Model-based clustering of DNA methylation array data: a recursive-partitioning algorithm for high-dimensional data arising as a mixture of beta distributions

Computational deconvolution of DNA methylation data from mixed DNA samples

In this review, we provide a comprehensive overview of the different computational tools that have been published for the deconvolution of bulk DNA methylation (DNAm) data. Here, deconvolution refers to the estimation of cell-type proportions that constitute a mixed sample. The paper reviews and compares 25 deconvolution methods (supervised, unsupervised or hybrid) developed between 2012 and 2023 and compares the strengths and limitations of each approach. Moreover, in this study, we describe the impact of the platform used for the generation of methylation data (including microarrays and sequencing), the applied data pre-processing steps and the used reference dataset on the deconvolution performance. Next to reference-based methods, we also examine methods that require only partial reference datasets or require no reference set at all. In this review, we provide guidelines for the use of specific methods dependent on the DNA methylation data type and data availability.

Discovery of novel DNA methylation biomarker panels for the diagnosis and differentiation between common adenocarcinomas and their liver metastases

Differentiation between adenocarcinomas is sometimes challenging. The promising avenue for discovering new biomarkers lies in bioinformatics using DNA methylation analysis. Utilizing a 2853-sample identification dataset and a 782-sample independent verification dataset, we have identified diagnostic DNA methylation biomarkers that are hypermethylated in cancer and differentiate between breast invasive carcinoma, cholangiocarcinoma, colorectal cancer, hepatocellular carcinoma, lung adenocarcinoma, pancreatic adenocarcinoma and stomach adenocarcinoma. The best panels for cancer type exhibit sensitivity of 77.8-95.9%, a specificity of 92.7-97.5% for tumors, a specificity of 91.5-97.7% for tumors and normal tissues and a diagnostic accuracy of 85.3-96.4%. We have shown that the results can be extended from the primary cancers to their liver metastases, as the best panels diagnose and differentiate between pancreatic adenocarcinoma liver metastases and breast invasive carcinoma liver metastases with a sensitivity and specificity of 83.3-100% and a diagnostic accuracy of 86.8-91.9%. Moreover, the panels could detect hypermethylation of selected regions in the cell-free DNA of patients with liver metastases. At the same time, these were unmethylated in the cell-free DNA of healthy donors, confirming their applicability for liquid biopsies.

The Origin of Highly Elevated Cell-Free DNA in Healthy Individuals and Patients with Pancreatic, Colorectal, Lung, or Ovarian Cancer

Cell-free DNA (cfDNA) concentrations from patients with cancer are often elevated compared with those of healthy controls, but the sources of this extra cfDNA have never been determined. To address this issue, we assessed cfDNA methylation patterns in 178 patients with cancers of the colon, pancreas, lung, or ovary and 64 patients without cancer. Eighty-three of these individuals had cfDNA concentrations much greater than those generally observed in healthy subjects. The major contributor of cfDNA in all samples was leukocytes, accounting for ∼76% of cfDNA, with neutrophils predominating. This was true regardless of whether the samples were derived from patients with cancer or the total plasma cfDNA concentration. High levels of cfDNA observed in patients with cancer did not come from either neoplastic cells or surrounding normal epithelial cells from the tumor's tissue of origin. These data suggest that cancers may have a systemic effect on cell turnover or DNA clearance.

SIGNIFICANCE

The origin of excess cfDNA in patients with cancer is unknown. Using cfDNA methylation patterns, we determined that neither the tumor nor the surrounding normal tissue contributes this excess cfDNA-rather it comes from leukocytes. This finding suggests that cancers have a systemic impact on cell turnover or DNA clearance. See related commentary by Thierry and Pisareva, p. 2122. This article is featured in Selected Articles from This Issue, p. 2109.

In Silico Gene Prioritization Highlights the Significance of Bone Morphogenetic Protein 4 (BMP4) Promoter Methylation across All Methylation Clusters in Colorectal Cancer

The cytosine-phosphate-guanine (CpG) island methylator phenotype (CIMP) represents one of the pathways involved in the development of colorectal cancer, characterized by genome-wide hypermethylation. To identify samples exhibiting hypermethylation, we used unsupervised hierarchical clustering on genome-wide methylation data. This clustering analysis revealed the presence of four distinct subtypes within the tumor samples, namely, CIMP-H, CIMP-L, cluster 3, and cluster 4. These subtypes demonstrated varying levels of methylation, categorized as high, intermediate, and very low. To gain further insights, we mapped significant probes from all clusters to Ensembl Regulatory build 89, with a specific focus on those located within promoter regions or bound regions. By intersecting the methylated promoter and bound regions across all methylation subtypes, we identified a total of 253 genes exhibiting aberrant methylation patterns in the promoter regions across all four subtypes of colorectal cancer. Among these genes, our comprehensive genome-wide analysis highlights bone morphogenic protein 4 (BMP4) as the most prominent candidate. This significant finding was derived through the utilization of various bioinformatics tools, emphasizing the potential role of BMP4 in colorectal cancer development and progression.

RABC: Rheumatoid Arthritis Bioinformatics Center

Advances in sequencing technologies have led to the rapid growth of multi-omics data on rheumatoid arthritis (RA). However, a comprehensive database that systematically collects and classifies the scattered data is still lacking. Here, we developed the Rheumatoid Arthritis Bioinformatics Center (RABC, http://www.onethird-lab.com/RABC/), the first multi-omics data resource platform (data hub) for RA. There are four categories of data in RABC: (i) 175 multi-omics sample sets covering transcriptome, epigenome, genome, and proteome; (ii) 175 209 differentially expressed genes (DEGs), 105 differentially expressed microRNAs (DEMs), 18 464 differentially DNA methylated (DNAm) genes, 1 764 KEGG pathways, 30 488 GO terms, 74 334 SNPs, 242 779 eQTLs, 105 m6A-SNPs and 18 491 669 meta-mQTLs; (iii) prior knowledge on seven types of RA molecular markers from nine public and credible databases; (iv) 127 073 literature information from PubMed (from 1972 to March 2022). RABC provides a user-friendly interface for browsing, searching and downloading these data. In addition, a visualization module also supports users to generate graphs of analysis results by inputting personalized parameters. We believe that RABC will become a valuable resource and make a significant contribution to the study of RA.

An Intelligent Classification System for Cancer Detection Based on DNA Methylation Using ML and Semantic Knowledge in Healthcare

To consistently assess a patient's internal and external wellness and diagnose chronic conditions like cancer, Alzheimer's disease, and cardiovascular disease, wearable sensing devices are being used. Wearable technologies and networking websites have become incredibly common in the medical sector in recent times. The condition of a patient's health can be influenced by a number of factors, including psychological response, emotional stability, and anxiety levels, which can be evaluated using social network analysis based on graph theory-based techniques and these ideas, known as "social network analysis" (SNA) are used to study relationship phenomena. Therefore, numerous uses for SNA in health research are possible, ranging from social science to exact science. For example, it can be used to research cooperative networks of healthcare providers and hazard-prone behaviors, infectious disease transmission, and the spread of initiatives for health promotion and prevention. Recently, a number of machine learning-based healthcare solutions have been proposed to track chronic illnesses utilizing data from social networks and wearable monitoring devices. In our suggested approach, we are using an intelligent system with the assistance of wearable sensors for the classification of cancer based on DNA methylation, an important epigenetic process in the human genome that controls gene expression and has been connected to a number of health issues. A mixed-sampling imbalanced data ensemble classification technique is created with the help of biomedical sensors to address the problem of class imbalance and high dimensionality in the Cancer Genome Atlas (TCGA) massive data. This technique is based on the Intelligent Synthetic Minority Oversampling (SMOTE) algorithm. The false-negative rate significantly rises as a result of this, to give a larger data set, a new minority class sample will be first obtained. The noise created during the sample expansion process is actually any data that has been acquired, preserved, or altered in a way that prevents the system that initially conceived it from accessing or utilizing it. Noisy data boosts the amount of space needed excessively and can also drastically influence the findings of any data collection investigation and therefore can also affect the sample sets of one or the other class, resulting in the class imbalance which acts as a common problem in ML datasets. The Tomek Link method is then used to eliminate this noise, producing a reasonably balanced data set. Each layer selects two random forest structures using the cascading forest structure of the deep forest (GC-Forest) algorithm to increase the generalization ability of the model and create the final classification model. Experiments using DNA methylation data collected by employing biosensors from six tumor patients reveal that the mixed-sampling unbalanced data ensemble classification technique may increase the sensitivity to the minority class while maintaining the majority class's classification accuracy.

Deconvolution of tumor composition using partially available DNA methylation data

Background

Deciphering proportions of constitutional cell types in tumor tissues is a crucial step for the analysis of tumor heterogeneity and the prediction of response to immunotherapy. In the process of measuring cell population proportions, traditional experimental methods have been greatly hampered by the cost and extensive dropout events. At present, the public availability of large amounts of DNA methylation data makes it possible to use computational methods to predict proportions.

