Kernel-based hierarchical structural component models for pathway analysis.

MOTIVATION: Pathway analyses have led to more insight into the underlying biological functions related to the phenotype of interest in various types of omics data. Pathway-based statistical approaches have been actively developed, but most of them do not consider correlations among pathways. Because it is well known that there are quite a few biomarkers that overlap between pathways, these approaches may provide misleading results. In addition, most pathway-based approaches tend to assume that biomarkers within a pathway have linear associations with the phenotype of interest, even though the relationships are more complex. RESULTS: To model complex effects including non-linear effects, we propose a new approach, Hierarchical structural CoMponent analysis using Kernel (HisCoM-Kernel). The proposed method models non-linear associations between biomarkers and phenotype by extending the kernel machine regression and analyzes entire pathways simultaneously by using the biomarker-pathway hierarchical structure. HisCoM-Kernel is a flexible model that can be applied to various omics data. It was successfully applied to three omics datasets generated by different technologies. Our simulation studies showed that HisCoM-Kernel provided higher statistical power than other existing pathway-based methods in all datasets. The application of HisCoM-Kernel to three types of omics dataset showed its superior performance compared to existing methods in identifying more biologically meaningful pathways, including those reported in previous studies. AVAILABILITY AND IMPLEMENTATION: The HisCoM-Kernel software is freely available at http://statgen.snu.ac.kr/software/HisCom-Kernel/. The RNA-seq data underlying this article are available at https://xena.ucsc.edu/, and the others will be shared on reasonable request to the corresponding author. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Bayesian Extended Redundancy Analysis: A Bayesian Approach to Component-based Regression with Dimension Reduction.

Extended redundancy analysis (ERA) combines linear regression with dimension reduction to explore the directional relationships between multiple sets of predictors and outcome variables in a parsimonious manner. It aims to extract a component from each set of predictors in such a way that it accounts for the maximum variance of outcome variables. In this article, we extend ERA into the Bayesian framework, called Bayesian ERA (BERA). The advantages of BERA are threefold. First, BERA enables to make statistical inferences based on samples drawn from the joint posterior distribution of parameters obtained from a Markov chain Monte Carlo algorithm. As such, it does not necessitate any resampling method, which is on the other hand required for (frequentist's) ordinary ERA to test the statistical significance of parameter estimates. Second, it formally incorporates relevant information obtained from previous research into analyses by specifying informative power prior distributions. Third, BERA handles missing data by implementing multiple imputation using a Markov Chain Monte Carlo algorithm, avoiding the potential bias of parameter estimates due to missing data. We assess the performance of BERA through simulation studies and apply BERA to real data regarding academic achievement.

HisCoM-G×E: Hierarchical Structural Component Analysis of Gene-Based Gene-Environment Interactions.

Gene-environment interaction (G×E) studies are one of the most important solutions for understanding the "missing heritability" problem in genome-wide association studies (GWAS). Although many statistical methods have been proposed for detecting and identifying G×E, most employ single nucleotide polymorphism (SNP)-level analysis. In this study, we propose a new statistical method, Hierarchical structural CoMponent analysis of gene-based Gene-Environment interactions (HisCoM-G×E). HisCoM-G×E is based on the hierarchical structural relationship among all SNPs within a gene, and can accommodate all possible SNP-level effects into a single latent variable, by imposing a ridge penalty, and thus more efficiently takes into account the latent interaction term of G×E. The performance of the proposed method was evaluated in simulation studies, and we applied the proposed method to investigate gene-alcohol intake interactions affecting systolic blood pressure (SBP), using samples from the Korea Associated REsource (KARE) consortium data.

Assessment of eating attitudes and dieting behaviors in healthy children: Confirmatory factor analysis of the Children's Eating Attitudes Test.

