1 rs9908998, lnc‐RP11‐150O12.3 rs2275959, rs1008547, and rs11776545 might be predictors for HBV‐related HCC risk or prognosis.

Power and Sample Size Calculations for Genetic Association Studies in the Presence of Genetic Model Misspecification

INTRODUCTION

When analyzing data from large-scale genetic association studies, such as targeted or genome-wide resequencing studies, it is common to assume a single genetic model, such as dominant or additive, for all tests of association between a given genetic variant and the phenotype. However, for many variants, the chosen model will result in poor model fit and may lack statistical power due to model misspecification.

OBJECTIVE

We develop power and sample size calculations for tests of gene and gene × environment interaction, allowing for misspecification of the true mode of genetic susceptibility.

METHODS

The power calculations are based on a likelihood ratio test framework and are implemented in an open-source R package ("genpwr").

RESULTS

We use these methods to develop an analysis plan for a resequencing study in idiopathic pulmonary fibrosis and show that using a 2-degree of freedom test can increase power to detect recessive genetic effects while maintaining power to detect dominant and additive effects.

CONCLUSIONS

Understanding the impact of model misspecification can aid in study design and developing analysis plans that maximize power to detect a range of true underlying genetic effects. In particular, these calculations help identify when a multiple degree of freedom test or other robust test of association may be advantageous.

Genome-wide association study of genetic variants related to anthracycline-induced cardiotoxicity in early breast cancer

We performed a genome‐wide association study to investigate the association between single nucleotide polymorphisms and anthracycline‐induced cardiotoxicity (ACT) in patients diagnosed with early breast cancer. From January 2000 to December 2015, 8490 patients underwent breast surgery at the National Cancer Center in Korea. Patients who received doxorubicin (cumulative dose 240 mg/m²‐300 mg/m²) with or without trastuzumab as a neoadjuvant/adjuvant therapy were included in our cohort. Sixty‐seven patients in our cohort were diagnosed with ACT. Clinical data, including age, body weight, height, cancer stage, trastuzumab treatment, comorbidities, and concomitant medications, were collected retrospectively. Patients were classified as having either persistent or transient ACT based on their clinical course. In total, 346 946 single nucleotide polymorphisms in 42 cases and 215 controls were tested in this study. Body mass index (BMI) ≥25 kg/m² [odds ratio (OR) = 2.45, 95% confidence interval (CI), 1.23‐4.88, P = .011] and trastuzumab use (OR = 2.40, 95% CI, 1.11‐5.17, P = .026) were identified as significant risk factors. We found 7 genetic variants for ACT including rs17530621 (SHISA3, P = 3.10E−06), rs11894115 (MPP4, P = 4.71E−06), rs58328254 (RPL7, P = 6.09E−06), and rs117299725 (PRUNE2, P = 8.53E−06), although none of these variants reached the Bonferroni‐corrected significance level when adjusted for BMI and trastuzumab use ( = α1.44E−07 based on 0.05/346 946). rs117299725 was a common variant when only the persistent ACT group was analyzed separately. It is meaningful that our study analyzed comprehensively the influence of genetic variation on ACT, along with some clinical factors in Asian breast cancer patients who received anthracycline with or without trastuzumab. Further research will be needed on candidate genetic variants found in this study.

The association of integration patterns of human papilloma virus and single nucleotide polymorphisms on immune- or DNA repair-related genes in cervical cancer patients

The present study investigated the association between single nucleotide polymorphisms (SNPs) in immune- or DNA repair-related genes and the integration pattern of human papillomavirus (HPV), a promising prognostic marker in cervical cancer. The HPV integration patterns of cervical cancer patients were determined by polymerase chain reaction and in situ hybridization, and categorized as episomal (group A), single-copy or multi-copy tandem repetition integrated (group B), and undetectable HPV types (group C). After sample and SNP quality control, 166,505 SNPs in 161 samples (38, 111, and 12 patients in groups A, B, and C, respectively) were examined. None of the SNPs reached genome-wide significance, and several candidate SNPs for future study were selected, including rs10999435 on chromosome 10q22, rs1322054 on chromosome 9q32-33, and rs10902171 on chromosome 11p15. Luciferase assay identified rs1322054 as the primary functional variant to regulate gene expression in immune cell. Further studies are needed to determine the genetic background of different integration patterns of HPV in cervical cancer patients.

