Differentially expressed heterogeneous overdispersion genes testing for count data.

The mRNA-seq data analysis is a powerful technology for inferring information from biological systems of interest. Specifically, the sequenced RNA fragments are aligned with genomic reference sequences, and we count the number of sequence fragments corresponding to each gene for each condition. A gene is identified as differentially expressed (DE) if the difference in its count numbers between conditions is statistically significant. Several statistical analysis methods have been developed to detect DE genes based on RNA-seq data. However, the existing methods could suffer decreasing power to identify DE genes arising from overdispersion and limited sample size, where overdispersion refers to the empirical phenomenon that the variance of read counts is larger than the mean of read counts. We propose a new differential expression analysis procedure: heterogeneous overdispersion genes testing (DEHOGT) based on heterogeneous overdispersion modeling and a post-hoc inference procedure. DEHOGT integrates sample information from all conditions and provides a more flexible and adaptive overdispersion modeling for the RNA-seq read count. DEHOGT adopts a gene-wise estimation scheme to enhance the detection power of differentially expressed genes when the number of replicates is limited as long as the number of conditions is large. DEHOGT is tested on the synthetic RNA-seq read count data and outperforms two popular existing methods, DESeq2 and EdgeR, in detecting DE genes. We apply the proposed method to a test dataset using RNAseq data from microglial cells. DEHOGT tends to detect more differently expressed genes potentially related to microglial cells under different stress hormones treatments.

Sleep disturbances and interrelationship between persons with dementia and family caregivers: The lived experiences of Korean American Dyads.

BACKGROUND: Little research has investigated sleep quality in dyadic interrelationships between persons with dementia (PWD) and family caregivers, particularly among immigrant ethnic minorities, such as Korean Americans. PURPOSE: The study aimed to describe lived experiences of sleep disturbances and sleep interrelationships between Korean American PWD and their family caregivers. METHODS: A descriptive qualitative design used semi-structured interviews with cohabitating PWD-caregiver dyads. RESULTS: Eleven Korean American dyads participated (PWD mean age: 82.7, SD=2.3; caregivers mean age: 69.1, SD=10.2). Major themes included (1) linked sleep disturbances between PWD and caregivers, (2) interrelationship in dyads, (3) language challenges within and outside the dyads, and (4) strategies that improve sleep quality for dyads. CONCLUSION: Findings demonstrated bidirectional influences in dyadic sleep disturbances, where caregiving reciprocally impacted PWD sleep as part of an interactional unit. Communication barriers and limited community resources posed challenges for these dyads. Future sleep interventions should consider culturally competent, dyadic approaches.

Longitudinal changes in objective sleep parameters during pregnancy.

BACKGROUND: Sleep disturbances are associated with adverse perinatal outcomes. Thus, it is necessary to understand the continuous patterns of sleep during pregnancy and how moderators such as maternal age and pre-pregnancy body mass index impact sleep. OBJECTIVE: This study aimed to examine the continuous changes in sleep parameters objectively (i.e. sleep stages, total sleep time, and awake time) in pregnant women and to describe the impact of maternal age and/or pre-pregnancy body mass index as moderators of these objective sleep parameters. DESIGN: This was a longitudinal observational design. METHODS: Seventeen women with a singleton pregnancy participated in this study. Mixed model repeated measures were used to describe weekly patterns, while aggregated changes describe these three pregnancy periods (10-19, 20-29, and 30-39 gestational weeks). RESULTS: For the weekly patterns, we found significantly decreased deep (1.26 ± 0.18 min/week, p < 0.001), light (0.72 ± 0.37 min/week, p = 0.05), and total sleep time (1.56 ± 0.47 min/week, p < 0.001) as well as increased awake time (1.32 ± 0.34 min/week, p < 0.001). For the aggregated changes, we found similar patterns to weekly changes. Women (⩾30 years) had an even greater decrease in deep sleep (1.50 ± 0.22 min/week, p < 0.001) than those younger (0.84 ± 0.29 min/week, p = 0.04). Women who were both overweight/obese and ⩾30 years experienced an increase in rapid eye movement sleep (0.84 ± 0.31 min/week, p = 0.008), but those of normal weight (<30 years) did not. CONCLUSION: This study appears to be the first to describe continuous changes in sleep parameters during pregnancy at home. Our study provides preliminary evidence that sleep parameters could be potential non-invasive physiological markers predicting perinatal outcomes.