Results

In this paper, we proposed PRMeth, a method to deconvolve tumor mixtures using partially available DNA methylation data. By adopting an iteratively optimized non-negative matrix factorization framework, PRMeth took DNA methylation profiles of a portion of the cell types in the tissue mixtures (including blood and solid tumors) as input to estimate the proportions of all cell types as well as the methylation profiles of unknown cell types simultaneously. We compared PRMeth with five different methods through three benchmark datasets and the results show that PRMeth could infer the proportions of all cell types and recover the methylation profiles of unknown cell types effectively. Then, applying PRMeth to four types of tumors from The Cancer Genome Atlas (TCGA) database, we found that the immune cell proportions estimated by PRMeth were largely consistent with previous studies and met biological significance.

Conclusions

Our method can circumvent the difficulty of obtaining complete DNA methylation reference data and obtain satisfactory deconvolution accuracy, which will be conducive to exploring the new directions of cancer immunotherapy. PRMeth is implemented in R and is freely available from GitHub (https://github.com/hedingqin/PRMeth).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-022-04893-7.

DNA 5-hydroxymethylcytosine in pediatric central nervous system tumors may impact tumor classification and is a positive prognostic marker

Background

Nucleotide-specific 5-hydroxymethylcytosine (5hmC) remains understudied in pediatric central nervous system (CNS) tumors. 5hmC is abundant in the brain, and alterations to 5hmC in adult CNS tumors have been reported. However, traditional approaches to measure DNA methylation do not distinguish between 5-methylcytosine (5mC) and its oxidized counterpart 5hmC, including those used to build CNS tumor DNA methylation classification systems. We measured 5hmC and 5mC epigenome-wide at nucleotide resolution in glioma, ependymoma, and embryonal tumors from children, as well as control pediatric brain tissues using tandem bisulfite and oxidative bisulfite treatments followed by hybridization to the Illumina Methylation EPIC Array that interrogates over 860,000 CpG loci.

Results

Linear mixed effects models adjusted for age and sex tested the CpG-specific differences in 5hmC between tumor and non-tumor samples, as well as between tumor subtypes. Results from model-based clustering of tumors was used to test the relation of cluster membership with patient survival through multivariable Cox proportional hazards regression. We also assessed the robustness of multiple epigenetic CNS tumor classification methods to 5mC-specific data in both pediatric and adult CNS tumors. Compared to non-tumor samples, tumors were hypohydroxymethylated across the epigenome and tumor 5hmC localized to regulatory elements crucial to cell identity, including transcription factor binding sites and super-enhancers. Differentially hydroxymethylated loci among tumor subtypes tended to be hypermethylated and disproportionally found in CTCF binding sites and genes related to posttranscriptional RNA regulation, such as DICER1. Model-based clustering results indicated that patients with low 5hmC patterns have poorer overall survival and increased risk of recurrence. Our results suggest 5mC-specific data from OxBS-treated samples impacts methylation-based tumor classification systems giving new opportunities for further refinement of classifiers for both pediatric and adult tumors.

Conclusions

We identified that 5hmC localizes to super-enhancers, and genes commonly implicated in pediatric CNS tumors were differentially hypohydroxymethylated. We demonstrated that distinguishing methylation and hydroxymethylation is critical in identifying tumor-related epigenetic changes. These results have implications for patient prognostication, considerations of epigenetic therapy in CNS tumors, and for emerging molecular neuropathology classification approaches.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-021-01156-9.

Genomic and phenotypic insights from an atlas of genetic effects on DNA methylation

Characterizing genetic influences on DNA methylation (DNAm) provides an opportunity to understand mechanisms underpinning gene regulation and disease. In the present study, we describe results of DNAm quantitative trait locus (mQTL) analyses on 32,851 participants, identifying genetic variants associated with DNAm at 420,509 DNAm sites in blood. We present a database of >270,000 independent mQTLs, of which 8.5% comprise long-range (trans) associations. Identified mQTL associations explain 15-17% of the additive genetic variance of DNAm. We show that the genetic architecture of DNAm levels is highly polygenic. Using shared genetic control between distal DNAm sites, we constructed networks, identifying 405 discrete genomic communities enriched for genomic annotations and complex traits. Shared genetic variants are associated with both DNAm levels and complex diseases, but only in a minority of cases do these associations reflect causal relationships from DNAm to trait or vice versa, indicating a more complex genotype-phenotype map than previously anticipated.

Artificial Intelligence in Epigenetic Studies: Shedding Light on Rare Diseases

More than 7,000 rare diseases (RDs) exist worldwide, affecting approximately 350 million people, out of which only 5% have treatment. The development of novel genome sequencing techniques has accelerated the discovery and diagnosis in RDs. However, most patients remain undiagnosed. Epigenetics has emerged as a promise for diagnosis and therapies in common disorders (e.g., cancer) with several epimarkers and epidrugs already approved and used in clinical practice. Hence, it may also become an opportunity to uncover new disease mechanisms and therapeutic targets in RDs. In this “big data” age, the amount of information generated, collected, and managed in (bio)medicine is increasing, leading to the need for its rapid and efficient collection, analysis, and characterization. Artificial intelligence (AI), particularly deep learning, is already being successfully applied to analyze genomic information in basic research, diagnosis, and drug discovery and is gaining momentum in the epigenetic field. The application of deep learning to epigenomic studies in RDs could significantly boost discovery and therapy development. This review aims to collect and summarize the application of AI tools in the epigenomic field of RDs. The lower number of studies found, specific for RDs, indicate that this is a field open to expansion, following the results obtained for other more common disorders.

Penalized logistic regression based on L1/2 penalty for high-dimensional DNA methylation data

BACKGROUND:

DNA methylation is a molecular modification of DNA that is vital and occurs in gene expression. In cancer tissues, the 5’–C–phosphate–G–3’(CpG) rich regions are abnormally hypermethylated or hypomethylated. Therefore, it is useful to find out the diseased CpG sites by employing specific methods. CpG sites are highly correlated with each other within the same gene or the same CpG island.

OBJECTIVE:

Based on this group effect, we proposed an efficient and accurate method for selecting pathogenic CpG sites.

METHODS:

Our method aimed to combine a L1/2 regularized solver and a central node fully connected network to penalize group constrained logistic regression model. Consequently, both sparsity and group effect were brought in with respect to the correlated regression coefficients.

RESULTS:

Extensive simulation studies were used to compare our proposed approach with existing mainstream regularization in respect of classification accuracy and stability. The simulation results show that a greater predictive accuracy was attained in comparison to previous methods. Furthermore, our method was applied to over 20000 CpG sites and verified using the ovarian cancer data generated from Illumina Infinium HumanMethylation 27K Beadchip. In the result of the real dataset, not only the indicators of predictive accuracy are higher than the previous methods, but also more CpG sites containing genes are confirmed pathogenic. Additionally, the total number of CpG sites chosen is less than other methods and the results show higher accuracy rates in comparison to other methods in simulation and DNA methylation data.

CONCLUSION:

The proposed method offers an advanced tool to researchers in DNA methylation and can be a powerful tool for recognizing pathogenic CpG sites.

Current Status and Future Targeted Therapy in Adrenocortical Cancer

Adrenocortical carcinoma (ACC) is a rare malignancy with a poor prognosis. The current treatment standards include complete surgical resection for localized resectable disease and systemic therapy with mitotane alone or in combination with etoposide, doxorubicin, and cisplatin in patients with advanced ACC. However, the efficacy of systemic therapy in ACC is very limited, with high rates of toxicities. The understanding of altered molecular pathways is critically important to identify effective treatment options that currently do not exist. In this review, we discuss the results of recent advanced in molecular profiling of ACC with the focus on dysregulated pathways from various genomic and epigenetic dysregulation. We discuss the potential translational therapeutic implication of molecular alterations. In addition, we review and summarize the results of recent clinical trials and ongoing trials.

Application of Random Forest and data integration identifies three dysregulated genes and enrichment of Central Carbon Metabolism pathway in Oral Cancer

BACKGROUND

Studies of epigenomic alterations associated with diseases primarily focus on methylation profiles of promoter regions of genes, but not of other genomic regions. In our past work (Das et al. 2019) on patients suffering from gingivo-buccal oral cancer - the most prevalent form of cancer among males in India - we have also focused on promoter methylation changes and resultant impact on transcription profiles. Here, we have investigated alterations in non-promoter (gene-body) methylation profiles and have carried out an integrative analysis of gene-body methylation and transcriptomic data of oral cancer patients.

METHODS

Tumor and adjacent normal tissue samples were collected from 40 patients. Data on methylation in the non-promoter (gene-body) regions of genes and transcriptome profiles were generated and analyzed. Because of high dimensionality and highly correlated nature of these data, we have used Random Forest (RF) and other data-analytical methods.

