Adjusting for Batch Effects in DNA Methylation Microarray Data, a Lesson Learned

The DNA methylation landscape across the TCR loci in patients with acute myeloid leukemia

The capacity of T cells to initiate anti-leukemia immune responses is determined by the ability of their receptors (TCRs) to recognize leukemia neoantigens. Epigenetic mechanisms including DNA methylation contribute to shaping the TCR repertoire composition and diversity. The DNA hypomethylating agents (HMAs) have been widely used in the treatment of acute myeloid leukemia (AML) and myelodysplastic syndrome (MDS). Whether DNA HMAs directly influence TCR gene loci methylation patterns remains unknown. By analyzing public datasets, we compared methylation patterns across TCR loci in AML patients and healthy controls. We also explored how HMAs influence TCR loci DNA methylation in patients with AML. While methylation patterns are largely conserved across the TCR loci, certain V genes exhibit high interindividual variability. Although overall methylation levels within the TCR loci did not show significant differences, specific sites, including 32 TRAV and 12 TRBV sites exhibited distinct methylation patterns when comparing T cells from healthy donors to those from patients with AML. In leukemic cells, decitabine treatment demethylates sites across the TRAV and TRBV genes. While not as significant, a similar pattern of demethylation is observed in T cells. Pretreatment AML samples exhibit higher methylation beta values in differentially methylated positions (DMPs) compared with non-DMPs. Methylation levels of certain TRAV and TRBV genes in leukemic cells are associated with patients' risk status. The presence of disease specific TCR loci methylated signatures that are associated with clinical outcome presents an opportunity for therapeutic intervention. HMAs can modulate the TCR loci methylation patterns, yet whether they could reprogram the TCR repertoire composition remains to be explored.

Epigenome-wide association study of lung cancer among never smokers in two prospective cohorts in Shanghai, China

BACKGROUND

The aetiology of lung cancer among individuals who never smoked remains elusive, despite 15% of lung cancer cases in men and 53% in women worldwide being unrelated to smoking. Epigenetic alterations, particularly DNA methylation (DNAm) changes, have emerged as potential drivers. Yet, few prospective epigenome-wide association studies (EWAS), primarily focusing on peripheral blood DNAm with limited representation of never smokers, have been conducted.

METHODS

We conducted a nested case-control study of 80 never-smoking incident lung cancer cases and 83 never-smoking controls within the Shanghai Women's Health Study and Shanghai Men's Health Study. DNAm was measured in prediagnostic oral rinse samples using Illumina MethylationEPIC array. Initially, we conducted an EWAS to identify differentially methylated positions (DMPs) associated with lung cancer in the discovery sample of 101 subjects. The top 50 DMPs were further evaluated in a replication sample of 62 subjects, and results were pooled using fixed-effect meta-analysis.

RESULTS

Our study identified three DMPs significantly associated with lung cancer at the epigenome-wide significance level of p<8.22×10-8. These DMPs were identified as cg09198866 (MYH9; TXN2), cg01411366 (SLC9A10) and cg12787323. Furthermore, examination of the top 1000 DMPs indicated significant enrichment in epithelial regulatory regions and their involvement in small GTPase-mediated signal transduction pathways. Additionally, GrimAge acceleration was identified as a risk factor for lung cancer (OR=1.19 per year; 95% CI 1.06 to 1.34).

CONCLUSIONS

While replication in a larger sample size is necessary, our findings suggest that DNAm patterns in prediagnostic oral rinse samples could provide novel insights into the underlying mechanisms of lung cancer in never smokers.

Transcriptome- and DNA methylation-based cell-type deconvolutions produce similar estimates of differential gene expression and differential methylation

BACKGROUND

Changing cell-type proportions can confound studies of differential gene expression or DNA methylation (DNAm) from peripheral blood mononuclear cells (PBMCs). We examined how cell-type proportions derived from the transcriptome versus the methylome (DNAm) influence estimates of differentially expressed genes (DEGs) and differentially methylated positions (DMPs).

METHODS

Transcriptome and DNAm data were obtained from PBMC RNA and DNA of Kenyan children (n = 8) before, during, and 6 weeks following uncomplicated malaria. DEGs and DMPs between time points were detected using cell-type adjusted modeling with Cibersortx or IDOL, respectively.

RESULTS

Most major cell types and principal components had moderate to high correlation between the two deconvolution methods (r = 0.60-0.96). Estimates of cell-type proportions and DEGs or DMPs were largely unaffected by the method, with the greatest discrepancy in the estimation of neutrophils.

CONCLUSION

Variation in cell-type proportions is captured similarly by both transcriptomic and methylome deconvolution methods for most major cell types.

