A new system identification approach to identify genetic variants in sequencing studies for a binary phenotype

GMEPS: a fast and efficient likelihood approach for genome-wide mediation analysis under extreme phenotype sequencing

Due to many advantages such as higher statistical power of detecting the association of genetic variants in human disorders and cost saving, extreme phenotype sequencing (EPS) is a rapidly emerging study design in epidemiological and clinical studies investigating how genetic variations associate with complex phenotypes. However, the investigation of the mediation effect of genetic variants on phenotypes is strictly restrictive under the EPS design because existing methods cannot well accommodate the non-random extreme tails sampling process incurred by the EPS design. In this paper, we propose a likelihood approach for testing the mediation effect of genetic variants through continuous and binary mediators on a continuous phenotype under the EPS design (GMEPS). Besides implementing in EPS design, it can also be utilized as a general mediation analysis procedure. Extensive simulations and two real data applications of a genome-wide association study of benign ethnic neutropenia under EPS design and a candidate-gene study of neurocognitive performance in patients with sickle cell disease under random sampling design demonstrate the superiority of GMEPS under the EPS design over widely used mediation analysis procedures, while demonstrating compatible capabilities under the general random sampling framework.

STEPS: an efficient prospective likelihood approach to genetic association analyses of secondary traits in extreme phenotype sequencing

It has been well acknowledged that methods for secondary trait (ST) association analyses under a case-control design (ST$_{\text{CC}}$) should carefully consider the sampling process to avoid biased risk estimates. A similar situation also exists in the extreme phenotype sequencing (EPS) designs, which is to select subjects with extreme values of continuous primary phenotype for sequencing. EPS designs are commonly used in modern epidemiological and clinical studies such as the well-known National Heart, Lung, and Blood Institute Exome Sequencing Project. Although naïve generalized regression or ST$_{\text{CC}}$ method could be applied, their validity is questionable due to difference in statistical designs. Herein, we propose a general prospective likelihood framework to perform association testing for binary and continuous STs under EPS designs (STEPS), which can also incorporate covariates and interaction terms. We provide a computationally efficient and robust algorithm to obtain the maximum likelihood estimates. We also present two empirical mathematical formulas for power/sample size calculations to facilitate planning of binary/continuous STs association analyses under EPS designs. Extensive simulations and application to a genome-wide association study of benign ethnic neutropenia under an EPS design demonstrate the superiority of STEPS over all its alternatives above.

Extreme-value sampling design is cost-beneficial only with a valid statistical approach for exposure-secondary outcome association analyses

In epidemiology cohort studies, exposure data are collected in sub-studies based on a primary outcome (PO) of interest, as with the extreme-value sampling design (EVSD), to investigate their correlation. Secondary outcomes (SOs) data are also readily available, enabling researchers to assess the correlations between the exposure and the SOs. However, when the EVSD is used, the data for SOs are not representative samples of a general population; thus, many commonly used statistical methods, such as the generalized linear model (GLM), are not valid. A prospective likelihood method has been developed to associate SOs with single-nucleotide polymorphisms under an extreme phenotype sequencing design. In this paper, we describe the application of the prospective likelihood method (ST_EVSD) to exposure-SO association analysis under an EVSD. We undertook extensive simulations to assess the performance of the ST_EVSD method in associating binary and continuous exposures with SOs, comparing it to the simple GLM method that ignores the EVSD. To demonstrate the cost-benefit of the ST_EVSD method, we also mimicked the design of two new retrospective studies, as would be done in actual practice, based on the PO of interest, which was the same as the SO in the EVSD study. We then analyzed these data by using the GLM method and compared its power to that of the ST_EVSD method. We demonstrated the usefulness of the ST_EVSD method by applying it to a benign ethnic neutropenia dataset. Our results indicate that the ST_EVSD method can control type I error well, whereas the GLM method cannot do so owing to its ignorance of EVSD, and that the ST_EVSD method is cost-effective because it has statistical power similar to that of two new retrospective studies that require collecting new exposure data for selected individuals.

Statistical selection of biological models for genome-wide association analyses

Genome-wide association studies have discovered many biologically important associations of genes with phenotypes. Typically, genome-wide association analyses formally test the association of each genetic feature (SNP, CNV, etc) with the phenotype of interest and summarize the results with multiplicity-adjusted p-values. However, very small p-values only provide evidence against the null hypothesis of no association without indicating which biological model best explains the observed data. Correctly identifying a specific biological model may improve the scientific interpretation and can be used to more effectively select and design a follow-up validation study. Thus, statistical methodology to identify the correct biological model for a particular genotype-phenotype association can be very useful to investigators. Here, we propose a general statistical method to summarize how accurately each of five biological models (null, additive, dominant, recessive, co-dominant) represents the data observed for each variant in a GWAS study. We show that the new method stringently controls the false discovery rate and asymptotically selects the correct biological model. Simulations of two-stage discovery-validation studies show that the new method has these properties and that its validation power is similar to or exceeds that of simple methods that use the same statistical model for all SNPs. Example analyses of three data sets also highlight these advantages of the new method. An R package is freely available at www.stjuderesearch.org/site/depts/biostats/maew.