RESULTS

Integrative analysis of non-promoter methylation and transcriptome data revealed significant methylation-driven alterations in some genes that also significantly impact on their transcription levels. These changes result in enrichment of the Central Carbon Metabolism (CCM) pathway, primarily by dysregulation of (a) NTRK3, which plays a dual role as an oncogene and a tumor suppressor; (b) SLC7A5 (LAT1) which is a transporter dedicated to essential amino acids, and is overexpressed in cancer cells to meet the increased demand for nutrients that include glucose and essential amino acids; and, (c) EGFR which has been earlier implicated in progression, recurrence, and stemness of oral cancer, but we provide evidence of epigenetic impact on overexpression of this gene for the first time.

CONCLUSIONS

In rapidly dividing cancer cells, metabolic reprogramming from normal cells takes place to enable enhanced proliferation. Here, we have identified that among oral cancer patients, genes in the CCM pathway - that plays a fundamental role in metabolic reprogramming - are significantly dysregulated because of perturbation of methylation in non-promoter regions of the genome. This result compliments our previous result that perturbation of promoter methylation results in significant changes in key genes that regulate the feedback process of DNA methylation for the maintenance of normal cell division.

Coupled co-clustering-based unsupervised transfer learning for the integrative analysis of single-cell genomic data

Unsupervised methods, such as clustering methods, are essential to the analysis of single-cell genomic data. The most current clustering methods are designed for one data type only, such as single-cell RNA sequencing (scRNA-seq), single-cell ATAC sequencing (scATAC-seq) or sc-methylation data alone, and a few are developed for the integrative analysis of multiple data types. The integrative analysis of multimodal single-cell genomic data sets leverages the power in multiple data sets and can deepen the biological insight. In this paper, we propose a coupled co-clustering-based unsupervised transfer learning algorithm (coupleCoC) for the integrative analysis of multimodal single-cell data. Our proposed coupleCoC builds upon the information theoretic co-clustering framework. In co-clustering, both the cells and the genomic features are simultaneously clustered. Clustering similar genomic features reduces the noise in single-cell data and facilitates transfer of knowledge across single-cell datasets. We applied coupleCoC for the integrative analysis of scATAC-seq and scRNA-seq data, sc-methylation and scRNA-seq data and scRNA-seq data from mouse and human. We demonstrate that coupleCoC improves the overall clustering performance and matches the cell subpopulations across multimodal single-cell genomic datasets. Our method coupleCoC is also computationally efficient and can scale up to large datasets. Availability: The software and datasets are available at https://github.com/cuhklinlab/coupleCoC.

Randomized Phase IIB Trial of the Lignan Secoisolariciresinol Diglucoside in Premenopausal Women at Increased Risk for Development of Breast Cancer

We conducted a multiinstitutional, placebo-controlled phase IIB trial of the lignan secoisolariciresinol diglucoside (SDG) found in flaxseed. Benign breast tissue was acquired by random periareolar fine needle aspiration (RPFNA) from premenopausal women at increased risk for breast cancer. Those with hyperplasia and ≥2% Ki-67 positive cells were eligible for randomization 2:1 to 50 mg SDG/day (Brevail) versus placebo for 12 months with repeat bio-specimen acquisition. The primary endpoint was difference in change in Ki-67 between randomization groups. A total of 180 women were randomized, with 152 ultimately evaluable for the primary endpoint. Median baseline Ki-67 was 4.1% with no difference between arms. Median Ki-67 change was -1.8% in the SDG arm (P = 0.001) and -1.2% for placebo (P = 0.034); with no significant difference between arms. As menstrual cycle phase affects proliferation, secondary analysis was performed for 117 women who by progesterone levels were in the same phase of the menstrual cycle at baseline and off-study tissue sampling. The significant Ki-67 decrease persisted for SDG (median = -2.2%; P = 0.002) but not placebo (median = -1.0%). qRT-PCR was performed on 77 pairs of tissue specimens. Twenty-two had significant ERα gene expression changes (<0.5 or >2.0) with 7 of 10 increases in placebo and 10 of 12 decreases for SDG (P = 0.028), and a difference between arms (P = 0.017). Adverse event incidence was similar in both groups, with no evidence that 50 mg/day SDG is harmful. Although the proliferation biomarker analysis showed no difference between the treatment group and the placebo, the trial demonstrated use of SDG is tolerable and safe.

MethylNet: an automated and modular deep learning approach for DNA methylation analysis

BACKGROUND

DNA methylation (DNAm) is an epigenetic regulator of gene expression programs that can be altered by environmental exposures, aging, and in pathogenesis. Traditional analyses that associate DNAm alterations with phenotypes suffer from multiple hypothesis testing and multi-collinearity due to the high-dimensional, continuous, interacting and non-linear nature of the data. Deep learning analyses have shown much promise to study disease heterogeneity. DNAm deep learning approaches have not yet been formalized into user-friendly frameworks for execution, training, and interpreting models. Here, we describe MethylNet, a DNAm deep learning method that can construct embeddings, make predictions, generate new data, and uncover unknown heterogeneity with minimal user supervision.

RESULTS

The results of our experiments indicate that MethylNet can study cellular differences, grasp higher order information of cancer sub-types, estimate age and capture factors associated with smoking in concordance with known differences.

CONCLUSION

The ability of MethylNet to capture nonlinear interactions presents an opportunity for further study of unknown disease, cellular heterogeneity and aging processes.

DNA methylation profiling to explore colorectal tumor differences according to menopausal hormone therapy use in women

Aim: Use of menopausal hormone therapy (MHT) has been associated with a reduced risk for colorectal cancer, but mechanisms underlying this relationship are not well understood. In the colon, MHT appears to act through estrogen receptor β (ERβ) which may influence DNA methylation by binding to DNA. Using genome-wide methylation profiling data, we aimed to identify genes that may be differentially methylated according to MHT use. Materials & methods: DNA methylation was measured using Illumina HumanMethylation450k arrays in two independent tumor sample sets of colorectal cancer patients. Differential methylation was determined using R/limma. Results: In the discovery analysis, two CpG sites showed differential DNA methylation according to MHT use, both were not replicated. In stratified analyses, 342 CpG sites were associated with current MHT use only in ERβ-positive tumors. Conclusion: The suggestive findings of differential methylation according to current MHT use in ERβ-positive tumors warrant further investigation.

Effect of Bazedoxifene and Conjugated Estrogen (Duavee) on Breast Cancer Risk Biomarkers in High-Risk Women: A Pilot Study

Interventions that relieve vasomotor symptoms while reducing risk for breast cancer would likely improve uptake of chemoprevention for perimenopausal and postmenopausal women. We conducted a pilot study with 6 months of the tissue selective estrogen complex bazedoxifene (20 mg) and conjugated estrogen (0.45 mg; Duavee) to assess feasibility and effects on risk biomarkers for postmenopausal breast cancer. Risk biomarkers included fully automated mammographic volumetric density (Volpara), benign breast tissue Ki-67 (MIB-1 immunochemistry), and serum levels of progesterone, IGF-1, and IGFBP3, bioavailable estradiol and testosterone. Twenty-eight perimenopausal and postmenopausal women at increased risk for breast cancer were enrolled: 13 in cohort A with baseline Ki-67 < 1% and 15 in cohort B with baseline Ki-67 of 1% to 4%. All completed the study with > 85% drug adherence. Significant changes in biomarkers, uncorrected for multiple comparisons, were a decrease in mammographic fibroglandular volume (P = 0.043); decreases in serum progesterone, bioavailable testosterone, and IGF-1 (P < 0.01), an increase in serum bioavailable estradiol (P < 0.001), and for women from cohort B a reduction in Ki-67 (P = 0.017). An improvement in median hot flash score from 15 at baseline to 0 at 6 months, and menopause-specific quality-of-life total, vasomotor, and sexual domain scores were also observed (P < 0.001). Given the favorable effects on risk biomarkers and patient reported outcomes, a placebo-controlled phase IIB trial is warranted.

DNA Methylation Changes in Regional Lung Macrophages Are Associated with Metabolic Differences

A number of pulmonary diseases occur with upper lobe predominance, including cystic fibrosis and smoking-related chronic obstructive pulmonary disease. In the healthy lung, several physiologic and metabolic factors exhibit disparity when comparing the upper lobe of the lung to lower lobe, including differences in oxygenation, ventilation, lymphatic flow, pH, and blood flow. In this study, we asked whether these regional differences in the lung are associated with DNA methylation changes in lung macrophages that could potentially lead to altered cell responsiveness upon subsequent environmental challenge. All analyses were performed using primary lung macrophages collected via bronchoalveolar lavage from healthy human subjects with normal pulmonary function. Epigenome-wide DNA methylation was examined via Infinium MethylationEPIC (850K) array and validated by targeted next-generation bisulfite sequencing. We observed 95 CpG loci with significant differential methylation in lung macrophages, comparing upper lobe to lower lobe (all false discovery rate < 0.05). Several of these genes, including CLIP4, HSH2D, NR4A1, SNX10, and TYK2, have been implicated as participants in inflammatory/immune-related biological processes. Functionally, we identified phenotypic differences in oxygen use, comparing upper versus lower lung macrophages. Our results support a hypothesis that epigenetic changes, specifically DNA methylation, at a multitude of gene loci in lung macrophages are associated with metabolic differences regionally in lung.