Comparison of Infinium MethylationEPIC v2.0 to v1.0 for human population epigenetics: considerations for addressing EPIC version differences in DNA methylation-based tools

Background

The recently launched DNA methylation profiling platform, Illumina MethylationEPIC BeadChip Infinium microarray v2.0 (EPICv2), is highly correlated with measurements obtained from its predecessor MethylationEPIC BeadChip Infinium microarray v1.0 (EPICv1). However, the concordance between the two versions in the context of DNA methylation-based tools, including cell type deconvolution algorithms, epigenetic clocks, and inflammation and lifestyle biomarkers has not yet been investigated.

Findings

We profiled DNA methylation on both EPIC versions using matched venous blood samples from individuals spanning early to late adulthood across three cohorts. On combining the DNA methylomes of the cohorts, we observed that samples primarily clustered by the EPIC version they were measured on. Within each cohort, when we calculated cell type proportions, epigenetic age acceleration (EAA), rate of aging estimates, and biomarker scores for the matched samples on each version, we noted significant differences between EPICv1 and EPICv2 in the majority of these estimates. These differences were not significant, however, when estimates were adjusted for EPIC version or when EAAs were calculated separately for each EPIC version.

Conclusions

Our findings indicate that EPIC version differences predominantly explain DNA methylation variation and influence estimates of DNA methylation-based tools, and therefore we recommend caution when combining cohorts run on different versions. We demonstrate the importance of calculating DNA methylation-based estimates separately for each EPIC version or accounting for EPIC version either as a covariate in statistical models or by using version correction algorithms.

Consistent cord blood DNA methylation signatures of gestational age between South Asian and white European cohorts

BACKGROUND

Epigenetic modifications, particularly DNA methylation (DNAm) in cord blood, are an important biological marker of how external exposures during gestation can influence the in-utero environment and subsequent offspring development. Despite the recognized importance of DNAm during gestation, comparative studies to determine the consistency of these epigenetic signals across different ethnic groups are largely absent. To address this gap, we first performed epigenome-wide association studies (EWAS) of gestational age (GA) using newborn cord blood DNAm comparatively in a white European (n = 342) and a South Asian (n = 490) birth cohort living in Canada. Then, we capitalized on established cord blood epigenetic GA clocks to examine the associations between maternal exposures, offspring characteristics and epigenetic GA, as well as GA acceleration, defined as the residual difference between epigenetic and chronological GA at birth.

RESULTS

Individual EWASs confirmed 1,211 and 1,543 differentially methylated CpGs previously reported to be associated with GA, in white European and South Asian cohorts, respectively, with a similar distribution of effects. We confirmed that Bohlin's cord blood GA clock was robustly correlated with GA in white Europeans (r = 0.71; p = 6.0 × 10-54) and South Asians (r = 0.66; p = 6.9 × 10-64). In both cohorts, Bohlin's clock was positively associated with newborn weight and length and negatively associated with parity, newborn female sex, and gestational diabetes. Exclusive to South Asians, the GA clock was positively associated with the newborn ponderal index, while pre-pregnancy weight and gestational weight gain were strongly predictive of increased epigenetic GA in white Europeans. Important predictors of GA acceleration included gestational diabetes mellitus, newborn sex, and parity in both cohorts.

CONCLUSIONS

These results demonstrate the consistent DNAm signatures of GA and the utility of Bohlin's GA clock across the two populations. Although the overall pattern of DNAm is similar, its connections with the mother's environment and the baby's anthropometrics can differ between the two groups. Further research is needed to understand these unique relationships.

Transcriptome- and DNA methylation-based cell-type deconvolutions produce similar estimates of differential gene expression and differential methylation

Background

Changing cell-type proportions can confound studies of differential gene expression or DNA methylation (DNAm) from peripheral blood mononuclear cells (PBMCs). We examined how cell-type proportions derived from the transcriptome versus the methylome (DNAm) influence estimates of differentially expressed genes (DEGs) and differentially methylated positions (DMPs).

Methods

Transcriptome and DNAm data were obtained from PBMC RNA and DNA of Kenyan children (n = 8) before, during, and 6 weeks following uncomplicated malaria. DEGs and DMPs between time points were detected using cell-type adjusted modeling with Cibersortx or IDOL, respectively.

Results

Most major cell types and principal components had moderate to high correlation between the two deconvolution methods (r = 0.60-0.96). Estimates of cell-type proportions and DEGs or DMPs were largely unaffected by the method, with the greatest discrepancy in the estimation of neutrophils.