OBJECTIVE: The Children's Eating Attitudes Test (ChEAT) is a self-report questionnaire that is conventionally summarized with a single score to identify "problematic" eating attitudes, masking informative variability in different eating attitude domains. This study evaluated the empirical support for single- versus multifactor models of the ChEAT. For validation, we compared how well the single- versus multifactor-based scores predicted body mass index (BMI). METHOD: Using data from 13,674 participants of the 11.5 year-follow-up of the Promotion of Breastfeeding Intervention Trial (PROBIT) in the Republic of Belarus, we conducted confirmatory factor analysis to evaluate the performance of 3- and 5-factor models, which were based on past studies, to a single-factor model representing the conventional summary of the ChEAT. We used cross-validated linear regression models and the reduction in mean squared error (MSE) to compare the prediction of BMI at 11.5 and 16 years by the conventional and confirmed factor-based ChEAT scores. RESULTS: The 5-factor model, based on 14 of the original 26 ChEAT items, had good fit to the data whereas the 3- and single-factor models did not. The MSE for concurrent (11.5 years) BMI regressed on the 5-factor ChEAT summary was 35% lower than that of the single-score models, which reduced the MSE from the null model by only 1%-5%. The MSE for BMI at 16 years was 20% lower. DISCUSSION: We found that a parsimonious 5-factor model of the ChEAT explained the data collected from healthy Belarusian children better than the conventional summary score and thus provides a more discriminating measure of eating attitudes.

Out-of-bag Prediction Error: A Cross Validation Index for Generalized Structured Component Analysis.

Cross validation is a useful way of comparing predictive generalizability of theoretically plausible a priori models in structural equation modeling (SEM). A number of overall or local cross validation indices have been proposed for existing factor-based and component-based approaches to SEM, including covariance structure analysis and partial least squares path modeling. However, there is no such cross validation index available for generalized structured component analysis (GSCA) which is another component-based approach. We thus propose a cross validation index for GSCA, called Out-of-bag Prediction Error (OPE), which estimates the expected prediction error of a model over replications of so-called in-bag and out-of-bag samples constructed through the implementation of the bootstrap method. The calculation of this index is well-suited to the estimation procedure of GSCA, which uses the bootstrap method to obtain the standard errors or confidence intervals of parameter estimates. We empirically evaluate the performance of the proposed index through the analyses of both simulated and real data.

Pathway-based approach using hierarchical components of rare variants to analyze multiple phenotypes.

BACKGROUND: As one possible solution to the "missing heritability" problem, many methods have been proposed that apply pathway-based analyses, using rare variants that are detected by next generation sequencing technology. However, while a number of methods for pathway-based rare-variant analysis of multiple phenotypes have been proposed, no method considers a unified model that incorporate multiple pathways. RESULTS: Simulation studies successfully demonstrated advantages of multivariate analysis, compared to univariate analysis, and comparison studies showed the proposed approach to outperform existing methods. Moreover, real data analysis of six type 2 diabetes-related traits, using large-scale whole exome sequencing data, identified significant pathways that were not found by univariate analysis. Furthermore, strong relationships between the identified pathways, and their associated metabolic disorder risk factors, were found via literature search, and one of the identified pathway, was successfully replicated by an analysis with an independent dataset. CONCLUSIONS: Herein, we present a powerful, pathway-based approach to investigate associations between multiple pathways and multiple phenotypes. By reflecting the natural hierarchy of biological behavior, and considering correlation between pathways and phenotypes, the proposed method is capable of analyzing multiple phenotypes and multiple pathways simultaneously.

Pathway-based approach using hierarchical components of collapsed rare variants.

MOTIVATION: To address 'missing heritability' issue, many statistical methods for pathway-based analyses using rare variants have been proposed to analyze pathways individually. However, neglecting correlations between multiple pathways can result in misleading solutions, and pathway-based analyses of large-scale genetic datasets require massive computational burden. We propose a Pathway-based approach using HierArchical components of collapsed RAre variants Of High-throughput sequencing data (PHARAOH) for the analysis of rare variants by constructing a single hierarchical model that consists of collapsed gene-level summaries and pathways and analyzes entire pathways simultaneously by imposing ridge-type penalties on both gene and pathway coefficient estimates; hence our method considers the correlation of pathways without constraint by a multiple testing problem. RESULTS: Through simulation studies, the proposed method was shown to have higher statistical power than the existing pathway-based methods. In addition, our method was applied to the large-scale whole-exome sequencing data with levels of a liver enzyme using two well-known pathway databases Biocarta and KEGG. This application demonstrated that our method not only identified associated pathways but also successfully detected biologically plausible pathways for a phenotype of interest. These findings were successfully replicated by an independent large-scale exome chip study. AVAILABILITY AND IMPLEMENTATION: An implementation of PHARAOH is available at http://statgen.snu.ac.kr/software/pharaoh/ CONTACT: tspark@stats.snu.ac.kr SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Two-Way Regularized Fuzzy Clustering of Multiple Correspondence Analysis.