Insights from deconvolution of cell subtype proportions enhance the interpretation of functional genomic data

Cell subtype proportion variability between samples contributes significantly to the variation of functional genomic properties such as gene expression or DNA methylation. Although the impact of the variation of cell subtype composition on measured genomic quantities is recognized, and some innovative tools have been developed for the analysis of heterogeneous samples, most functional genomics studies using samples with mixed cell types still ignore the influence of cell subtype proportion variation, or just deal with it as a nuisance variable to be eliminated. Here we demonstrate how harvesting information about cell subtype proportions from functional genomics data can provide insights into cellular changes associated with phenotypes. We focused on two types of mixed cell populations, human blood and mouse kidney. Cell type prediction is well developed in the former, but not currently in the latter. Estimating the cellular repertoire is easier when a reference dataset from purified samples of all cell types in the tissue is available, as is the case for blood. However, reference datasets are not available for most other tissues, such as the kidney. In this study, we showed that the proportion of alterations attributable to changes in the cellular composition varies strikingly in the two disorders (asthma and systemic lupus erythematosus), suggesting that the contribution of cell subtype proportion changes to functional genomic properties can be disease-specific. We also showed that a reference dataset from a single-cell RNA-seq study successfully estimated the cell subtype proportions in mouse kidney and allowed us to distinguish altered cell subtype differences between two different knock-out mouse models, both of which had reported a reduced number of glomeruli compared to their wild-type counterparts. These findings demonstrate that testing for changes in cell subtype proportions between conditions can yield important insights in functional genomics studies.

Genes CEP55, FOXD3, FOXF2, GNAO1, GRIA4, and KCNA5 as potential diagnostic biomarkers in colorectal cancer

Background

Colorectal cancer (CRC) is one of the leading causes of death by cancer worldwide and in need of novel potential diagnostic biomarkers for early discovery.

Methods

We conducted a two-step study. We first employed bioinformatics on data from The Cancer Genome Atlas to obtain potential biomarkers and then experimentally validated some of them on our clinical samples. Our aim was to find a methylation alteration common to all clusters, with the potential of becoming a diagnostic biomarker in CRC.

Results

Unsupervised clustering of methylation data resulted in four clusters, none of which had a known common genetic or epigenetic event, such as mutations or methylation. The intersect among clusters and regulatory regions resulted in 590 aberrantly methylated probes, belonging to 198 differentially expressed genes. After performing pathway and functional analysis on differentially expressed genes, we selected six genes: CEP55, FOXD3, FOXF2, GNAO1, GRIA4 and KCNA5, for further experimental validation on our own clinical samples. In silico analysis demonstrated that CEP55 was hypomethylated in 98.7% and up-regulated in 95.0% of samples. Genes FOXD3, FOXF2, GNAO1, GRIA4 and KCNA5 were hypermethylated in 97.9, 81.1, 80.3, 98.4 and 94.0%, and down-regulated in 98.3, 98.9, 98.1, 98.1 and 98.6% of samples, respectively. Our experimental data show CEP55 was hypomethylated in 97.3% of samples and down-regulated in all samples, while FOXD3, FOXF2, GNAO1, GRIA4 and KCNA5 were hypermethylated in 100.0, 90.2, 100.0, 99.1 and 100.0%, and down-regulated in 68.0, 76.0, 96.0, 95.2 and 84.0% of samples, respectively. Results of in silico and our experimental analyses showed that more than 97% of samples had at least four methylation markers altered.

Conclusions

Using bioinformatics followed by experimental validation, we identified a set of six genes that were differentially expressed in CRC compared to normal mucosa and whose expression seems to be methylation dependent. Moreover, all of these six genes were common in all methylation clusters and mutation statuses of CRC and as such are believed to be an early event in human CRC carcinogenesis and to represent potential CRC biomarkers.

Electronic supplementary material

The online version of this article (10.1186/s12920-019-0501-z) contains supplementary material, which is available to authorized users.

Integrative Genome-Scale DNA Methylation Analysis of a Large and Unselected Cohort Reveals 5 Distinct Subtypes of Colorectal Adenocarcinomas

BACKGROUND & AIMS

Colorectal cancer is an epigenetically heterogeneous disease, however, the extent and spectrum of the CpG island methylator phenotype (CIMP) is not clear.

METHODS

Genome-scale methylation and transcript expression were measured by DNA Methylation and RNA expression microarray in 216 unselected colorectal cancers, and findings were validated using The Cancer Genome Atlas 450K and RNA sequencing data. Mutations in epigenetic regulators were assessed using CIMP-subtyped Cancer Genome Atlas exomes.

RESULTS

CIMP-high cancers dichotomized into CIMP-H1 and CIMP-H2 based on methylation profile. KRAS mutation was associated significantly with CIMP-H2 cancers, but not CIMP-H1 cancers. Congruent with increasing methylation, there was a stepwise increase in patient age from 62 years in the CIMP-negative subgroup to 75 years in the CIMP-H1 subgroup (P < .0001). CIMP-H1 predominantly comprised consensus molecular subtype 1 cancers (70%) whereas consensus molecular subtype 3 was over-represented in the CIMP-H2 subgroup (55%). Polycomb Repressive Complex-2 (PRC2)-marked loci were subjected to significant gene body methylation in CIMP cancers (P < 1.6 × 10^-78). We identified oncogenes susceptible to gene body methylation and Wnt pathway antagonists resistant to gene body methylation. CIMP cluster-specific mutations were observed in chromatin remodeling genes, such as in the SWItch/Sucrose Non-Fermentable and Chromodomain Helicase DNA-Binding gene families.

CONCLUSIONS

There are 5 clinically and molecularly distinct subgroups of colorectal cancer. We show a striking association between CIMP and age, sex, and tumor location, and identify a role for gene body methylation in the progression of serrated neoplasia. These data support our recent findings that CIMP is uncommon in young patients and that BRAF mutant polyps in young patients may have limited potential for malignant progression.

A Network-guided Association Mapping Approach from DNA Methylation to Disease

Aberrant DNA methylation may contribute to development of cancer. However, understanding the associations between DNA methylation and cancer remains a challenge because of the complex mechanisms involved in the associations and insufficient sample sizes. The unprecedented wealth of DNA methylation, gene expression and disease status data give us a new opportunity to design machine learning methods to investigate the underlying associated mechanisms. In this paper, we propose a network-guided association mapping approach from DNA methylation to disease (NAMDD). Compared with existing methods, NAMDD finds methylation-disease path associations by integrating analysis of multiple data combined with a stability selection strategy, thereby mining more information in the datasets and improving the quality of resultant methylation sites. The experimental results on both synthetic and real ovarian cancer data show that NAMDD substantially outperforms former disease-related methylation site research methods (including NsRRR and PCLOGIT) under false positive control. Furthermore, we applied NAMDD to ovarian cancer data, identified significant path associations and provided hypothetical biological path associations to explain our findings.

Genome-wide DNA methylation differences according to oestrogen receptor beta status in colorectal cancer

Involvement of sex hormones in colorectal cancer (CRC) development has been linked to oestrogen receptor β (ERβ). Expression of ERβ is found reduced in tumour tissue and inversely related to mortality. However, mechanisms are not well understood. Our study aimed to detect differentially methylated genes associated with ERβ expression, which could point to mechanisms by which ERβ could influence risk and prognosis of CRC. Epigenome-wide DNA methylation profiling was performed using Illumina HumanMethylation450k BeadChip arrays in two independent tumour sample sets of CRC patients recruited in 2003-2010 by the German DACHS study (discovery cohort n = 917, replication cohort n = 907). ERβ expression was measured using immunohistochemistry and scored as negative, moderate and high. Differentially methylated CpG sites and genomic regions were determined using limma in the R-package RnBeads. For the comparison of tumours with moderate/high ERβ versus negative expression, differentially methylated CpG sites were identified but not confirmed by replication. Comparing tumours of high with tumours of negative ERβ expression revealed 2,904 differentially methylated CpG sites of which 403 were replicated (FDR adjusted p-value<0.05). Replicated CpGs were annotated to genes such as CD36, HK1 or LRP5. A survival analysis indicates that 30 of the replicated CpGs are also associated with overall survival (FDR-adjusted p-value<0.05). The regional analysis identified 60 differentially methylated promotor regions. The epigenome-wide analysis identified both novel genes as well as genes already implicated in CRC. Follow-up mechanistic studies to better understand the regulatory role of ERβ could inform potential targets for improving treatment or prevention of CRC.