Conclusion

Variation in cell-type proportions is captured similarly by both transcriptomic and methylome deconvolution methods for most major cell types.

Analytical challenges in omics research on asthma and allergy: A National Institute of Allergy and Infectious Diseases workshop

Studies of asthma and allergy are generating increasing volumes of omics data for analysis and interpretation. The National Institute of Allergy and Infectious Diseases (NIAID) assembled a workshop comprising investigators studying asthma and allergic diseases using omics approaches, omics investigators from outside the field, and NIAID medical and scientific officers to discuss the following areas in asthma and allergy research: genomics, epigenomics, transcriptomics, microbiomics, metabolomics, proteomics, lipidomics, integrative omics, systems biology, and causal inference. Current states of the art, present challenges, novel and emerging strategies, and priorities for progress were presented and discussed for each area. This workshop report summarizes the major points and conclusions from this NIAID workshop. As a group, the investigators underscored the imperatives for rigorous analytic frameworks, integration of different omics data types, cross-disciplinary interaction, strategies for overcoming current limitations, and the overarching goal to improve scientific understanding and care of asthma and allergic diseases.

Epigenomic profiling of the infrapatellar fat pad in osteoarthritis

Osteoarthritis is a prevalent, complex disease of the joints, and affects multiple intra-articular tissues. Here, we have examined genome-wide DNA methylation profiles of primary infrapatellar fat pad and matched blood samples from 70 osteoarthritis patients undergoing total knee replacement surgery. Comparing the DNA methylation profiles between these tissues reveal widespread epigenetic differences. We produce the first genome-wide methylation quantitative trait locus (mQTL) map of fat pad, and make the resource available to the wider community. Using two-sample Mendelian randomization and colocalization analyses, we resolve osteoarthritis GWAS signals and provide insights into the molecular mechanisms underpinning disease aetiopathology. Our findings provide the first view of the epigenetic landscape of infrapatellar fat pad primary tissue in osteoarthritis.

Sex-dependent placental methylation quantitative trait loci provide insight into the prenatal origins of childhood onset traits and conditions

Molecular quantitative trait loci (QTLs) allow us to understand the biology captured in genome-wide association studies (GWASs). The placenta regulates fetal development and shows sex differences in DNA methylation. We therefore hypothesized that placental methylation QTL (mQTL) explain variation in genetic risk for childhood onset traits, and that effects differ by sex. We analyzed 411 term placentas from two studies and found 49,252 methylation (CpG) sites with mQTL and 2,489 CpG sites with sex-dependent mQTL. All mQTL were enriched in regions that typically affect gene expression in prenatal tissues. All mQTL were also enriched in GWAS results for growth- and immune-related traits, but male- and female-specific mQTL were more enriched than cross-sex mQTL. mQTL colocalized with trait loci at 777 CpG sites, with 216 (28%) specific to males or females. Overall, mQTL specific to male and female placenta capture otherwise overlooked variation in childhood traits.

Using saliva epigenetic data to develop and validate a multivariable predictor of esophageal cancer status

Background: Salivary epigenetic biomarkers may detect esophageal cancer. Methods: A total of 256 saliva samples from esophageal adenocarcinoma patients and matched volunteers were analyzed with Illumina EPIC methylation arrays. Three datasets were created, using 64% for discovery, 16% for testing and 20% for validation. Modules of gene-based methylation probes were created using weighted gene coexpression network analysis. Module significance to disease and gene importance to module were determined and a random forest classifier generated using best-scoring gene-related epigenetic probes. A cost-sensitive wrapper algorithm maximized cancer diagnosis. Results: Using age, sex and seven probes, esophageal adenocarcinoma was detected with area under the curve of 0.72 in discovery, 0.73 in testing and 0.75 in validation datasets. Cancer sensitivity was 88% with specificity of 31%. Conclusion: We have demonstrated a potentially clinically viable classifier of esophageal cancer based on saliva methylation.

Dynamic change in genome-wide methylation in response to increased suicidal ideation in schizophrenia spectrum disorders

Suicide is a significant public health crisis, with 800,000 people dying annually. Most people completing suicide have previous psychiatric conditions, and those with psychotic and mood disorders are particularly vulnerable. Unfortunately, there are currently no biomarkers available for accurately detecting suicidal ideation. Given the genetic and environmental factors that play a role in suicidal ideation, we attempted to determine epigenetic modifications, specifically DNA methylation, in response to changes in suicidal ideation. Using a longitudinal study design, 31 participants with schizophrenia spectrum disorders were interviewed at a baseline visit and again at a follow-up visit 3-12 months later. Current suicidal ideation was recorded at both visits with the Columbia Suicide Severity Rating Scale and the Beck Scale for Suicide Ideation, and whole blood was collected for methylation analysis. Our analysis shows a significant negative correlation between cg26910920 methylation and increasing Columbia Suicide Severity Rating Scale scores and a positive correlation between cg13673029 methylation and increasing Beck Scale for Suicide Ideation scores. This pilot study indicates that there is  the possibility  that DNA methylation can respond to changes in suicidal ideation over time and potentially be used as a biomarker of suicidal ideation in the future.