Healthcare center clustering for Cox's proportional hazards model by fusion penalty.

There has been growing research interest in developing methodology to evaluate healthcare centers' performance with respect to patient outcomes. Conventional assessments can be conducted using fixed or random effects models, as seen in provider profiling. We propose a new method, using fusion penalty to cluster healthcare centers with respect to a survival outcome. Without any prior knowledge of the grouping information, the new method provides a desirable data-driven approach for automatically clustering healthcare centers into distinct groups based on their performance. An efficient alternating direction method of multipliers algorithm is developed to implement the proposed method. The validity of our approach is demonstrated through simulation studies, and its practical application is illustrated by analyzing data from the national kidney transplant registry.

Differentially Expressed Heterogeneous Overdispersion Genes Testing for Count Data.

The mRNA-seq data analysis is a powerful technology for inferring information from biological systems of interest. Specifically, the sequenced RNA fragments are aligned with genomic reference sequences, and we count the number of sequence fragments corresponding to each gene for each condition. A gene is identified as differentially expressed (DE) if the difference in its count numbers between conditions is statistically significant. Several statistical analysis methods have been developed to detect DE genes based on RNA-seq data. However, the existing methods could suffer decreasing power to identify DE genes arising from overdispersion and limited sample size. We propose a new differential expression analysis procedure: heterogeneous overdispersion genes testing (DEHOGT) based on heterogeneous overdispersion modeling and a post-hoc inference procedure. DEHOGT integrates sample information from all conditions and provides a more flexible and adaptive overdispersion modeling for the RNA-seq read count. DEHOGT adopts a gene-wise estimation scheme to enhance the detection power of differentially expressed genes. DEHOGT is tested on the synthetic RNA-seq read count data and outperforms two popular existing methods, DESeq and EdgeR, in detecting DE genes. We apply the proposed method to a test dataset using RNAseq data from microglial cells. DEHOGT tends to detect more differently expressed genes potentially related to microglial cells under different stress hormones treatments.

HIMA2: high-dimensional mediation analysis and its application in epigenome-wide DNA methylation data.

Mediation analysis plays a major role in identifying significant mediators in the pathway between environmental exposures and health outcomes. With advanced data collection technology for large-scale studies, there has been growing research interest in developing methodology for high-dimensional mediation analysis. In this paper we present HIMA2, an extension of the HIMA method (Zhang in Bioinformatics 32:3150-3154, 2016). First, the proposed HIMA2 reduces the dimension of mediators to a manageable level based on the sure independence screening (SIS) method (Fan in J R Stat Soc Ser B 70:849-911, 2008). Second, a de-biased Lasso procedure is implemented for estimating regression parameters. Third, we use a multiple-testing procedure to accurately control the false discovery rate (FDR) when testing high-dimensional mediation hypotheses. We demonstrate its practical performance using Monte Carlo simulation studies and apply our method to identify DNA methylation markers which mediate the pathway from smoking to reduced lung function in the Coronary Artery Risk Development in Young Adults (CARDIA) Study.

DNA methylation of Nuclear Factor of Activated T Cells 1 mediates the prospective relation between exposure to different traumatic event types and post-traumatic stress disorder.

The mechanisms through which exposure to differing trauma types become biologically embedded to shape the risk for post-traumatic stress disorder (PTSD) is unclear. DNA methylation (5-mC), particularly in stress-relevant genes, may play a role in this relationship. Here, we conducted path analysis using generalized structural equation modeling to investigate whether blood-derived 5-mC in Nuclear Factor of Activated T Cells 1 (NFATC1) mediates the prospective association between each of five different trauma types ("assaultive violence", "other injury or shocking experience", "learning of trauma to loved one", "sudden, unexpected death of a close friend or relative", and "other") and lifetime PTSD. All five trauma types were significantly associated with reduced methylation at NFATC1 CpG site, cg17057218. Two of the five trauma types were significantly associated with increased methylation at NFATC1 CpG site, cg22324981. Moreover, methylation at cg17057218 significantly mediated 21-32% of the total effect for four of the five trauma types, while methylation at cg22324981 mediated 27-40% of the total effect for two of the five trauma types. These CpG sites were differentially associated with transcription factor binding sites and chromatin state signatures. NFATC1 5-mC may be a potential mechanism in the relationship between some trauma types and prospective risk for PTSD.