Investigation of dominant and recessive inheritance models in genome-wide association studies data of nonsyndromic cleft lip with or without cleft palate

Nonsyndromic cleft lip with or without cleft palate (nsCL/P) is one of the most common congenital malformation worldwide, and its etiology involves both genetic and environmental factors. Recent genome-wide and targeted genetic studies of nsCL/P have identified numerous genetic risk loci, under the hypothesis of a multiplicative mode of inheritance. The present study investigated whether novel nsCL/P risk loci could be identified by analyzing dominant/recessive genetic effects in single nucleotide polymorphism (SNP) data from genome-wide association studies. For this purpose, a genome-wide investigation of dominant/recessive common SNP effects was performed in our previously published meta-analysis data set. Twenty-four loci were identified as candidate regions. In a subsequent association analysis in an independent study cohort of 224 nsCL/P patients and 986 controls of European descent, none of the loci could be replicated. Therefore, our strategy of identifying novel loci by applying different genetic models did not yield any novel findings, suggesting that recessive/dominant common variation only make a limited contribution to nsCL/P in Europeans. However, we cannot rule out that such effects are present at some of the loci that have previously been identified, or are present in different populations.

A robust distribution-free test for genetic association studies of quantitative traits

In association studies of quantitative traits, the association of each genetic marker with the trait of interest is typically tested using the F-test assuming an additive genetic model. In practice, the true model is rarely known, and specifying an incorrect model can lead to a loss of power. For case-control studies, the maximum of test statistics optimal for additive, dominant, and recessive models has been shown to be robust to model misspecification. The approach has later been extended to quantitative traits. However, the existing procedures assume that the trait is normally distributed and may not maintain correct type I error rates and can also have reduced power when the assumption of normality is violated. Here, we introduce a maximum (MAX3) test that is based on ranks and is therefore distribution-free. We examine the behavior of the proposed method using a Monte Carlo simulation with both normal and non-normal data and compare the results to the usual parametric procedures and other nonparametric alternatives. We show that the rank-based maximum test has favorable properties relative to other tests, especially in the case of symmetric distributions with heavy tails. We illustrate the method with data from a real association study of symmetric dimethylarginine (SDMA).

Adaptive robust genetic association tests using case-parents triad families

Using the conditional likelihood, a model-specific score test can be derived under a given genetic model to test genetic association for case-parents triad family data. When the underlying genetic model is correctly specified, the score test is most powerful. However, it can lose substantial power when the model is misspecified. Several robust tests have been proposed to deal with the problem, such as the maximum test statistic, the maximin efficiency robust test, and the constrained likelihood ratio test. These tests have been shown to be robust against model misspecification compared with those model-specific score tests, but they are either time-consuming in computation or not sufficiently high in power robustness under some situations. In this study, a data-driven procedure is proposed to construct two adaptive robust genetic association tests WMERT and WMAX . The WMERT is simple in calculation and has fairly high power robustness. The empirical power of WMAX is quite stable and close to those of the model-specific score tests. The two proposed tests should be beneficial to practical genetic association studies. A real dataset consisting of neural tube defect triad families is used for illustration of the methods. R-scripts are also provided for numerical calculation of the proposed methods in practical studies.

Combining dependent F-tests for robust association of quantitative traits under genetic model uncertainty

In association mapping of quantitative traits, the F-test based on an assumed genetic model is a basic statistical tool for testing association of each candidate locus with the trait of interest. However, the true underlying genetic model is often unknown, and using an incorrect model may cause serious loss of power. For case-control studies, it is known that the combination of several tests that are optimal for different models is robust to model misspecification. In this paper, we extend the test combination approach to quantitative trait association. We first derive the exact correlations among transformed test statistics and discuss interesting special cases. We then propose and evaluate a multivariate normality based approximation to the joint distribution of test statistics, such that the marginal distributions and pairwise correlations among test statistics are accounted for. Through simulations, we show that the sizes of the resulting approximate combined tests are accurate for practical purposes under a variety of situations. We find that the combination of the tests from the additive model and the genotypic model performs well, because it demonstrates both robustness to incorrect models and satisfactory power. A mouse lipoprotein data set is used to demonstrate the method.

