An in silico procedure for generating protein-mediated chromatin interaction data and comparison of significant interaction calling methods.

The ability to simulate high-throughput data with high fidelity to real experimental data is fundamental for benchmarking methods used to detect true long-range chromatin interactions mediated by a specific protein. Yet, such tools are not currently available. To fill this gap, we develop an in silico experimental procedure, ChIA-Sim, which imitates the experimental procedures that produce real ChIA-PET, Hi-ChIP, or PLAC-seq data. We show the fidelity of ChIA-Sim to real data by using guiding characteristics of several real datasets to generate data using the simulation procedure. We also used ChIA-Sim data to demonstrate the use of our in silico procedure in benchmarking methods for significant interactions analysis by evaluating four methods for significant interaction calling (SIC). In particular, we assessed each method's performance in terms of correct identification of long-range interactions. We further analyzed four experimental datasets from publicly available databases and shew that the trend of the results are consistent with those seen in data generated from ChIA-Sim. This serves as additional evidence that ChIA-Sim closely resembles data produced from the experimental protocols it models after.

Bayesian LASSO for population stratification correction in rare haplotype association studies.

Population stratification (PS) is one major source of confounding in both single nucleotide polymorphism (SNP) and haplotype association studies. To address PS, principal component regression (PCR) and linear mixed model (LMM) are the current standards for SNP associations, which are also commonly borrowed for haplotype studies. However, the underfitting and overfitting problems introduced by PCR and LMM, respectively, have yet to be addressed. Furthermore, there have been only a few theoretical approaches proposed to address PS specifically for haplotypes. In this paper, we propose a new method under the Bayesian LASSO framework, QBLstrat, to account for PS in identifying rare and common haplotypes associated with a continuous trait of interest. QBLstrat utilizes a large number of principal components (PCs) with appropriate priors to sufficiently correct for PS, while shrinking the estimates of unassociated haplotypes and PCs. We compare the performance of QBLstrat with the Bayesian counterparts of PCR and LMM and a current method, haplo.stats. Extensive simulation studies and real data analyses show that QBLstrat is superior in controlling false positives while maintaining competitive power for identifying true positives under PS.

Methane prediction equations including genera of rumen bacteria as predictor variables improve prediction accuracy.

Methane (CH4) emissions from ruminants are of a significant environmental concern, necessitating accurate prediction for emission inventories. Existing models rely solely on dietary and host animal-related data, ignoring the predicting power of rumen microbiota, the source of CH4. To address this limitation, we developed novel CH4 prediction models incorporating rumen microbes as predictors, alongside animal- and feed-related predictors using four statistical/machine learning (ML) methods. These include random forest combined with boosting (RF-B), least absolute shrinkage and selection operator (LASSO), generalized linear mixed model with LASSO (glmmLasso), and smoothly clipped absolute deviation (SCAD) implemented on linear mixed models. With a sheep dataset (218 observations) of both animal data and rumen microbiota data (relative sequence abundance of 330 genera of rumen bacteria, archaea, protozoa, and fungi), we developed linear mixed models to predict CH4 production (g CH4/animal·d, ANIM-B models) and CH4 yield (g CH4/kg of dry matter intake, DMI-B models). We also developed models solely based on animal-related data. Prediction performance was evaluated 200 times with random data splits, while fitting performance was assessed without data splitting. The inclusion of microbial predictors improved the models, as indicated by decreased root mean square prediction error (RMSPE) and mean absolute error (MAE), and increased Lin's concordance correlation coefficient (CCC). Both glmmLasso and SCAD reduced the Akaike information criterion (AIC) and Bayesian information criterion (BIC) for both the ANIM-B and the DMI-B models, while the other two ML methods had mixed outcomes. By balancing prediction performance and fitting performance, we obtained one ANIM-B model (containing 10 genera of bacteria and 3 animal data) fitted using glmmLasso and one DMI-B model (5 genera of bacteria and 1 animal datum) fitted using SCAD. This study highlights the importance of incorporating rumen microbiota data in CH4 prediction models to enhance accuracy and robustness. Additionally, ML methods facilitate the selection of microbial predictors from high-dimensional metataxonomic data of the rumen microbiota without overfitting. Moreover, the identified microbial predictors can serve as biomarkers of CH4 emissions from sheep, providing valuable insights for future research and mitigation strategies.