Heterogeneous Mediation Analysis on Epigenomic PTSD and Traumatic Stress in a Predominantly African American Cohort.

DNA methylation (DNAm) has been suggested to play a critical role in post-traumatic stress disorder (PTSD), through mediating the relationship between trauma and PTSD. However, this underlying mechanism of PTSD for African Americans still remains unknown. To fill this gap, in this article, we investigate how DNAm mediates the effects of traumatic experiences on PTSD symptoms in the Detroit Neighborhood Health Study (DNHS) (2008-2013) which involves primarily African Americans adults. To achieve this, we develop a new mediation analysis approach for high-dimensional potential DNAm mediators. A key novelty of our method is that we consider heterogeneity in mediation effects across subpopulations. Specifically, mediators in different subpopulations could have opposite effects on the outcome, and thus could be difficult to identify under a traditional homogeneous model framework. In contrast, the proposed method can estimate heterogeneous mediation effects and identifies subpopulations in which individuals share similar effects. Simulation studies demonstrate that the proposed method outperforms existing methods for both homogeneous and heterogeneous data. We also present our mediation analysis results of a dataset with 125 participants and more than 450,000 CpG sites from the DNHS study. The proposed method finds three subgroups of subjects and identifies DNAm mediators corresponding to genes such as HSP90AA1 and NFATC1 which have been linked to PTSD symptoms in literature. Our finding could be useful in future finer-grained investigation of PTSD mechanism and in the development of new treatments for PTSD.

Endogenous opioids contribute to the feeling of pain relief in humans.

Endogenous opioids mediate the pleasurable responses to positively reinforcing stimuli such as palatable food. Yet, the reduction or omission of a negative experience can also be rewarding (negative reinforcement). As such, pain relief leads to negative reinforcement and evokes a pleasant feeling in humans. Although it has been shown that the feeling of pleasure associated with positive reinforcement is at least partly mediated through endogenous opioids, it is currently unknown whether similar neurochemical mechanisms are involved in the pleasant feeling evoked by pain relief. In this study, 27 healthy participants completed 2 identical experimental sessions, 1 with placebo and 1 with naltrexone, an endogenous opioid antagonist. Pain relief was induced by superficial cooling after heat stimulation of capsaicin-sensitized skin. Participants rated the relief and pleasantness in response to the cooling. Endogenous opioid blockade by naltrexone decreased relief and pleasantness ratings compared with placebo (P = 0.0027). This study provides evidence that endogenous opioids play a role in mediating the pleasant feeling of pain relief in humans. Clinically, the rewarding nature of pain relief and its underlying mechanisms require consideration because of their potential reinforcing effects on behaviors that might be beneficial short-term but maladaptive long-term.

The impact of psychopathology, social adversity and stress-relevant DNA methylation on prospective risk for post-traumatic stress: A machine learning approach.

BACKGROUND: A range of factors have been identified that contribute to greater incidence, severity, and prolonged course of post-traumatic stress disorder (PTSD), including: comorbid and/or prior psychopathology; social adversity such as low socioeconomic position, perceived discrimination, and isolation; and biological factors such as genomic variation at glucocorticoid receptor regulatory network (GRRN) genes. This complex etiology and clinical course make identification of people at higher risk of PTSD challenging. Here we leverage machine learning (ML) approaches to identify a core set of factors that may together predispose persons to PTSD. METHODS: We used multiple ML approaches to assess the relationship among DNA methylation (DNAm) at GRRN genes, prior psychopathology, social adversity, and prospective risk for PTS severity (PTSS). RESULTS: ML models predicted prospective risk of PTSS with high accuracy. The Gradient Boost approach was the top-performing model with mean absolute error of 0.135, mean square error of 0.047, root mean square error of 0.217, and R2 of 95.29%. Prior PTSS ranked highest in predicting the prospective risk of PTSS, accounting for >88% of the prediction. The top ranked GRRN CpG site was cg05616442, in AKT1, and the top ranked social adversity feature was loneliness. CONCLUSION: Multiple factors including prior PTSS, social adversity, and DNAm play a role in predicting prospective risk of PTSS. ML models identified factors accounting for increased PTSS risk with high accuracy, which may help to target risk factors that reduce the likelihood or course of PTSD, potentially pointing to approaches that can lead to early intervention. LIMITATION: One of the limitations of this study is small sample size.