Subtypes of Barrett's oesophagus and oesophageal adenocarcinoma based on genome-wide methylation analysis

OBJECTIVE

To identify and characterise DNA methylation subtypes in oesophageal adenocarcinoma (EAC) and its precursor Barrett's oesophagus (BE).

DESIGN

We performed genome-wide DNA methylation profiling on samples of non-dysplastic BE from cancer-free patients (n=59), EAC (n=23), normal squamous oesophagus (n=33) and normal fundus (n=9), and identified methylation subtypes using a recursively partitioned mixture model. We assessed genomic alterations for 9 BE and 22 EAC samples with massively parallel sequencing of 243 EAC-associated genes, and we conducted integrative analyses with transcriptome data to identify epigenetically repressed genes. We also carried out in vitro experiments treating EAC cell lines with 5-Aza-2'-Deoxycytidine (5-Aza-dC), short hairpin RNA knockdown and anticancer therapies.

RESULTS

We identified and validated four methylation subtypes of EAC and BE. The high methylator subtype (HM) of EAC had the greatest number of activating events in ERBB2 (p<0.05, Student's t-test) and the highest global mutation load (p<0.05, Fisher's exact test). PTPN13 was silenced by aberrant methylation in the HM subtype preferentially and in 57% of EACs overall. In EAC cell lines, 5-Aza-dC treatment restored PTPN13 expression and significantly decreased its promoter methylation in HM cell lines (p<0.05, Welch's t-test). Inhibition of PTPN13 expression in the SK-GT-4 EAC cell line promoted proliferation, colony formation and migration, and increased phosphorylation in ERBB2/EGFR/Src kinase pathways. Finally, EAC cell lines showed subtype-specific responses to topotecan, SN-38 and palbociclib treatment.

CONCLUSIONS

We identified and characterised methylator subtypes in BE and EAC. We further demonstrated the biological and clinical relevance of EAC methylator subtypes, which may ultimately help guide clinical management of patients with EAC.

Epigenetic analysis in rheumatoid arthritis synoviocytes

Rheumatoid arthritis (RA) is a complex chronic systematic disease with progressive destruction of the joints by invasive synoviocytes. To characterize the key regulators involved in the development of RA, we obtained multilayer epigenomics data including DNA methylation by whole-genome bisulfite sequencing, miRNA profiles, genetic variations by whole-exome sequencing, and mRNA profiles from synoviocytes of RA and osteoarthritis (OA) patients. The overall DNA methylation patterns were not much different between RA and OA, but 523 low-methylated regions (LMRs) were specific to RA. The LMRs were preferentially localized at the 5′ introns and overlapped with transcription factor binding motifs for GLI1, RUNX2, and TFAP2A/C. Single base-scale differentially methylated CpGs were linked with several networks related to wound response, tissue development, collagen fibril organization, and the TGF-β receptor signaling pathway. Further, the DNA methylation of 201 CpGs was significantly correlated with 27 expressed miRNA genes. Our interpretation of epigenomic data of the synoviocytes from RA and OA patients is an informative resource to further investigate regulatory elements and biomarkers responsible for the pathophysiology of RA and OA.

Whole genome analysis of synoviocytes, specialized cells in the joint-lubricating synovial fluid, sheds light on the pathogenic mechanisms of rheumatoid arthritis (RA). Around 350 million people worldwide suffer joint pain and stiffness due to RA, but the inheritance pattern of the disease remains unclear. A study led by Tae-Young Roh at Pohang University of Science and Technology, South Korea, reveals a distinct pattern of chemical tags on the DNA of synoviocytes from RA patients. Differences in methyl group tags in over 500 regions of the genome influenced the expression of RA-associated genes and of microRNAs, small RNA molecules that are also involved in the regulation of gene expression. These differentially methylated sites may not only represent potential disease biomarkers, but also offer new insights into the regulation of RA-relevant genes.

Description of exposure profiles for seven environmental chemicals in a US population using recursive partition mixture modeling (RPMM)

Biomonitoring studies have shown that humans are exposed to numerous environmental chemicals. Previous work provides limited insights into the dynamic relationship between different chemicals within a population. The objective of this study is to develop an analytical method identifying exposure profiles of seven common environmental chemicals and determine how exposure profiles differ by sociodemographic groups and National Health and Nutrition Examination Survey (NHANES) 2003-2012 cycle year. We used recursive partition mixture modeling (RPMM) to define classes of the population with similar exposure profiles of lead, cadmium, 2,4-dichlorophenol, 2,5-dichlorophenol, bisphenol A (BPA), triclosan, and benzophenone-3 in individuals aged ≥6 years. Additionally, quasibinomial logistic regression was used to examine the association between each class and selected demographic characteristics. Eight exposure profiles were identified. Individuals who clustered together and had the highest chemical exposures were more likely to be older, to be Non-Hispanic Black (NHB) or Other Hispanic (OH), more likely to live below the poverty line, more likely to be male, and more likely to have participated in the earlier NHANES cycle (2003-2004). The developed method described the dynamic relationship between chemicals and shows that this relationship is different for subpopulations based on their sociodemographic characteristics.

Genome-wide DNA methylation profiling shows a distinct epigenetic signature associated with lung macrophages in cystic fibrosis

Background

Lung macrophages are major participants in the pulmonary innate immune response. In the cystic fibrosis (CF) lung, the inability of lung macrophages to successfully regulate the exaggerated inflammatory response suggests dysfunctional innate immune cell function. In this study, we aim to gain insight into innate immune cell dysfunction in CF by investigating alterations in DNA methylation in bronchoalveolar lavage (BAL) cells, composed primarily of lung macrophages of CF subjects compared with healthy controls. All analyses were performed using primary alveolar macrophages from human subjects collected via bronchoalveolar lavage. Epigenome-wide DNA methylation was examined via Illumina MethylationEPIC (850 K) array. Targeted next-generation bisulfite sequencing was used to validate selected differentially methylated CpGs. Methylation-based sample classification was performed using the recursively partitioned mixture model (RPMM) and was tested against sample case-control status. Differentially methylated loci were identified by fitting linear models with adjustment of age, sex, estimated cell type proportions, and repeat measurement.

Results

RPMM class membership was significantly associated with the CF disease status (P = 0.026). One hundred nine CpG loci were differentially methylated in CF BAL cells (all FDR ≤ 0.1). The majority of differentially methylated loci in CF were hypo-methylated and found within non-promoter CpG islands as well as in putative enhancer regions and DNase hyper-sensitive regions.

Conclusions

These results support a hypothesis that epigenetic changes, specifically DNA methylation at a multitude of gene loci in lung macrophages, may participate, at least in part, in driving dysfunctional innate immune cells in the CF lung.

Electronic supplementary material

The online version of this article (10.1186/s13148-018-0580-2) contains supplementary material, which is available to authorized users.

Epigenetic Changes Associated with Early Life Experiences: Saliva, A Biospecimen for DNA Methylation Signatures

BACKGROUND

Adverse Childhood Experiences (ACEs), which include traumatic injury, are associated with poor health outcomes in later life, yet the biological mechanisms mediating this association are unknown. Neurocircuitry, immune system and hormone regulation differ from normal in adults reporting ACEs. These systems could be affected by epigenetic changes, including methylation of cytosine (5mC) in genomic DNA, activated by ACEs. Since 5mC levels influence gene expression and can be long-lasting, altered 5mC status at specific sites or throughout the genome is hypothesized to influence mental and physical outcomes after ACE(s). Human and animal studies support this, with animal models allowing experiments for attributing causality. Here we provide a lengthy introduction and background on 5mC and the impact of early life adversity.

OBJECTIVE

Next we address the issue of a mixture of cell types in saliva, the most accessible biospecimen for 5mC analysis. Typical human bio-specimens for 5mC analysis include saliva or buccal swabs, whole blood or types of blood cells, tumors and post-mortem brain. In children saliva is the most accessible biospecimen, but contains a mixture of keratinocytes and white blood cells, as do buccal swabs. Even in saliva from the same individual at different time points, cell composition may differ widely. Similar issues affect analysis in blood, where nucleated cells represent a wide array of white blood cell types. Unless variations in ratios of these cells between each sample are included in the analysis, results can be unreliable.

METHODS

Several different biochemical assays are available to test for site-specific methylation levels genome-wide, each producing different information, with high-density arrays being the easiest to use, and bisulfite whole genome sequencing the most comprehensive. We compare results from different assays and use high-throughput computational processing to deconvolve cell composition in saliva samples.

RESULTS

Here we present examples demonstrating the critical importance of determining the relative contribution of blood cells versus keratinocytes to the 5mC profile found in saliva. We further describe a strategy to perform a reference-based computational correction for cell composition, and therefore to identify differential methylation patterns due to experience, or for the diagnosis of phenotypes that correlate between traits, such as hormone levels, trauma status and various mental health outcomes.

CONCLUSION

Specific sites that respond to adversity with altered methylation levels in either blood cells, keratinocytes or both can be identified by this rigorous approach, which will then be useful as diagnostic biomarkers and therapeutic targets.