Population stratification correction using Bayesian shrinkage priors for genetic association studies.

INTRODUCTION: Population stratification (PS) is a major source of confounding in population-based genetic association studies of quantitative traits. Principal component regression (PCR) and linear mixed model (LMM) are two commonly used approaches to account for PS in association studies. Previous studies have shown that LMM can be interpreted as including all principal components (PCs) as random-effect covariates. However, including all PCs in LMM may dilute the influence of relevant PCs in some scenarios, while including only a few preselected PCs in PCR may fail to fully capture the genetic diversity. MATERIALS AND METHODS: To address these shortcomings, we introduce Bayestrat-a method to detect associated variants with PS correction under the Bayesian LASSO framework. To adjust for PS, Bayestrat accommodates a large number of PCs and utilizes appropriate shrinkage priors to shrink the effects of nonassociated PCs. RESULTS: Simulation results show that Bayestrat consistently controls type I error rates and achieves higher power compared to its non-shrinkage counterparts, especially when the number of PCs included in the model is large. As a demonstration of the utility of Bayestrat, we apply it to the Multi-Ethnic Study of Atherosclerosis (MESA). Variants and genes associated with serum triglyceride or HDL cholesterol are identified in our analyses. DISCUSSION: The automatic and self-selection features of Bayestrat make it particularly suited in situations with complex underlying PS scenarios, where it is unknown a priori which PCs are potential confounders, yet the number that needs to be considered could be large in order to fully account for PS.

Human alveolar macrophage response to Mycobacterium tuberculosis: immune characteristics underlying large inter-individual variability.

Background: Mycobacterium tuberculosis (M.tb), the causative bacterium of tuberculosis (TB), establishes residence and grows in human alveolar macrophages (AMs). Inter-individual variation in M.tb-human AM interactions can indicate TB risk and the efficacy of therapies and vaccines; however, we currently lack an understanding of the gene and protein expression programs that dictate this variation in the lungs. Results: Herein, we systematically analyze interactions of a virulent M.tb strain H37Rv with freshly isolated human AMs from 28 healthy adult donors, measuring host RNA expression and secreted candidate proteins associated with TB pathogenesis over 72h. A large set of genes possessing highly variable inter-individual expression levels are differentially expressed in response to M.tb infection. Eigengene modules link M.tb growth rate with host transcriptional and protein profiles at 24 and 72h. Systems analysis of differential RNA and protein expression identifies a robust network with IL1B, STAT1, and IDO1 as hub genes associated with M.tb growth. RNA time profiles document stimulation towards an M1-type macrophage gene expression followed by emergence of an M2-type profile. Finally, we replicate these results in a cohort from a TB-endemic region, finding a substantial portion of significant differentially expressed genes overlapping between studies. Conclusions: We observe large inter-individual differences in bacterial uptake and growth, with tenfold variation in M.tb load by 72h.The fine-scale resolution of this work enables the identification of genes and gene networks associated with early M.tb growth dynamics in defined donor clusters, an important step in developing potential biological indicators of individual susceptibility to M.tb infection and response to therapies.

Do Patients With Low-Risk Pulmonary Arterial Hypertension Really Benefit From Upfront Combination Therapy?: Insight From the AMBITION Trial.