Evaluating the impact of trauma and PTSD on epigenetic prediction of lifespan and neural integrity.

Post-traumatic stress disorder (PTSD) is a debilitating disorder that develops in some people following trauma exposure. Trauma and PTSD have been associated with accelerated cellular aging. This study evaluated the effect of trauma and PTSD on accelerated GrimAge, an epigenetic predictor of lifespan, in traumatized civilians. This study included 218 individuals with current PTSD, 427 trauma-exposed controls without any history of PTSD and 209 subjects with lifetime PTSD history who are not categorized as current PTSD cases. The Traumatic Events Inventory (TEI) and Clinician-Administered PTSD Scale (CAPS) were used to measure lifetime trauma burden and PTSD, respectively. DNA from whole blood was interrogated using the MethylationEPIC or HumanMethylation450 BeadChips. GrimAge estimates were calculated using the methylation age calculator. Cortical thickness of 69 female subjects was assessed by using T1-weighted structural MRI images. Associations between trauma exposure, PTSD, cortical thickness, and GrimAge acceleration were tested with multiple regression models. Lifetime trauma burden (p = 0.03), current PTSD (p = 0.02) and lifetime PTSD (p = 0.005) were associated with GrimAge acceleration, indicative of a shorter predicted lifespan. The association with lifetime PTSD was replicated in an independent cohort (p = 0.04). In the MRI sub sample, GrimAge acceleration also associated with cortical atrophy in the right lateral orbitofrontal cortex (padj = 0.03) and right posterior cingulate (padj = 0.04), brain areas associated with emotion-regulation and threat-regulation. Our findings suggest that lifetime trauma and PTSD may contribute to a higher epigenetic-based mortality risk. We also demonstrate a relationship between cortical atrophy in PTSD-relevant brain regions and shorter predicted lifespan.

Subgroup analysis based on structured mixed-effects models for longitudinal data.

In recent years, subgroup analysis has emerged as an important tool to identify unknown subgroup memberships. However, subgroup analysis is still under-studied for longitudinal data. In this paper, we propose a structured mixed-effects approach for longitudinal data to model subgroup distribution and identify subgroup membership simultaneously. In the proposed structured mixed-effects model, the heterogeneous treatment effect is modeled as a random effect from a two-component mixture model, while the membership of the mixture model is incorporated using a logistic model with respect to some covariates. One advantage of our approach is that we are able to derive the estimation of the treatment effects through an EM-type algorithm which keeps the subgroup membership unchanged over time. Our numerical studies and real data example demonstrate that the proposed model outperforms other competing methods.

Time-varying feature selection for longitudinal analysis.

We propose time-varying coefficient model selection and estimation based on the spline approach, which is capable of capturing time-dependent covariate effects. The new penalty function utilizes local-region information for varying-coefficient estimation, in contrast to the traditional model selection approach focusing on the entire region. The proposed method is extremely useful when the signals associated with relevant predictors are time-dependent, and detecting relevant covariate effects in the local region is more scientifically relevant than those of the entire region. Our simulation studies indicate that the proposed model selection incorporating local features outperforms the global feature model selection approaches. The proposed method is also illustrated through a longitudinal growth and health study from National Heart, Lung, and Blood Institute.

Feasibility and validity of a statistical adjustment to reduce self-report bias of height and weight in wave 1 of the Add Health study.

BACKGROUND: Bias in adolescent self-reported height and weight is well documented. Given the importance and widespread use of the National Longitudinal Study of Adolescent to Adult Health (Add Health) data for obesity research, we developed and tested the feasibility and validity of an empirically derived statistical correction for self-report bias in wave 1 (W1) of Add Health, a large panel study in the United States. METHODS: Participants in grades 7-12 with complete height and weight data at W1 were included (n = 20,175). We used measured and self-reported (SR) height and weight and relevant biopsychosocial factors from wave 2 (W2) of Add Health (n = 14,190) to identify sources of bias and derive the most efficient sex-specific estimates of corrected height and weight. Measured, SR, and corrected W2 BMI values were calculated and compared, including sensitivity and specificity. Final correction equations were applied to W1. RESULTS: After correction, weight status misclassification rates among those who underestimated their weight status were reduced from 6.6 to 5.7 % for males and from 8.0 to 5.6 % for females compared to self-report; and the correlation between SR and measured BMI in W2 increased slightly from 0.92 to 0.93. Among females, correction procedures resulted in a 3.4 % increase in sensitivity to detect overweight/obesity (BMI ≥ 25) and 5.9 % increase in sensitivity for obesity (BMI ≥ 30). CONCLUSIONS: Findings suggest that application of the proposed statistical corrections can reduce bias of self-report height and weight in W1 of the Add Health data and may be useful in some analyses. In particular, the corrected BMI values improve sensitivity --the ability to detect a true positive-for overweight/obesity among females, which addresses a major concern about self-report bias in obesity research. However, the correction does not improve sensitivity to identify underweight or healthy weight adolescents and so should be applied selectively based on research questions.