Multiple correspondence analysis (MCA) is a useful tool for investigating the interrelationships among dummy-coded categorical variables. MCA has been combined with clustering methods to examine whether there exist heterogeneous subclusters of a population, which exhibit cluster-level heterogeneity. These combined approaches aim to classify either observations only (one-way clustering of MCA) or both observations and variable categories (two-way clustering of MCA). The latter approach is favored because its solutions are easier to interpret by providing explicitly which subgroup of observations is associated with which subset of variable categories. Nonetheless, the two-way approach has been built on hard classification that assumes observations and/or variable categories to belong to only one cluster. To relax this assumption, we propose two-way fuzzy clustering of MCA. Specifically, we combine MCA with fuzzy k-means simultaneously to classify a subgroup of observations and a subset of variable categories into a common cluster, while allowing both observations and variable categories to belong partially to multiple clusters. Importantly, we adopt regularized fuzzy k-means, thereby enabling us to decide the degree of fuzziness in cluster memberships automatically. We evaluate the performance of the proposed approach through the analysis of simulated and real data, in comparison with existing two-way clustering approaches.

Pathway-based association study of multiple candidate genes and multiple traits using structural equation models.

There is increasing interest in the joint analysis of multiple genetic variants from multiple genes and multiple correlated quantitative traits in association studies. The classical approach involves testing univariate associations between genotypes and phenotypes and correcting for multiple testing that results in loss of power to detect associations. In this paper, we propose modeling complex relationships between genetic variants in candidate genes and measured correlated traits using structural equation models (SEM), taking advantage of prior knowledge on clinical and genetic pathways. We adopt generalized structured component analysis (GSCA) as an approach to SEM and develop a single association test between multiple genetic variants in a gene and a set of correlated traits, taking into account all available data from other genes and other traits. The performance of this test is investigated by simulations. We apply the proposed method to the Quebec Child and Adolescent Health and Social Survey (1999) data to investigate genetic associations with cardiovascular disease-related traits.

Symptom changes in five dimensions of the Positive and Negative Syndrome Scale in refractory psychosis.

Refractory psychosis units currently have little information regarding which symptoms profiles should be expected to respond to treatment. In the current study, we provide this information using structural equation modeling of Positive and Negative Syndrome Scale (PANSS) ratings at admission and discharge on a sample of 610 patients admitted to a treatment refractory psychosis program at a Canadian tertiary care unit between 1990 and 2011. The hypothesized five-dimensional structure of the PANSS fit the data well at both admission and discharge, and the latent variable scores are reported as a function of symptom dimension and diagnostic category. The results suggest that, overall, positive symptoms (POS) responded to treatment better than all other symptoms dimensions, but for the schizoaffective and bipolar groups, greater response on POS was observed relative to the schizophrenia and major depression groups. The major depression group showed the most improvement on negative symptoms and emotional distress, and the bipolar group showed the most improvement on disorganization. Schizophrenia was distinct from schizoaffective disorder in showing reduced treatment response on all symptom dimensions. These results can assist refractory psychosis units by providing information on how PANSS symptom dimensions respond to treatment and how this depends on diagnostic category.

Generalized Structured Component Analysis Accommodating Convex Components: A Knowledge-Based Multivariate Method with Interpretable Composite Indexes.

Generalized structured component analysis (GSCA) is a multivariate method for examining theory-driven relationships between variables including components. GSCA can provide the deterministic component score for each individual once model parameters are estimated. As the traditional GSCA always standardizes all indicators and components, however, it could not utilize information on the indicators' scale in parameter estimation. Consequently, its component scores could just show the relative standing of each individual for a component, rather than the individual's absolute standing in terms of the original indicators' measurement scales. In the paper, we propose a new version of GSCA, named convex GSCA, which can produce a new type of unstandardized components, termed convex components, which can be intuitively interpreted in terms of the original indicators' scales. We investigate the empirical performance of the proposed method through the analyses of simulated and real data.