BACKGROUND: Based on results of the Ambrisentan and Tadalafil in Patients with Pulmonary Arterial Hypertension (AMBITION) trial, upfront combination therapy is recommended for treatment-naive patients with low-risk pulmonary arterial hypertension (PAH). However, conflicting data exist whether adopting this treatment strategy in this risk group is beneficial or well tolerated. RESEARCH QUESTION: Do patients with low-risk PAH really benefit from upfront combination therapy? STUDY DESIGN AND METHODS: Using the data from the original AMBITION trial, patients with PAH were classified as low, intermediate, or high risk using the Registry to Evaluate Early and Long-term PAH Disease Management 2.0 (REVEAL 2.0) score and the Pulmonary Hypertension Outcomes and Risk Assessment (PHORA) tool. The primary end point was time to clinical worsening (including death, hospitalization for PAH worsening, and disease progression) censored at 1- and 3-year post-enrollment. Side effects that led to withdrawal of treatment were also considered. RESULTS: Patients with low-risk PAH categorized by REVEAL 2.0 and PHORA did not see a statistically significant benefit of upfront combination therapy vs monotherapy for time to clinical worsening at 1 and 3 years' post-enrollment using Cox proportional analysis (3-year hazard ratio of 0.40 [95% CI, 0.15-1.06; P = .07] and 0.55 [95% CI, 0.26-1.18; P = .12] for REVEAL 2.0 and PHORA, respectively) or considering time to clinical worsening or side effects (3-year hazard ratio of 0.75 [95% CI, 0.39-1.47; P = .4] and 0.87 [95% CI, 0.49-1.54; P = .63] for REVEAL 2.0 and PHORA). Patients with low-risk PAH on upfront combination therapy experienced a higher but not significant incidence of side effects using REVEAL 2.0 and PHORA. In contrast, patients at intermediate or high risk saw a statistically significant benefit of upfront combination therapy considering each of the end points regardless of side effects. INTERPRETATION: This analysis suggests that perhaps some patients with low-risk PAH should be further stratified using other modalities prior to committing to upfront combination therapy, especially when the occurrence of side effects is considered. Further prospective data are needed to validate this hypothesis prior to changes in current guideline directed therapy are contemplated.

An adaptive and robust method for multi-trait analysis of genome-wide association studies using summary statistics.

Genome-wide association studies (GWAS) have identified thousands of genetic variants associated with human traits or diseases in the past decade. Nevertheless, much of the heritability of many traits is still unaccounted for. Commonly used single-trait analysis methods are conservative, while multi-trait methods improve statistical power by integrating association evidence across multiple traits. In contrast to individual-level data, GWAS summary statistics are usually publicly available, and thus methods using only summary statistics have greater usage. Although many methods have been developed for joint analysis of multiple traits using summary statistics, there are many issues, including inconsistent performance, computational inefficiency, and numerical problems when considering lots of traits. To address these challenges, we propose a multi-trait adaptive Fisher method for summary statistics (MTAFS), a computationally efficient method with robust power performance. We applied MTAFS to two sets of brain imaging derived phenotypes (IDPs) from the UK Biobank, including a set of 58 Volumetric IDPs and a set of 212 Area IDPs. Through annotation analysis, the underlying genes of the SNPs identified by MTAFS were found to exhibit higher expression and are significantly enriched in brain-related tissues. Together with results from a simulation study, MTAFS shows its advantage over existing multi-trait methods, with robust performance across a range of underlying settings. It controls type 1 error well and can efficiently handle a large number of traits.

Sparse estimation in semiparametric finite mixture of varying coefficient regression models.

Finite mixture of regressions (FMR) are commonly used to model heterogeneous effects of covariates on a response variable in settings where there are unknown underlying subpopulations. FMRs, however, cannot accommodate situations where covariates' effects also vary according to an "index" variable-known as finite mixture of varying coefficient regression (FM-VCR). Although complex, this situation occurs in real data applications: the osteocalcin (OCN) data analyzed in this manuscript presents a heterogeneous relationship where the effect of a genetic variant on OCN in each hidden subpopulation varies over time. Oftentimes, the number of covariates with varying coefficients also presents a challenge: in the OCN study, genetic variants on the same chromosome are considered jointly. The relative proportions of hidden subpopulations may also change over time. Nevertheless, existing methods cannot provide suitable solutions for accommodating all these features in real data applications. To fill this gap, we develop statistical methodologies based on regularized local-kernel likelihood for simultaneous parameter estimation and variable selection in sparse FM-VCR models. We study large-sample properties of the proposed methods. We then carry out a simulation study to evaluate the performance of various penalties adopted for our regularized approach and ascertain the ability of a BIC-type criterion for estimating the number of subpopulations. Finally, we applied the FM-VCR model to analyze the OCN data and identified several covariates, including genetic variants, that have age-dependent effects on OCN.

An Adaptive and Robust Test for Microbial Community Analysis.