Individualizing drug dosage with longitudinal data.

We propose a two-step procedure to personalize drug dosage over time under the framework of a log-linear mixed-effect model. We model patients' heterogeneity using subject-specific random effects, which are treated as the realizations of an unspecified stochastic process. We extend the conditional quadratic inference function to estimate both fixed-effect coefficients and individual random effects on a longitudinal training data sample in the first step and propose an adaptive procedure to estimate new patients' random effects and provide dosage recommendations for new patients in the second step. An advantage of our approach is that we do not impose any distribution assumption on estimating random effects. Moreover, the new approach can accommodate more general time-varying covariates corresponding to random effects. We show in theory and numerical studies that the proposed method is more efficient compared with existing approaches, especially when covariates are time varying. In addition, a real data example of a clozapine study confirms that our two-step procedure leads to more accurate drug dosage recommendations. Copyright © 2016 John Wiley & Sons, Ltd.

Estimating and Identifying Unspecified Correlation Structure for Longitudinal Data.

Identifying correlation structure is important to achieving estimation efficiency in analyzing longitudinal data, and is also crucial for drawing valid statistical inference for large size clustered data. In this paper, we propose a nonparametric method to estimate the correlation structure, which is applicable for discrete longitudinal data. We utilize eigenvector-based basis matrices to approximate the inverse of the empirical correlation matrix and determine the number of basis matrices via model selection. A penalized objective function based on the difference between the empirical and model approximation of the correlation matrices is adopted to select an informative structure for the correlation matrix. The eigenvector representation of the correlation estimation is capable of reducing the risk of model misspecification, and also provides useful information on the specific within-cluster correlation pattern of the data. We show that the proposed method possesses the oracle property and selects the true correlation structure consistently. The proposed method is illustrated through simulations and two data examples on air pollution and sonar signal studies.

Diagnosis of Basal-Like Breast Cancer Using a FOXC1-Based Assay.

BACKGROUND: Diagnosis of basal-like breast cancer (BLBC) remains a bottleneck to conducting effective clinical trials for this aggressive subtype. We postulated that elevated expression of Forkhead Box transcription factor C1 (FOXC1) is a simple and accurate diagnostic biomarker for BLBC. METHODS: Accuracy of FOXC1 expression in identifying BLBC was compared with the PAM50 gene expression panel in gene expression microarray (GEM) (n = 1992) and quantitative real-time polymerase chain reaction (qRT-PCR) (n = 349) datasets. A FOXC1-based immunohistochemical (IHC) assay was developed and assessed in 96 archival formalin-fixed, paraffin-embedded (FFPE) breast cancer samples that also underwent PAM50 profiling. All statistical tests were two-sided. RESULTS: A FOXC1-based two-tier assay (IHC +/- qRT-PCR) accurately identified BLBC (AUC = 0.88) in an independent cohort of FFPE samples, validating the accuracy of FOXC1-defined BLBC in GEM (AUC = 0.90) and qRT-PCR (AUC = 0.88) studies, when compared with platform-specific PAM50-defined BLBC. The hazard ratio (HR) for disease-specific survival in patients having FOXC1-defined BLBC was 1.71 (95% CI = 1.31 to 2.23, P < .001), comparable to PAM50 assay-defined BLBC (HR = 1.74, 95% CI = 1.40 to 2.17, P < .001). FOXC1 expression also predicted the development of brain metastasis. Importantly, unlike triple-negative or Core Basal IHC definitions, a FOXC1-based definition is able to identify BLBC in both ER+ and HER2+ patients. CONCLUSION: A FOXC1-based two-tier assay, by virtue of being rapid, simple, accurate, and cost-effective may emerge as the diagnostic assay of choice for BLBC. Such a test could substantially improve clinical trial enrichment of BLBC patients and accelerate the identification of effective chemotherapeutic options for this aggressive disease.