Gene-environment pathways to cognitive intelligence and psychotic-like experiences in children.

In children, psychotic-like experiences (PLEs) are related to risk of psychosis, schizophrenia, and other mental disorders. Maladaptive cognitive functioning, influenced by genetic and environmental factors, is hypothesized to mediate the relationship between these factors and childhood PLEs. Using large-scale longitudinal data, we tested the relationships of genetic and environmental factors (such as familial and neighborhood environment) with cognitive intelligence and their relationships with current and future PLEs in children. We leveraged large-scale multimodal data of 6,602 children from the Adolescent Brain and Cognitive Development Study. Linear mixed model and a novel structural equation modeling (SEM) method that allows estimation of both components and factors were used to estimate the joint effects of cognitive phenotypes polygenic scores (PGSs), familial and neighborhood socioeconomic status (SES), and supportive environment on NIH Toolbox cognitive intelligence and PLEs. We adjusted for ethnicity (genetically defined), schizophrenia PGS, and additionally unobserved confounders (using computational confound modeling). Our findings indicate that lower cognitive intelligence and higher PLEs are significantly associated with lower PGSs for cognitive phenotypes, lower familial SES, lower neighborhood SES, and less supportive environments. Specifically, cognitive intelligence mediates the effects of these factors on PLEs, with supportive parenting and positive school environments showing the strongest impact on reducing PLEs. This study underscores the influence of genetic and environmental factors on PLEs through their effects on cognitive intelligence. Our findings have policy implications in that improving school and family environments and promoting local economic development may enhance cognitive and mental health in children.

Childhood is a critical period for brain development. Difficult experiences during this developmental phase may contribute to reduced intelligence and poorer mental health later in life. Genetics and environmental factors also play roles. For example, having family support or a higher family income has been linked to better brain health outcomes for children. Delusions or hallucinations, or other psychotic-like experiences during childhood, are linked with poor mental health later in life. Children who experience psychotic-like episodes between the ages of nine and eleven have a higher risk of developing schizophrenia or related conditions. Environmental circumstances during childhood also appear to play a crucial role in shaping the risk of schizophrenia or related conditions. Park, Lee et al. show that positive parenting and supportive school and neighborhood environments boost child intelligence and mental health. In the experiments, Park, Lee et al. analyzed data on 6,602 children to determine how genetics and environmental factors shaped their intelligence and mental health. The models show that children with higher intelligence have a lower risk of psychosis. Both genetics and supportive environments contribute to higher intelligence. Complex interactions between biology and social factors shape children's intelligence and mental health. Beneficial genetics and coming from a family with more financial resources are helpful. Yet, social environments, such as having parents who use positive child-rearing practices, or having supportive schools or neighborhoods, have protective effects that can offset other disadvantages. Policies that help parents, encourage supportive school environments, and strengthen neighborhoods may boost children's intelligence and mental health later in life.

Functional Extended Redundancy Analysis.

We propose a functional version of extended redundancy analysis that examines directional relationships among several sets of multivariate variables. As in extended redundancy analysis, the proposed method posits that a weighed composite of each set of exogenous variables influences a set of endogenous variables. It further considers endogenous and/or exogenous variables functional, varying over time, space, or other continua. Computationally, the method reduces to minimizing a penalized least-squares criterion through the adoption of a basis function expansion approach to approximating functions. We develop an alternating regularized least-squares algorithm to minimize this criterion. We apply the proposed method to real datasets to illustrate the empirical feasibility of the proposed method.

Evaluation of Prediction-Oriented Model Selection Metrics for Extended Redundancy Analysis.