A note on the efficiencies of sampling strategies in two-stage Bayesian regional fine mapping of a quantitative trait

In focused studies designed to follow up associations detected in a genome-wide association study (GWAS), investigators can proceed to fine-map a genomic region by targeted sequencing or dense genotyping of all variants in the region, aiming to identify a functional sequence variant. For the analysis of a quantitative trait, we consider a Bayesian approach to fine-mapping study design that incorporates stratification according to a promising GWAS tag SNP in the same region. Improved cost-efficiency can be achieved when the fine-mapping phase incorporates a two-stage design, with identification of a smaller set of more promising variants in a subsample taken in stage 1, followed by their evaluation in an independent stage 2 subsample. To avoid the potential negative impact of genetic model misspecification on inference we incorporate genetic model selection based on posterior probabilities for each competing model. Our simulation study shows that, compared to simple random sampling that ignores genetic information from GWAS, tag-SNP-based stratified sample allocation methods reduce the number of variants continuing to stage 2 and are more likely to promote the functional sequence variant into confirmation studies.

A family-based robust multivariate association test using maximum statistic

For characterizing the genetic mechanisms of complex diseases familial data with multiple correlated quantitative traits are usually collected in genetic studies. To analyze such data, various multivariate tests have been proposed to investigate the association between the underlying disease genes and the multiple traits. Although these multivariate association tests may have better power performance than the univariate association tests, they suffer from loss of testing power when the genetic models of the putative genes are misspecified. To address the problem, in this paper we aim to develop a family-based robust multivariate association test. We will first establish the optimal multivariate score tests for the recessive, additive, and dominant genetic models. Based on these optimal tests, a maximum-type robust multivariate association test is then obtained. Simulations are conducted to compare the power of our method with that of other existing multivariate methods. The results show that the robust multivariate test does manifest the robustness in power over all plausible genetic models. A practical data set is applied to demonstrate the applicability of our approach. The results suggest that the robust multivariate test is more powerful than the robust univariate test when dealing with multiple quantitative traits.

Fisher's method of combining dependent statistics using generalizations of the gamma distribution with applications to genetic pleiotropic associations

A classical approach to combine independent test statistics is Fisher's combination of $p$-values, which follows the $\chi ^2$ distribution. When the test statistics are dependent, the gamma distribution (GD) is commonly used for the Fisher's combination test (FCT). We propose to use two generalizations of the GD: the generalized and the exponentiated GDs. We study some properties of mis-using the GD for the FCT to combine dependent statistics when one of the two proposed distributions are true. Our results show that both generalizations have better control of type I error rates than the GD, which tends to have inflated type I error rates at more extreme tails. In practice, common model selection criteria (e.g. Akaike information criterion/Bayesian information criterion) can be used to help select a better distribution to use for the FCT. A simple strategy of the two generalizations of the GD in genome-wide association studies is discussed. Applications of the results to genetic pleiotrophic associations are described, where multiple traits are tested for association with a single marker.

A unifying framework for robust association testing, estimation, and genetic model selection using the generalized linear model

The analysis of genome-wide genetic association studies generally starts with univariate statistical tests of each single-nucleotide polymorphism. The standard approach is the Cochran-Armitage trend test or its logistic regression equivalent although this approach can lose considerable power if the underlying genetic model is not additive. An alternative is the MAX test, which is robust against the three basic modes of inheritance. Here, the asymptotic distribution of the MAX test is derived using the generalized linear model together with the Delta method and multiple contrasts. The approach is applicable to binary, quantitative, and survival traits. It may be used for unrelated individuals, family-based studies, and matched pairs. The approach provides point and interval effect estimates and allows selecting the most plausible genetic model using the minimum P-value. R code is provided. A Monte-Carlo simulation study shows that the asymptotic MAX test framework meets type I error levels well, has good power, and good model selection properties for minor allele frequencies ≥0.3. Pearson's χ(2)-test is superior for lower minor allele frequencies with low frequencies for the rare homozygous genotype. In these cases, the model selection procedure should be used with caution. The use of the MAX test is illustrated by reanalyzing findings from seven genome-wide association studies including case-control, matched pairs, and quantitative trait data.

Detecting case-control expression quantitative trait loci using locally most powerful or maximin robust rank tests

In testing genome-wide gene expression quantitative trait loci, efficiency robust statistical methods and their computational convenience are most relevant. For this purpose, we propose to use a modified locally most powerful rank test for the analysis of case-control expression data. This modified rank test statistic is computationally simple, robust for non-normally distributed expression data, and asymptotically locally most powerful. It depends on the specification of a location distribution form for data but is not sensitive to misspecifications. When such a location distribution form cannot be specified, we apply Gastwirth's maximin efficiency robust rank test to gene expression data to maximize the worst Pitman asymptotic relative efficiency among a family of location distributions. We conduct simulation studies to assess their performance and use an application to real data for illustration.