In microbiome studies, researchers measure the abundance of each operational taxon unit (OTU) and are often interested in testing the association between the microbiota and the clinical outcome while conditional on certain covariates. Two types of approaches exists for this testing purpose: the OTU-level tests that assess the association between each OTU and the outcome, and the community-level tests that examine the microbial community all together. It is of considerable interest to develop methods that enjoy both the flexibility of OTU-level tests and the biological relevance of community-level tests. We proposed MiAF, a method that adaptively combines p-values from the OTU-level tests to construct a community-level test. By borrowing the flexibility of OTU-level tests, the proposed method has great potential to generate a series of community-level tests that suit a range of different microbiome profiles, while achieving the desirable high statistical power of community-level testing methods. Using simulation study and real data applications in a smoker throat microbiome study and a HIV patient stool microbiome study, we demonstrated that MiAF has comparable or better power than methods that are specifically designed for community-level tests. The proposed method also provides a natural heuristic taxa selection.

HiCImpute: A Bayesian hierarchical model for identifying structural zeros and enhancing single cell Hi-C data.

Single cell Hi-C techniques enable one to study cell to cell variability in chromatin interactions. However, single cell Hi-C (scHi-C) data suffer severely from sparsity, that is, the existence of excess zeros due to insufficient sequencing depth. Complicating the matter further is the fact that not all zeros are created equal: some are due to loci truly not interacting because of the underlying biological mechanism (structural zeros); others are indeed due to insufficient sequencing depth (sampling zeros or dropouts), especially for loci that interact infrequently. Differentiating between structural zeros and dropouts is important since correct inference would improve downstream analyses such as clustering and discovery of subtypes. Nevertheless, distinguishing between these two types of zeros has received little attention in the single cell Hi-C literature, where the issue of sparsity has been addressed mainly as a data quality improvement problem. To fill this gap, in this paper, we propose HiCImpute, a Bayesian hierarchical model that goes beyond data quality improvement by also identifying observed zeros that are in fact structural zeros. HiCImpute takes spatial dependencies of scHi-C 2D data structure into account while also borrowing information from similar single cells and bulk data, when such are available. Through an extensive set of analyses of synthetic and real data, we demonstrate the ability of HiCImpute for identifying structural zeros with high sensitivity, and for accurate imputation of dropout values. Downstream analyses using data improved from HiCImpute yielded much more accurate clustering of cell types compared to using observed data or data improved by several comparison methods. Most significantly, HiCImpute-improved data have led to the identification of subtypes within each of the excitatory neuronal cells of L4 and L5 in the prefrontal cortex.

BCurve: Bayesian Curve Credible Bands Approach for the Detection of Differentially Methylated Regions.

High-throughput assays have been developed to measure DNA methylation, among which bisulfite-based sequencing (BS-seq) and microarray technologies are the most popular for genome-wide profiling. A major goal in DNA methylation analysis is the detection of differentially methylated genomic regions under two different conditions. To accomplish this, many state-of-the-art methods have been proposed in the past few years; only a handful of these methods are capable of analyzing both types of data (BS-seq and microarray), though. On the other hand, covariates, such as sex and age, are known to be potentially influential on DNA methylation; and thus, it would be important to adjust for their effects on differential methylation analysis. In this chapter, we describe a Bayesian curve credible bands approach and the accompanying software, BCurve, for detecting differentially methylated regions for data generated from either microarray or BS-Seq. The unified theme underlying the analysis of these two different types of data is the model that accounts for correlation between DNA methylation in nearby sites, covariates, and between-sample variability. The BCurve R software package also provides tools for simulating both microarray and BS-seq data, which can be useful for facilitating comparisons of methods given the known "gold standard" in the simulated data. We provide detailed description of the main functions in BCurve and demonstrate the utility of the package for analyzing data from both platforms using simulated data from the functions provided in the package. Analyses of two real datasets, one from BS-seq and one from microarray, are also furnished to further illustrate the capability of BCurve.

Detecting X-linked common and rare variant effects in family-based sequencing studies.