Extended redundancy analysis (ERA) is a statistical method that relates multiple sets of predictors to response variables. In ERA, the conventional approach of model evaluation tends to overestimate the performance of a model since the performance is assessed using the same sample used for model development. To avoid the overly optimistic assessment, we introduce a new model evaluation approach for ERA, which utilizes computer-intensive resampling methods to assess how well a model performs on unseen data. Specifically, we suggest several new model evaluation metrics for ERA that compute a model's performance on out-of-sample data, i.e., data not used for model development. Although considerable work has been done in machine learning and statistics to examine the utility of cross-validation and bootstrap variants for assessing such out-of-sample predictive performance, to date, no research has been carried out in the context of ERA. We use simulated and real data examples to compare the proposed model evaluation approach with the conventional one. Results show the conventional approach always favor more complex ERA models, thereby failing to prevent the problem of overfitting in model selection. Conversely, the proposed approach can select the true ERA model among many mis-specified (i.e., underfitted and overfitted) models.

A comparative evaluation of factor- and component-based structural equation modelling approaches under (in)correct construct representations.

Structural equation modelling (SEM) has evolved into two domains, factor-based and component-based, dependent on whether constructs are statistically represented as common factors or components. The two SEM domains are conceptually distinct, each assuming their own population models with either of the statistical construct proxies, and statistical SEM approaches should be used for estimating models whose construct representations correspond to what they assume. However, SEM approaches have often been evaluated and compared only under population factor models, providing misleading conclusions about their relative performance. This is partly because population component models and their relationships have not been clearly formulated. Also, it is of fundamental importance to examine how robust SEM approaches can be to potential misrepresentation of constructs because researchers may often lack clear theories to determine whether a factor or component is more representative of a given construct. Addressing these issues, this study begins by clarifying several population component models and their relationships and then provides a comprehensive evaluation of four SEM approaches - the maximum likelihood approach and factor score regression for factor-based SEM as well as generalized structured component analysis (GSCA) and partial least squares path modelling (PLSPM) for component-based SEM - under various experimental conditions. We confirm that the factor-based SEM approaches should be preferred for estimating factor models, whereas the component-based SEM approaches should be chosen for component models. Importantly, the component-based approaches are generally more robust to construct misrepresentation than the factor-based ones. Of the component-based approaches, GSCA should be chosen over PLSPM, regardless of whether or not constructs are misrepresented.

[11C]Martinostat PET analysis reveals reduced HDAC I availability in Alzheimer's disease.

Alzheimer's disease (AD) is characterized by the brain accumulation of amyloid-ß and tau proteins. A growing body of literature suggests that epigenetic dysregulations play a role in the interplay of hallmark proteinopathies with neurodegeneration and cognitive impairment. Here, we aim to characterize an epigenetic dysregulation associated with the brain deposition of amyloid-ß and tau proteins. Using positron emission tomography (PET) tracers selective for amyloid-ß, tau, and class I histone deacetylase (HDAC I isoforms 1-3), we find that HDAC I levels are reduced in patients with AD. HDAC I PET reduction is associated with elevated amyloid-ß PET and tau PET concentrations. Notably, HDAC I reduction mediates the deleterious effects of amyloid-ß and tau on brain atrophy and cognitive impairment. HDAC I PET reduction is associated with 2-year longitudinal neurodegeneration and cognitive decline. We also find HDAC I reduction in the postmortem brain tissue of patients with AD and in a transgenic rat model expressing human amyloid-ß plus tau pathology in the same brain regions identified in vivo using PET. These observations highlight HDAC I reduction as an element associated with AD pathophysiology.

Model-based recursive partitioning of extended redundancy analysis with an application to nicotine dependence among US adults.

Extended redundancy analysis (ERA) is used to reduce multiple sets of predictors to a smaller number of components and examine the effects of these components on a response variable. In various social and behavioural studies, auxiliary covariates (e.g., gender, ethnicity) can often lead to heterogeneous subgroups of observations, each of which involves distinctive relationships between predictor and response variables. ERA is currently unable to consider such covariate-dependent heterogeneity to examine whether the model parameters vary across subgroups differentiated by covariates. To address this issue, we combine ERA with model-based recursive partitioning in a single framework. This combined method, MOB-ERA, aims to partition observations into heterogeneous subgroups recursively based on a set of covariates while fitting a specified ERA model to data. Upon the completion of the partitioning procedure, one can easily examine the difference in the estimated ERA parameters across covariate-dependent subgroups. Moreover, it produces a tree diagram that aids in visualizing a hierarchy of partitioning covariates, as well as interpreting their interactions. In the analysis of public data concerning nicotine dependence among US adults, the method uncovered heterogeneous subgroups characterized by several sociodemographic covariates, each of which yielded different directional relationships between three predictor sets and nicotine dependence.