Comprehensive human cell-type methylation atlas reveals origins of circulating cell-free DNA in health and disease

Methylation patterns of circulating cell-free DNA (cfDNA) contain rich information about recent cell death events in the body. Here, we present an approach for unbiased determination of the tissue origins of cfDNA, using a reference methylation atlas of 25 human tissues and cell types. The method is validated using in silico simulations as well as in vitro mixes of DNA from different tissue sources at known proportions. We show that plasma cfDNA of healthy donors originates from white blood cells (55%), erythrocyte progenitors (30%), vascular endothelial cells (10%) and hepatocytes (1%). Deconvolution of cfDNA from patients reveals tissue contributions that agree with clinical findings in sepsis, islet transplantation, cancer of the colon, lung, breast and prostate, and cancer of unknown primary. We propose a procedure which can be easily adapted to study the cellular contributors to cfDNA in many settings, opening a broad window into healthy and pathologic human tissue dynamics.

The methylation status of circulating cell-free DNA (cfDNA) can be informative about recent cell death events. Here the authors present an approach to determine the tissue origins of cfDNA, using a reference methylation atlas of 25 human tissues and cell types, and find that cfDNA from patients reveals tissue contributions that agree with clinical findings.

CancerDetector: ultrasensitive and non-invasive cancer detection at the resolution of individual reads using cell-free DNA methylation sequencing data

The detection of tumor-derived cell-free DNA in plasma is one of the most promising directions in cancer diagnosis. The major challenge in such an approach is how to identify the tiny amount of tumor DNAs out of total cell-free DNAs in blood. Here we propose an ultrasensitive cancer detection method, termed 'CancerDetector', using the DNA methylation profiles of cell-free DNAs. The key of our method is to probabilistically model the joint methylation states of multiple adjacent CpG sites on an individual sequencing read, in order to exploit the pervasive nature of DNA methylation for signal amplification. Therefore, CancerDetector can sensitively identify a trace amount of tumor cfDNAs in plasma, at the level of individual reads. We evaluated CancerDetector on the simulated data, and showed a high concordance of the predicted and true tumor fraction. Testing CancerDetector on real plasma data demonstrated its high sensitivity and specificity in detecting tumor cfDNAs. In addition, the predicted tumor fraction showed great consistency with tumor size and survival outcome. Note that all of those testing were performed on sequencing data at low to medium coverage (1× to 10×). Therefore, CancerDetector holds the great potential to detect cancer early and cost-effectively.

armDNA: A functional beta model for detecting age-related genomewide DNA methylation marks

DNA methylation has been shown to play an important role in many complex diseases. The rapid development of high-throughput DNA methylation scan technologies provides great opportunities for genomewide DNA methylation-disease association studies. As methylation is a dynamic process involving time, it is quite plausible that age contributes to its variation to a large extent. Therefore, in analyzing genomewide DNA methylation data, it is important to identify age-related DNA methylation marks and delineate their functional relationship. This helps us to better understand the underlying biological mechanism and facilitate early diagnosis and prognosis analysis of complex diseases. We develop a functional beta model for analyzing DNA methylation data and detecting age-related DNA methylation marks on the whole genome by naturally taking sampling scheme into account and accommodating flexible age-methylation dynamics. We focus on DNA methylation data obtained through the widely used bisulfite conversion technique and propose to use a beta model to relate the DNA methylation level to the age. Adjusting for certain confounders, the functional age effect is left completely unspecified, offering great flexibility and allowing extra data dynamics. An efficient algorithm is developed for estimating unknown parameters, and the Wald test is used to detect age-related DNA methylation marks. Simulation studies and several real data applications were provided to demonstrate the performance of the proposed method.

Heterogeneous expression of cytokines accounts for clinical diversity and refines prognostication in CMML

Chronic myelomonocytic leukemia (CMML) is a clinically heterogeneous neoplasm in which JAK2 inhibition has demonstrated reductions in inflammatory cytokines and promising clinical activity. We hypothesize that annotation of inflammatory cytokines may uncover mutation-independent cytokine subsets associated with novel CMML prognostic features. A Luminex cytokine profiling assay was utilized to profile cryopreserved peripheral blood plasma from 215 CMML cases from three academic centers, along with center-specific, age-matched plasma controls. Significant differences were observed between CMML patients and healthy controls in 23 out of 45 cytokines including increased cytokine levels in IL-8, IP-10, IL-1RA, TNF-α, IL-6, MCP-1/CCL2, hepatocyte growth factor (HGF), M-CSF, VEGF, IL-4, and IL-2RA. Cytokine associations were identified with clinical and genetic features, and Euclidian cluster analysis identified three distinct cluster groups associated with important clinical and genetic features in CMML. CMML patients with decreased IL-10 expression had a poor overall survival when compared to CMML patients with elevated expression of IL-10 (P = 0.017), even when adjusted for ASXL1 mutation and other prognostic features. Incorporating IL-10 with the Mayo Molecular Model statistically improved the prognostic ability of the model. These established cytokines, such as IL-10, as prognostically relevant and represent the first comprehensive study exploring the clinical implications of the CMML inflammatory state.

Epigenetic dysregulation in adrenocortical carcinoma, a systematic review of the literature

Adrenocortical carcinoma (ACC) is a rare and aggressive endocrine malignancy with a poor prognosis. Diagnosis and treatment of this tumor remains challenging. The Weiss score, the current gold standard for the histopathological diagnosis of ACC, lacks diagnostic accuracy of borderline tumors (Weiss score 2 or 3) and is subject to inter observer variability. Furthermore, adjuvant and palliative systemic therapy have limited effect and no proven overall survival benefit. A better insight in the molecular background of ACC might identify markers that improve diagnostic accuracy, predict treatment response or even provide novel therapeutic targets. This systematic review of the literature aims to provide an overview of alterations in DNA methylation, histone modifications and their potential clinical relevance in ACC.

Accounting for tumor purity improves cancer subtype classification from DNA methylation data

Motivation

Tumor sample classification has long been an important task in cancer research. Classifying tumors into different subtypes greatly benefits therapeutic development and facilitates application of precision medicine on patients. In practice, solid tumor tissue samples obtained from clinical settings are always mixtures of cancer and normal cells. Thus, the data obtained from these samples are mixed signals. The 'tumor purity', or the percentage of cancer cells in cancer tissue sample, will bias the clustering results if not properly accounted for.

Results

In this article, we developed a model-based clustering method and an R function which uses DNA methylation microarray data to infer tumor subtypes with the consideration of tumor purity. Simulation studies and the analyses of The Cancer Genome Atlas data demonstrate improved results compared with existing methods.

Availability and implementation

InfiniumClust is part of R package InfiniumPurify , which is freely available from CRAN ( https://cran.r-project.org/web/packages/InfiniumPurify/index.html ).

Contact

hao.wu@emory.edu or xqzheng@shnu.edu.cn.

Supplementary information

Supplementary data are available at Bioinformatics online.

Region-Specific Methylation Profiling in Acute Myeloid Leukemia

Alteration of DNA methylation level in cancer diseases leads to deregulation of gene expression-silencing of tumor suppressor genes and enhancing of protooncogenes. There are several tools devoted to the problem of identification of CpG sites' demethylation but majority of them focuses on single site level and does not allow for quantification of region methylation changes. The aim was to create an adaptive algorithm supporting detection of differentially methylated CpG sites and genomic regions specific for acute myeloid leukemia. Knowledge on AML methylation fingerprint helps in better understanding the epigenetics of leukemogenesis. Proposed algorithm is data driven and does not use predefined quantification thresholds. Gaussian mixture modeling supports classification of CpG sites to several levels of demethylation. p value integration allows for translation from single site demethylation to the demethylation of gene promoter and body regions. Methylation profiles of healthy controls and AML patients were examined (GEO:GSE63409). The differences in whole genome methylation profiles were observed. The methylation profile differs significantly among genomic regions. The lowest methylation level was observed for promoter regions, while sites from intergenic regions were by average higher methylated. The observed number of AML related down methylated sites has not substantially exceeded the expected number by chance. Intergenic regions were characterized by the highest percentage of AML up methylated sites. Methylation enhancement/diminution is the most frequent for intergenic region while methylation compensation (positive or negative) is specific for promoter regions. Functional analysis performed for AML down methylated or extreme high up methylated genes showed strong connection to the leukemic processes.

Considerations for Design and Analysis of DNA Methylation Studies

The number of epigenetic studies is exponentially increasing. There is anticipation that DNA methylation may close gaps in our understanding of disease etiology, and how certain risk factors affect health and disease, but also that it has potential as a biomarker for disease. Human DNA methylation studies require careful considerations for design and analysis including population and tissue selection, population stratification, cell heterogeneity, confounding, temporality, sample size, appropriate statistical analysis, and validation of results. In this chapter, we discuss relevant aspects for the design of DNA methylation studies and delineate essential steps for their analysis. Specifically, we summarize methods used to extricate biologic signals from technical noise, and statistical approaches to capture meaningful variability based on the research hypothesis.