Pulmonary disease after treatment for Wilms tumor: a report from the national wilms tumor long-term follow-up study.

PURPOSE: This study was undertaken to evaluate the incidence of pulmonary disease among patients treated with radiation therapy (RT) for pulmonary metastases (PM) from Wilms tumor (WT). PATIENTS AND METHODS: We reviewed records of 6,449 patients treated on National Wilms Tumor Studies-1, -2, -3, and -4 whose flow sheets or annual status reports documented one of several pulmonary conditions. Cases were fully evaluable if pulmonary function test (PFT) results were available, pulmonary fibrosis was identified on a chest radiograph or was listed as the primary or a contributing factor to death. Partially evaluable cases were those for whom PFT results could not be obtained. We evaluated the relationship between RT factors and the occurrence of pulmonary disease using hazard ratios (HRs) and cumulative incidence, treating death as a competing risk. RESULTS: Sixty-four fully evaluable and 16 partially evaluable cases of pulmonary disease were identified. The cumulative incidence of pulmonary disease at 15 years since WT diagnosis was 4.0% (95% confidence interval [CI] 2.6-5.4%) among fully evaluable and 4.8% (95% CI 3.3-6.4%) among fully and partially evaluable patients who received lung RT for PM at initial diagnosis. Rates of pulmonary disease were substantially higher among those who received lung RT for PM present at initial diagnosis or relapse compared to those who received no RT or only abdominal RT (HR 30.2, 95% CI 16.9-53.9). CONCLUSION: The risk of pulmonary disease must be considered in evaluating the risk:benefit ratio of lung RT for the management of PM from WT.

Smoking cessation programs targeted to women: a systematic review.

The authors of this systematic review aimed to examine tobacco interventions developed to meet the needs of women, to identify sex- and gender-specific components, and to evaluate their effects on smoking cessation in women. The authors searched electronic databases in the Cochrane Central Register of Controlled Trials, MEDLINE, PubMed, EBSCO, PsychINFO, CINHAL, and EMBASE; the search was not restricted by publication date. Data was extracted from published peer-reviewed articles on participants, setting, treatment models, interventions, length of follow-up, and outcomes. The main outcome variable was abstinence from smoking. A total of 39 studies were identified. In efficacy studies, therapists addressed weight concerns and non-pharmacological aspects of smoking, taught mood/stress management strategies, and scheduled the quit date to be timed to the menstrual cycle. In effectiveness studies, therapists were peer counselors, provided telephone counseling, and/or distributed gendered booklets, videos, and posters. Among efficacy studies, interventions addressing weight gain/concerns showed the most promising results. If medication can support smoking cessation in women and how it interacts with non-pharmacological treatment also warrant further research. For effectiveness studies, the available evidence suggests that smoking should be addressed in low-income women accessing public health clinics. Further attention should be devoted to identifying new settings for providing smoking cessation interventions to women from disadvantaged groups. Women-specific tobacco programs help women stop smoking, although they appear to produce similar abstinence rates as non-sex/gender specific programs. Offering interventions for women specifically may reduce barriers to treatment entry and better meet individual preferences of smokers. Developing approaches that fully account for the multiple challenges treatment-seeking women face is still an area of research.

Penalized generalized estimating equations for high-dimensional longitudinal data analysis.

We consider the penalized generalized estimating equations (GEEs) for analyzing longitudinal data with high-dimensional covariates, which often arise in microarray experiments and large-scale health studies. Existing high-dimensional regression procedures often assume independent data and rely on the likelihood function. Construction of a feasible joint likelihood function for high-dimensional longitudinal data is challenging, particularly for correlated discrete outcome data. The penalized GEE procedure only requires specifying the first two marginal moments and a working correlation structure. We establish the asymptotic theory in a high-dimensional framework where the number of covariates p(n) increases as the number of clusters n increases, and p(n) can reach the same order as n. One important feature of the new procedure is that the consistency of model selection holds even if the working correlation structure is misspecified. We evaluate the performance of the proposed method using Monte Carlo simulations and demonstrate its application using a yeast cell-cycle gene expression data set.