A Robust and Powerful Set-Valued Approach to Rare Variant Association Analyses of Secondary Traits in Case-Control Sequencing Studies

In many case-control designs of genome-wide association (GWAS) or next generation sequencing (NGS) studies, extensive data on secondary traits that may correlate and share the common genetic variants with the primary disease are available. Investigating these secondary traits can provide critical insights into the disease etiology or pathology, and enhance the GWAS or NGS results. Methods based on logistic regression (LG) were developed for this purpose. However, for the identification of rare variants (RVs), certain inadequacies in the LG models and algorithmic instability can cause severely inflated type I error, and significant loss of power, when the two traits are correlated and the RV is associated with the disease, especially at stringent significance levels. To address this issue, we propose a novel set-valued (SV) method that models a binary trait by dichotomization of an underlying continuous variable, and incorporate this into the genetic association model as a critical component. Extensive simulations and an analysis of seven secondary traits in a GWAS of benign ethnic neutropenia show that the SV method consistently controls type I error well at stringent significance levels, has larger power than the LG-based methods, and is robust in performance to effect pattern of the genetic variant (risk or protective), rare or common variants, rare or common diseases, and trait distributions. Because of the SV method's striking and profound advantage, we strongly recommend the SV method be employed instead of the LG-based methods for secondary traits analyses in case-control sequencing studies.

Post hoc Analysis for Detecting Individual Rare Variant Risk Associations Using Probit Regression Bayesian Variable Selection Methods in Case-Control Sequencing Studies

Rare variants (RVs) have been shown to be significant contributors to complex disease risk. By definition, these variants have very low minor allele frequencies and traditional single-marker methods for statistical analysis are underpowered for typical sequencing study sample sizes. Multimarker burden-type approaches attempt to identify aggregation of RVs across case-control status by analyzing relatively small partitions of the genome, such as genes. However, it is generally the case that the aggregative measure would be a mixture of causal and neutral variants, and these omnibus tests do not directly provide any indication of which RVs may be driving a given association. Recently, Bayesian variable selection approaches have been proposed to identify RV associations from a large set of RVs under consideration. Although these approaches have been shown to be powerful at detecting associations at the RV level, there are often computational limitations on the total quantity of RVs under consideration and compromises are necessary for large-scale application. Here, we propose a computationally efficient alternative formulation of this method using a probit regression approach specifically capable of simultaneously analyzing hundreds to thousands of RVs. We evaluate our approach to detect causal variation on simulated data and examine sensitivity and specificity in instances of high RV dimensionality as well as apply it to pathway-level RV analysis results from a prostate cancer (PC) risk case-control sequencing study. Finally, we discuss potential extensions and future directions of this work.

Evaluation of a two-step iterative resampling procedure for internal validation of genome-wide association studies

Genome-wide association studies (GWAS) have successfully identified many common genetic variants associated with complex diseases over the past decade. The 'gold standard' method for validating the top single nucleotide polymorphisms (SNPs) identified in GWAS is to independently replicate the findings in similar or diverse large-scale external cohorts. However, for rare diseases, it can be difficult to find an external validation cohort within a reasonable timeframe. In such situations, resampling methods, such as the two-step iterative resampling (TSIR) approach have been used to identify SNPs associated with the outcome of interest. However, the TSIR approach involves choosing several parameters in each step, which can influence the performance of the approach. In this paper, we undertook extensive simulation studies to assess the effect of choice of different parameters on the type I error and power for both binary and continuous phenotypes and also compared the TSIR approach with the traditional one-stage (OS) and two-stage (TS) GWAS analysis. We illustrate the usefulness of the TSIR approach by applying it to a GWAS of childhood cancer survivors. Our results indicate that the TSIR approach with an at least 70:30 split and a cutoff of discovering and replicating SNPs at least 20 times in 100 replications provides conservative type I error control and has near 'optimal' power for internally validated SNPs. Its performance is comparable with the TS GWAS for which an external validation cohort is available with only slight reduction in power in some situations. It has almost the same power as OS GWAS with conservative type I error which leads to fewer false positive findings. TSIR is a powerful and efficient method for identifying and internally validating SNPs for GWAS when independent cohorts for external validation may not be available.

SVSI: fast and powerful set-valued system identification approach to identifying rare variants in sequencing studies for ordered categorical traits

In genetic association studies of an ordered categorical phenotype, it is usual to either regroup multiple categories of the phenotype into two categories and then apply the logistic regression (LG), or apply ordered logistic (oLG), or ordered probit (oPRB) regression, which accounts for the ordinal nature of the phenotype. However, they may lose statistical power or may not control type I error due to their model assumption and/or instable parameter estimation algorithm when the genetic variant is rare or sample size is limited. To solve this problem, we propose a set-valued (SV) system model to identify genetic variants associated with an ordinal categorical phenotype. We couple this model with a SV system identification algorithm to identify all the key system parameters. Simulations and two real data analyses show that SV and LG accurately controlled the Type I error rate even at a significance level of 10(-6) but not oLG and oPRB in some cases. LG had significantly less power than the other three methods due to disregarding of the ordinal nature of the phenotype, and SV had similar or greater power than oLG and oPRB. We argue that SV should be employed in genetic association studies for ordered categorical phenotype.