Joint analysis of binary and quantitative traits with data sharing and outcome-dependent sampling

We study the analysis of a joint association between a genetic marker with both binary (case-control) and quantitative (continuous) traits, where the quantitative trait values are only available for the cases due to data sharing and outcome-dependent sampling. Data sharing becomes common in genetic association studies, and the outcome-dependent sampling is the consequence of data sharing, under which a phenotype of interest is not measured for some subgroup. The trend test (or Pearson's test) and F-test are often, respectively, used to analyze the binary and quantitative traits. Because of the outcome-dependent sampling, the usual F-test can be applied using the subgroup with the observed quantitative traits. We propose a modified F-test by also incorporating the genotype frequencies of the subgroup whose traits are not observed. Further, a combination of this modified F-test and Pearson's test is proposed by Fisher's combination of their P-values as a joint analysis. Because of the correlation of the two analyses, we propose to use a Gamma (scaled chi-squared) distribution to fit the asymptotic null distribution for the joint analysis. The proposed modified F-test and the joint analysis can also be applied to test single trait association (either binary or quantitative trait). Through simulations, we identify the situations under which the proposed tests are more powerful than the existing ones. Application to a real dataset of rheumatoid arthritis is presented.

Single marker association analysis for unrelated samples

Methods for single marker association analysis are presented for binary and quantitative traits. For a binary trait, we focus on the analysis of retrospective case-control data using Pearson's chi-squared test, the trend test, and a robust test. For a continuous trait, typical methods are based on a linear regression model or the analysis of variance. We illustrate how these tests can be applied using a public available R package "Rassoc" and some existing R functions. Guidelines for choosing these test statistics are provided.

Robust association tests under different genetic models, allowing for binary or quantitative traits and covariates

The association of genetic variants with outcomes is usually assessed under an additive model, for example by the trend test. However, misspecification of the genetic model will lead to a reduction in power. More robust tests for association might therefore be preferred. A useful approach is to consider the maximum of the three test statistics under additive, dominant and recessive models (MAX3). The p-value however has to be adjusted to maintain the type I error rate. Previous studies and software on robust association tests have focused on binary traits without covariates. In this study we developed an analytic approach to robust association tests using MAX3, allowing for quantitative or binary traits as well as covariates. The p-values from our theoretical calculations match very well with those from a bootstrap resampling procedure. The methodology is implemented in the R package RobustSNP which is able to handle both small-scale studies and GWAS. The package and documentation are available at http://sites.google.com/site/honcheongso/software/robustsnp.

Robust tests for matched case-control genetic association studies

BACKGROUND

The Cochran-Armitage trend test (CATT) is powerful in detecting association between a susceptible marker and a disease. This test, however, may suffer from a substantial loss of power when the underlying genetic model is unknown and incorrectly specified. Thus, it is useful to derive tests obtaining the plausible power against all common genetic models. For this purpose, the genetic model selection (GMS) and genetic model exclusion (GME) methods were proposed recently. Simulation results showed that GMS and GME can obtain the plausible power against three common genetic models while the overall type I error is well controlled.

RESULTS

Although GMS and GME are powerful statistically, they could be seriously affected by known confounding factors such as gender, age and race. Therefore, in this paper, via comparing the difference of Hardy-Weinberg disequilibrium coefficients between the cases and the controls within each sub-population, we propose the stratified genetic model selection (SGMS) and exclusion (SGME) methods which could eliminate the effect of confounding factors by adopting a matching framework. Our goal in this paper is to investigate the robustness of the proposed statistics and compare them with other commonly used efficiency robust tests such as MAX3 and χ2 with 2 degrees of freedom (df) test in matched case-control association designs through simulation studies.

CONCLUSION

Simulation results showed that if the mean genetic effect of the heterozygous genotype is between those of the two homozygous genotypes, then the proposed tests and MAX3 are preferred. Otherwise, χ2 with 2 df test may be used. To illustrate the robust procedures, the proposed tests are applied to a real matched pair case-control etiologic study of sarcoidosis.