Clustering high-dimensional mixed data to uncover sub-phenotypes: joint analysis of phenotypic and genotypic data

The LIPGENE-SU.VI.MAX study, like many others, recorded high-dimensional continuous phenotypic data and categorical genotypic data. LIPGENE-SU.VI.MAX focuses on the need to account for both phenotypic and genetic factors when studying the metabolic syndrome (MetS), a complex disorder that can lead to higher risk of type 2 diabetes and cardiovascular disease. Interest lies in clustering the LIPGENE-SU.VI.MAX participants into homogeneous groups or sub-phenotypes, by jointly considering their phenotypic and genotypic data, and in determining which variables are discriminatory. A novel latent variable model that elegantly accommodates high dimensional, mixed data is developed to cluster LIPGENE-SU.VI.MAX participants using a Bayesian finite mixture model. A computationally efficient variable selection algorithm is incorporated, estimation is via a Gibbs sampling algorithm and an approximate BIC-MCMC criterion is developed to select the optimal model. Two clusters or sub-phenotypes ('healthy' and 'at risk') are uncovered. A small subset of variables is deemed discriminatory, which notably includes phenotypic and genotypic variables, highlighting the need to jointly consider both factors. Further, 7 years after the LIPGENE-SU.VI.MAX data were collected, participants underwent further analysis to diagnose presence or absence of the MetS. The two uncovered sub-phenotypes strongly correspond to the 7-year follow-up disease classification, highlighting the role of phenotypic and genotypic factors in the MetS and emphasising the potential utility of the clustering approach in early screening. Additionally, the ability of the proposed approach to define the uncertainty in sub-phenotype membership at the participant level is synonymous with the concepts of precision medicine and nutrition. Copyright © 2017 John Wiley & Sons, Ltd.

Epigenetic changes in myelofibrosis: Distinct methylation changes in the myeloid compartments and in cases with ASXL1 mutations

This is the first study to compare genome-wide DNA methylation profiles of sorted blood cells from myelofibrosis (MF) patients and healthy controls. We found that differentially methylated CpG sites located to genes involved in 'cancer' and 'embryonic development' in MF CD34+ cells, in 'inflammatory disease' in MF mononuclear cells, and in 'immunological diseases' in MF granulocytes. Only few differentially methylated CpG sites were common among the three cell populations. Mutations in the epigenetic regulators ASXL1 (47%) and TET2 (20%) were not associated with a specific DNA methylation pattern using an unsupervised approach. However, in a supervised analysis of ASXL1 mutated versus wild-type cases, differentially methylated CpG sites were enriched in regions marked by histone H3K4me1, histone H3K27me3, and the bivalent histone mark H3K27me3 + H3K4me3 in human CD34+ cells. Hypermethylation of selected CpG sites was confirmed in a separate validation cohort of 30 MF patients by pyrosequencing. Altogether, we show that individual MF cell populations have distinct differentially methylated genes relative to their normal counterparts, which likely contribute to the phenotypic characteristics of MF. Furthermore, differentially methylated CpG sites in ASXL1 mutated MF cases are found in regulatory regions that could be associated with aberrant gene expression of ASXL1 target genes.

Identification of relevant subtypes via preweighted sparse clustering

Cluster analysis methods are used to identify homogeneous subgroups in a data set. In biomedical applications, one frequently applies cluster analysis in order to identify biologically interesting subgroups. In particular, one may wish to identify subgroups that are associated with a particular outcome of interest. Conventional clustering methods generally do not identify such subgroups, particularly when there are a large number of high-variance features in the data set. Conventional methods may identify clusters associated with these high-variance features when one wishes to obtain secondary clusters that are more interesting biologically or more strongly associated with a particular outcome of interest. A modification of sparse clustering can be used to identify such secondary clusters or clusters associated with an outcome of interest. This method correctly identifies such clusters of interest in several simulation scenarios. The method is also applied to a large prospective cohort study of temporomandibular disorders and a leukemia microarray data set.

The development of lower respiratory tract microbiome in mice

BACKGROUND

Although culture-independent methods have paved the way for characterization of the lung microbiome, the dynamic changes in the lung microbiome from neonatal stage to adult age have not been investigated.

RESULTS

In this study, we tracked changes in composition and diversity of the lung microbiome in C57BL/6N mice, starting from 1-week-old neonates to 8-week-old mice. Towards this, the lungs were sterilely excised from mice of different ages from 1 to 8 weeks. High-throughput DNA sequencing of the 16S rRNA gene followed by composition and diversity analysis was utilized to decipher the microbiome in these samples. Microbiome analysis suggests that the changes in the lung microbiome correlated with age. The lung microbiome was primarily dominated by phyla Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria in all the stages from week 1 to week 8 after birth. Although Defluvibacter was the predominant genus in 1-week-old neonatal mice, Streptococcus became the dominant genus at the age of 2 weeks. Lactobacillus, Defluvibacter, Streptococcus, and Achromobacter were the dominant genera in 3-week-old mice, while Lactobacillus and Achromobacter were the most abundant genera in 4-week-old mice. Interestingly, relatively greater diversity (at the genus level) during the age of 5 to 6 weeks was observed as compared to the earlier weeks. The diversity of the lung microbiome remained stable between 6 and 8 weeks of age.

CONCLUSIONS

In summary, we have tracked the development of the lung microbiome in mice from an early age of 1 week to adulthood. The lung microbiome is dominated by the phyla Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria. However, dynamic changes were observed at the genus level. Relatively higher richness in the microbial diversity was achieved by age of 6 weeks and then maintained at later ages. We believe that this study improves our understanding of the development of the mice lung microbiome and will facilitate further analyses of the role of the lung microbiome in chronic lung diseases.

Biomarkers used in studying air pollution exposure during pregnancy and perinatal outcomes: a review

PURPOSE

This review focuses on studies among pregnant women that used biomarkers to assess air pollution exposure, or to understand the mechanisms by which it affects perinatal outcomes.

METHODS

We searched PubMed and Google scholar databases to find articles.

RESULTS AND CONCLUSIONS

We found 29 articles, mostly consisting of cohort studies. Interpolation models were most frequently used to assess exposure. The most consistent positive association was between polycyclic aromatic hydrocarbon (PAH) exposure during entire pregnancy and cord blood PAH DNA adducts. Exposure to particulate matter (PM) and nitrogen dioxide (NO₂) showed consistent inverse associations with mitochondrial DNA (mtDNA) content, particularly in the third trimester of pregnancy. No single pollutant showed strong associations with all the biomarkers included in this review. C-reactive proteins (CRPs) and oxidative stress markers increased, whereas telomere length decreased with increasing air pollution exposure. Placental global DNA methylation and mtDNA methylation showed contrasting results with air pollution exposure, the mechanism behind which is unclear. Most studies except those on PAH DNA adducts and mtDNA content provided insufficient evidence for characterizing a critical exposure window. Further research using biomarkers is warranted to understand the relationship between air pollution and perinatal outcomes.

Placental lipoprotein lipase DNA methylation alterations are associated with gestational diabetes and body composition at 5 years of age

Gestational diabetes mellitus (GDM) is associated with obesity in childhood. This suggests that consequences of in utero exposure to maternal hyperglycemia extend beyond the fetal development, possibly through epigenetic programming. The aims of this study were to assess whether placental DNA methylation (DNAm) marks were associated with maternal GDM status and to offspring body composition at 5 years old in a prospective birth cohort. DNAm levels were measured in the fetal side of the placenta in 66 samples (24 from GDM mothers) using bisDNA-pyrosequencing. Anthropometric and body composition (bioimpedance) were measured in children at 5 years of age. Mann-Whitney and Spearman tests were used to assess associations between GDM, placental DNAm levels at the lipoprotein lipase (LPL) locus and children's weight, height, body mass index (BMI), body fat, and lean masses at 5 years of age. Weight, height, and BMI z-scores were computed according to the World Health Organization growth chart. Analyses were adjusted for gestational age at birth, child sex, maternal age, and pre-pregnancy BMI. LPL DNAm levels were positively correlated with birth weight z-scores (r = 0.252, P = 0.04), and with mid-childhood weight z-scores (r = 0.314, P = 0.01) and fat mass (r = 0.275, P = 0.04), and negatively correlated with lean mass (r = -0.306, P = 0.02). We found a negative correlation between LPL DNAm and mRNA levels in placenta (r = -0.459; P < 0.001), which highlights the regulation of transcriptional activity by these epivariations. We demonstrated that alterations in fetal placental DNAm levels at the LPL gene locus are associated with the anthropometric profile in children at 5 years of age. These findings support the concept of fetal metabolic programming through epigenetic changes.