The breakthroughs in next generation sequencing have allowed us to access data consisting of both common and rare variants, and in particular to investigate the impact of rare genetic variation on complex diseases. Although rare genetic variants are thought to be important components in explaining genetic mechanisms of many diseases, discovering these variants remains challenging, and most studies are restricted to population-based designs. Further, despite the shift in the field of genome-wide association studies (GWAS) towards studying rare variants due to the "missing heritability" phenomenon, little is known about rare X-linked variants associated with complex diseases. For instance, there is evidence that X-linked genes are highly involved in brain development and cognition when compared with autosomal genes; however, like most GWAS for other complex traits, previous GWAS for mental diseases have provided poor resources to deal with identification of rare variant associations on X-chromosome. In this paper, we address the two issues described above by proposing a method that can be used to test X-linked variants using sequencing data on families. Our method is much more general than existing methods, as it can be applied to detect both common and rare variants, and is applicable to autosomes as well. Our simulation study shows that the method is efficient, and exhibits good operational characteristics. An application to the University of Miami Study on Genetics of Autism and Related Disorders also yielded encouraging results.

Are dropout imputation methods for scRNA-seq effective for scHi-C data?

The prevalence of dropout events is a serious problem for single-cell Hi-C (scHiC) data due to insufficient sequencing depth and data coverage, which brings difficulties in downstream studies such as clustering and structural analysis. Complicating things further is the fact that dropouts are confounded with structural zeros due to underlying properties, leading to observed zeros being a mixture of both types of events. Although a great deal of progress has been made in imputing dropout events for single cell RNA-sequencing (RNA-seq) data, little has been done in identifying structural zeros and imputing dropouts for scHiC data. In this paper, we adapted several methods from the single-cell RNA-seq literature for inference on observed zeros in scHiC data and evaluated their effectiveness. Through an extensive simulation study and real data analysis, we have shown that a couple of the adapted single-cell RNA-seq algorithms can be powerful for correctly identifying structural zeros and accurately imputing dropout values. Downstream analysis using the imputed values showed considerable improvement for clustering cells of the same types together over clustering results before imputation.

Modeling and analysis of Hi-C data by HiSIF identifies characteristic promoter-distal loops.

Current computational methods on Hi-C analysis focused on identifying Mb-size domains often failed to unveil the underlying functional and mechanistic relationship of chromatin structure and gene regulation. We developed a novel computational method HiSIF to identify genome-wide interacting loci. We illustrated HiSIF outperformed other tools for identifying chromatin loops. We applied it to Hi-C data in breast cancer cells and identified 21 genes with gained loops showing worse relapse-free survival in endocrine-treated patients, suggesting the genes with enhanced loops can be used for prognostic signatures for measuring the outcome of the endocrine treatment. HiSIF is available at https://github.com/yufanzhouonline/HiSIF .

Detecting rare haplotypes associated with complex diseases using both population and family data: Combined logistic Bayesian Lasso.

Haplotype-based association methods have been developed to understand the genetic architecture of complex diseases. Compared to single-variant-based methods, haplotype methods are thought to be more biologically relevant, since there are typically multiple non-independent genetic variants involved in complex diseases, and the use of haplotypes implicitly accounts for non-independence caused by linkage disequilibrium. In recent years, with the focus moving from common to rare variants, haplotype-based methods have also evolved accordingly to uncover the roles of rare haplotypes. One particular approach is regularization-based, with the use of Bayesian least absolute shrinkage and selection operator (Lasso) as an example. This type of methods has been developed for either case-control population data (the logistic Bayesian Lasso (LBL)) or family data (family-triad-based logistic Bayesian Lasso (famLBL)). In some situations, both family data and case-control data are available; therefore, it would be a waste of resources if only one of them could be analyzed. To make full usage of available data to increase power, we propose a unified approach that can combine both case-control and family data (combined logistic Bayesian Lasso (cLBL)). Through simulations, we characterized the performance of cLBL and showed the advantage of cLBL over existing methods. We further applied cLBL to the Framingham Heart Study data to demonstrate its utility in real data applications.

Characterization of histone modification patterns and prediction of novel promoters using functional principal component analysis.