A knowledge-based multivariate statistical method for examining gene-brain-behavioral/cognitive relationships: Imaging genetics generalized structured component analysis.

With advances in neuroimaging and genetics, imaging genetics is a naturally emerging field that combines genetic and neuroimaging data with behavioral or cognitive outcomes to examine genetic influence on altered brain functions associated with behavioral or cognitive variation. We propose a statistical approach, termed imaging genetics generalized structured component analysis (IG-GSCA), which allows researchers to investigate such gene-brain-behavior/cognitive associations, taking into account well-documented biological characteristics (e.g., genetic pathways, gene-environment interactions, etc.) and methodological complexities (e.g., multicollinearity) in imaging genetic studies. We begin by describing the conceptual and technical underpinnings of IG-GSCA. We then apply the approach for investigating how nine depression-related genes and their interactions with an environmental variable (experience of potentially traumatic events) influence the thickness variations of 53 brain regions, which in turn affect depression severity in a sample of Korean participants. Our analysis shows that a dopamine receptor gene and an interaction between a serotonin transporter gene and the environment variable have statistically significant effects on a few brain regions' variations that have statistically significant negative impacts on depression severity. These relationships are largely supported by previous studies. We also conduct a simulation study to safeguard whether IG-GSCA can recover parameters as expected in a similar situation.

An approach to structural equation modeling with both factors and components: Integrated generalized structured component analysis.

In this article, we propose integrated generalized structured component analysis (IGSCA), which is a general statistical approach for analyzing data with both components and factors in the same model, simultaneously. This approach combines generalized structured component analysis (GSCA) and generalized structured component analysis with measurement errors incorporated (GSCAM) in a unified manner and can estimate both factor- and component-model parameters, including component and factor loadings, component and factor path coefficients, and path coefficients connecting factors and components. We conduct 2 simulation studies to investigate the performance of IGSCA under models with both factors and components. The first simulation study assesses how existing approaches for structural equation modeling and IGSCA recover parameters. This study shows that only consistent partial least squares (PLSc) and IGSCA yield unbiased estimates of all parameters, whereas the other approaches always provided biased estimates of several parameters. As such, we conduct a second, extensive simulation study to evaluate the relative performance of the 2 competitors (PLSc and IGSCA), considering a variety of experimental factors (model specification, sample size, the number of indicators per factor/component, and exogenous factor/component correlation). IGSCA exhibits better performance than PLSc under most conditions. We also present a real data application of IGSCA to the study of genes and their influence on depression. Finally, we discuss the implications and limitations of this approach, and recommendations for future research. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Item-specific overlap between hallucinatory experiences and cognition in the general population: A three-step multivariate analysis of international multi-site data.

Hallucinatory experiences (HEs) can be pronounced in psychosis, but similar experiences also occur in nonclinical populations. Cognitive mechanisms hypothesized to underpin HEs include dysfunctional source monitoring, heightened signal detection, and impaired attentional processes. Using data from an international multisite study on non-clinical participants (N = 419), we described the overlap between two sets of variables - one measuring cognition and the other HEs - at the level of individual items. We used a three-step method to extract and examine item-specific signal, which is typically obscured when summary scores are analyzed using traditional methodologies. The three-step method involved: (1) constraining variance in cognition variables to that which is predictable from HE variables, followed by dimension reduction, (2) determining reliable HE items using split-halves and permutation tests, and (3) selecting cognition items for interpretation using a leave-one-out procedure followed by repetition of Steps 1 and 2. The results showed that the overlap between HEs and cognition variables can be conceptualized as bi-dimensional, with two distinct mechanisms emerging as candidates for separate pathways to the development of HEs: HEs involving perceptual distortions on one hand (including voices), underpinned by a low threshold for signal detection in cognition, and HEs involving sensory overload on the other hand, underpinned by reduced laterality in cognition. We propose that these two dimensions of HEs involving distortions/liberal signal detection, and sensation overload/reduced laterality may map onto psychosis-spectrum and dissociation-spectrum anomalous experiences, respectively.