PLMET: A Novel Pseudolikelihood-Based EM Test for Homogeneity in Generalilzed Exponential Tilt Mixture Models

Motivated by analyses of DNA methylation data, we propose a semiparametric mixture model, namely the generalized exponential tilt mixture model, to account for heterogeneity between differentially methylated and non-differentially methylated subjects in the cancer group, and capture the differences in higher order moments (e.g. mean and variance) between subjects in cancer and normal groups. A pairwise pseudolikelihood is constructed to eliminate the unknown nuisance function. To circumvent boundary and non-identifiability problems as in parametric mixture models, we modify the pseudolikelihood by adding a penalty function. In addition, the test with simple asymptotic distribution has computational advantages compared with permutation-based test for high-dimensional genetic or epigenetic data. We propose a pseudolikelihood based expectation-maximization test, and show the proposed test follows a simple chi-squared limiting distribution. Simulation studies show that the proposed test controls Type I errors well and has better power compared to several current tests. In particular, the proposed test outperforms the commonly used tests under all simulation settings considered, especially when there are variance differences between two groups. The proposed test is applied to a real data set to identify differentially methylated sites between ovarian cancer subjects and normal subjects.

DNA Methylation Identifies Loci Distinguishing Hereditary Nonpolyposis Colorectal Cancer Without Germ-Line MLH1/MSH2 Mutation from Sporadic Colorectal Cancer

Objectives:

Roughly half of hereditary nonpolyposis colorectal cancer (HNPCC) cases are Lynch syndrome and exhibit germ-line mutations in DNA mismatch repair (MMR) genes; the other half are familial colorectal cancer (CRC) type X (FCCTX) and are MMR proficient. About 70% of Lynch syndrome tumors have germ-line MLH1 or MSH2 mutations. The clinical presentation, histopathological features, and carcinogenesis of FCCTX resemble those of sporadic MMR-proficient colorectal tumors. It is of interest to obtain biomarkers that distinguish FCCTX from sporadic microsatellite stable (MSS) CRC, to develop preventive strategies.

Methods:

The tumors and adjacent normal tissues of 40 patients with HNPCC were assayed using the Illumina Infinium HumanMethylation27 (HM27) BeadChip to assess the DNA methylation level at about 27,000 loci. The germ-line mutation status of MLH1 and MSH2 and the microsatellite instability status in these patients were obtained. Genome-wide DNA methylation measurements of three groups of patients with general CRC were downloaded from public domain databases. Probes with DNA methylation levels that differed significantly between patients with sporadic MSS CRC and FCCTX were examined, to explore their potential as biomarkers.

Results:

We found that MSS HNPCC tumors were overwhelmingly hypomethylated compared with those from patient groups with other types of CRC, including germ-line MLH1/MSH2-mutated HNPCC and sporadic MSS CRC. Five gene-marker panels that exhibited a sensitivity of 100% and a specificity higher than 90% in both discovery and validation cohorts were proposed to distinguish MSS HNPCC tumors from sporadic MSS CRC.

Conclusions:

Our results warrant further investigation and validation. The loci identified here may become useful biomarkers for distinguishing between FCCTX and sporadic MSS CRC tumors.

Towards understanding the breast cancer epigenome: a comparison of genome-wide DNA methylation and gene expression data

Until recently, an elevated disease risk has been ascribed to a genetic predisposition, however, exciting progress over the past years has discovered alternate elements of inheritance that involve epigenetic regulation. Epigenetic changes are heritably stable alterations that include DNA methylation, histone modifications and RNA-mediated silencing. Aberrant DNA methylation is a common molecular basis for a number of important human diseases, including breast cancer. Changes in DNA methylation profoundly affect global gene expression patterns. What is emerging is a more dynamic and complex association between DNA methylation and gene expression than previously believed. Although many tools have already been developed for analyzing genome-wide gene expression data, tools for analyzing genome-wide DNA methylation have not yet reached the same level of refinement.

Here we provide an in-depth analysis of DNA methylation in parallel with gene expression data characteristics and describe the particularities of low-level and high-level analyses of DNA methylation data. Low-level analysis refers to pre-processing of methylation data (i.e. normalization, transformation and filtering), whereas high-level analysis is focused on illustrating the application of the widely used class comparison, class prediction and class discovery methods to DNA methylation data. Furthermore, we investigate the influence of DNA methylation on gene expression by measuring the correlation between the degree of CpG methylation and the level of expression and to explore the pattern of methylation as a function of the promoter region.

Differentially Methylated DNA Regions in Monozygotic Twin Pairs Discordant for Rheumatoid Arthritis: An Epigenome-Wide Study

Objectives

In an explorative epigenome-wide association study (EWAS) to search for gene independent, differentially methylated DNA positions and regions (DMRs) associated with rheumatoid arthritis (RA) by studying monozygotic (MZ) twin pairs discordant for RA.

Methods

Genomic DNA was isolated from whole blood samples from 28 MZ twin pairs discordant for RA. DNA methylation was measured using the HumanMethylation450 BeadChips. Smoking, anti-cyclic citrullinated peptide antibodies, and immunosuppressive treatment were included as covariates. Pathway analysis was performed using GREAT.

Results

Smoking was significantly associated with hypomethylation of a DMR overlapping the promoter region of the RNF5 and the AGPAT1, which are implicated in inflammation and autoimmunity, whereas DMARD treatment induced hypermethylation of the same region. Additionally, the promotor region of both S100A6 and EFCAB4B were hypomethylated, and both genes have previously been associated with RA. We replicated several candidate genes identified in a previous EWAS in treatment-naïve RA singletons. Gene-set analysis indicated the involvement of immunologic signatures and cancer-related pathways in RA.

Conclusion

We identified several differentially methylated regions associated with RA, which may represent environmental effects or consequences of the disease and plausible biological pathways pertinent to the pathogenesis of RA.

Bacterially-Associated Transcriptional Remodelling in a Distinct Genomic Subtype of Colorectal Cancer Provides a Plausible Molecular Basis for Disease Development

The relevance of specific microbial colonisation to colorectal cancer (CRC) disease pathogenesis is increasingly recognised, but our understanding of possible underlying molecular mechanisms that may link colonisation to disease in vivo remains limited. Here, we investigate the relationships between the most commonly studied CRC-associated bacteria (Enterotoxigenic Bacteroides fragilis, pks+ Escherichia coli, Fusobacterium spp., afaC+ E. coli, Enterococcus faecalis & Enteropathogenic E. coli) and altered transcriptomic and methylation profiles of CRC patients, in order to gain insight into the potential contribution of these bacteria in the aetiopathogenesis of CRC. We show that colonisation by E. faecalis and high levels of Fusobacterium is associated with a specific transcriptomic subtype of CRC that is characterised by CpG island methylation, microsatellite instability and a significant increase in inflammatory and DNA damage pathways. Analysis of the significant, bacterially-associated changes in host gene expression, both at the level of individual genes as well as pathways, revealed a transcriptional remodeling that provides a plausible mechanistic link between specific bacterial colonisation and colorectal cancer disease development and progression in this subtype; these included upregulation of REG3A, REG1A and REG1P in the case of high-level colonization by Fusobacterium, and CXCL10 and BMI1 in the case of colonisation by E. faecalis. The enrichment of both E. faecalis and Fusobacterium in this CRC subtype suggests that polymicrobial colonisation of the colonic epithelium may well be an important aspect of colonic tumourigenesis.

Drug discovery using clinical outcome-based Connectivity Mapping: application to ovarian cancer

Background

Epithelial ovarian cancer (EOC) is the fifth leading cause of cancer death among women in the United States (5 % of cancer deaths). The standard treatment for patients with advanced EOC is initial debulking surgery followed by carboplatin-paclitaxel combination chemotherapy. Unfortunately, with chemotherapy most patients relapse and die resulting in a five-year overall survival around 45 %. Thus, finding novel therapeutics for treating EOC is essential. Connectivity Mapping (CMAP) has been used widely in cancer drug discovery and generally has relied on cancer cell line gene expression and drug phenotype data. Therefore, we took a CMAP approach based on tumor information and clinical endpoints from high grade serous EOC patients.

Methods

We determined tumor gene expression signatures (e.g., sets of genes) associated with time to recurrence (with and without adjustment for additional clinical covariates) among patients within TCGA (n = 407) and, separately, from the Mayo Clinic (n = 326). Each gene signature was inputted into CMAP software (Broad Institute) to determine a set of drugs for which our signature “matches” the “reference” signature, and drugs that overlapped between the CMAP analyses and the two studies were carried forward for validation studies involving drug screens on a set of 10 EOC cell lines.

Results

Of the 11 drugs carried forward, five (mitoxantrone, podophyllotoxin, wortmannin, doxorubicin, and 17-AAG) were known a priori to be cytotoxics and were indeed shown to effect EOC cell viability.

Conclusions

Future research is needed to investigate the use of these CMAP and similar analyses for determining combination therapies that might work synergistically to kill cancer cells and to apply this in silico bioinformatics approach using clinical outcomes to other cancer drug screening studies.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-016-3149-5) contains supplementary material, which is available to authorized users.