Characterization of distinct histone methylation and acetylation binding patterns in promoters and prediction of novel regulatory regions remains an important area of genomic research, as it is hypothesized that distinct chromatin signatures may specify unique genomic functions. However, methods that have been proposed in the literature are either descriptive in nature or are fully parametric and hence more restrictive in pattern discovery. In this article, we propose a two-step non-parametric statistical inference procedure to characterize unique histone modification patterns and apply it to analyzing the binding patterns of four histone marks, H3K4me2, H3K4me3, H3K9ac, and H4K20me1, in human B-lymphoblastoid cells. In the first step, we used a functional principal component analysis method to represent the concatenated binding patterns of these four histone marks around the transcription start sites as smooth curves. In the second step, we clustered these curves to reveal several unique classes of binding patterns. These uncovered patterns were used in turn to scan the whole-genome to predict novel and alternative promoters. Our analyses show that there are three distinct promoter binding patterns of active genes. Further, 19654 regions not within known gene promoters were found to overlap with human ESTs, CpG islands, or common SNPs, indicative of their potential role in gene regulation, including being potential novel promoter regions.

The Association of ARMC5 with the Renin-Angiotensin-Aldosterone System, Blood Pressure, and Glycemia in African Americans.

CONTEXT: Armadillo repeat containing 5 (ARMC5) on chromosome 16 is an adrenal gland tumor suppressor gene associated with primary aldosteronism, especially among African Americans (AAs). We examined the association of ARMC5 variants with aldosterone, plasma renin activity (PRA), blood pressure, glucose, and glycosylated hemoglobin A1c (HbA1c) in community-dwelling AAs. METHODS: The Jackson Heart Study is a prospective cardiovascular cohort study in AAs with baseline data collection from 2000 to 2004. Kernel machine method was used to perform a single joint test to analyze for an overall association between the phenotypes of interest (aldosterone, PRA, systolic and diastolic blood pressure [SBP, DBP], glucose, and HbA1c) and the ARMC5 single nucleotide variants (SNVs) adjusted for age, sex, BMI, and medications; followed by Baysian Lasso methodology to identify sets of SNVs in terms of associated haplotypes with specific phenotypes. RESULTS: Among 3223 participants (62% female; mean age 55.6 (SD ± 12.8) years), the average SBP and DBP were 127 and 76 mmHg, respectively. The average fasting plasma glucose and HbA1c were 101 mg/dL and 6.0%, respectively. ARMC5 variants were associated with all 6 phenotypes. Haplotype TCGCC (ch16:31476015-31476093) was negatively associated, whereas haplotype CCCCTTGCG (ch16:31477195-31477460) was positively associated with SBP, DBP, and glucose. Haplotypes GGACG (ch16:31477790-31478013) and ACGCG (ch16:31477834-31478113) were negatively associated with aldosterone and positively associated with HbA1c and glucose, respectively. Haplotype GCGCGAGC (ch16:31471193-ch16:31473597(rs114871627) was positively associated with PRA and negatively associated with HbA1c. CONCLUSIONS: ARMC5 variants are associated with aldosterone, PRA, blood pressure, fasting glucose, and HbA1c in community-dwelling AAs, suggesting that germline mutations in ARMC5 may underlie cardiometabolic disease in AAs.

Author Correction: Temporal dynamic reorganization of 3D chromatin architecture in hormone-induced breast cancer and endocrine resistance.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Incorporating information from markers in LD with test locus for detecting imprinting and maternal effects.

Numerous statistical methods have been developed to explore genomic imprinting and maternal effects by identifying parent-of-origin patterns in complex human diseases. However, because most of these methods only use available locus-specific genotype data, it is sometimes impossible for them to infer the distribution of parental origin of a variant allele, especially when some genotypes are missing. In this article, we propose a two-step approach, LIMEhap, to improve upon a recent partial likelihood inference method. In the first step, the distribution of the missing genotypes is inferred through the construction of haplotypes by using information from nearby loci. In the second step, a partial likelihood method is applied to the inferred data. To substantiate the validity of the proposed procedures, we simulated data in a genomic region of gene GPX1. The results show that, by borrowing genetic information from nearby loci, the power of the proposed method can be close to that with complete genotype data at the locus of interest. Since the inference on the genotype distribution is made under the assumption of Hardy-Weinberg Equilibrium (HWE), we further studied the robustness of LIMEhap to violation of HWE. Finally, we demonstrate the utility of LIMEhap by applying it to an autism dataset.