Sex differences in muscle protein expression and DNA methylation in response to exercise training

BACKGROUND

Exercise training elicits changes in muscle physiology, epigenomics, transcriptomics, and proteomics, with males and females exhibiting differing physiological responses to exercise training. However, the molecular mechanisms contributing to the differing adaptations between the sexes are poorly understood.

METHODS

We performed a meta-analysis for sex differences in skeletal muscle DNA methylation following an endurance training intervention (Gene SMART cohort and E-MTAB-11282 cohort). We investigated for sex differences in the skeletal muscle proteome following an endurance training intervention (Gene SMART cohort). Lastly, we investigated whether the methylome and proteome are associated with baseline cardiorespiratory fitness (maximal oxygen consumption; VO2max) in a sex-specific manner.

RESULTS

Here, we investigated for the first time, DNA methylome and proteome sex differences in response to exercise training in human skeletal muscle (n = 78; 50 males, 28 females). We identified 92 DNA methylation sites (CpGs) associated with exercise training; however, no CpGs changed in a sex-dependent manner. In contrast, we identified 189 proteins that are differentially expressed between the sexes following training, with 82 proteins differentially expressed between the sexes at baseline. Proteins showing the most robust sex-specific response to exercise include SIRT3, MRPL41, and MBP. Irrespective of sex, cardiorespiratory fitness was associated with robust methylome changes (19,257 CpGs) and no proteomic changes. We did not observe sex differences in the association between cardiorespiratory fitness and the DNA methylome. Integrative multi-omic analysis identified sex-specific mitochondrial metabolism pathways associated with exercise responses. Lastly, exercise training and cardiorespiratory fitness shifted the DNA methylomes to be more similar between the sexes.

CONCLUSIONS

We identified sex differences in protein expression changes, but not DNA methylation changes, following an endurance exercise training intervention; whereas we identified no sex differences in the DNA methylome or proteome response to lifelong training. Given the delicate interaction between sex and training as well as the limitations of the current study, more studies are required to elucidate whether there is a sex-specific training effect on the DNA methylome. We found that genes involved in mitochondrial metabolism pathways are differentially modulated between the sexes following endurance exercise training. These results shed light on sex differences in molecular adaptations to exercise training in skeletal muscle.

Exploring perinatal biopsychosocial factors and epigenetic age in 1-year-old offspring

Background: Little is known about the determinants of epigenetic aging in pediatric populations. Methods: Epigenetic age was estimated from 258 1-year-olds, using pediatric buccal epigenetic and Horvath clocks. We explored associations between epigenetic age and maternal indicators of mental and relational health, substance use and general physical health assessed during trimester three. Results: Higher anxiety and stress, BMI and higher parent-parent relationship quality were associated with pediatric buccal epigenetic clock differences. High blood pressure during pregnancy was associated with Horvath age acceleration. Third-trimester smoking and pre-pregnancy weight were associated with acceleration and deceleration respectively, and concordant across clocks. Conclusion: A broad range of maternal factors may shape epigenetic age in infancy; further research is needed to explore the possible effects on health and development.

Evaluation of the pooled sample method in Infinium MethylationEPIC BeadChip array by comparison with individual samples

BACKGROUND

The pooled sample method is used in epigenomic research and expression analysis and is a cost-effective screening approach for small amounts of DNA. Evaluation of the pooled sample method in epigenomic studies is performed using the Illumina Infinium Methylation 450K BeadChip array; however, subsequent reports on the updated 850K array are lacking. A previous study demonstrated that the methylation levels obtained from individual samples were accurately replicated using pooled samples but did not address epigenome-wide association study (EWAS) statistics. The DNA quantification method, which is important for the homogeneous mixing of DNA in the pooled sample method, has since become fluorescence-based, and additional factors need to be considered including the resolution of batch effects of microarray chips and the heterogeneity of the cellular proportions from which the DNA samples are derived. In this study, four pooled samples were created from 44 individual samples, and EWAS statistics for differentially methylated positions (DMPs) and regions (DMRs) were conducted for individual samples and compared with the statistics obtained from the pooled samples.

RESULTS

The methylation levels could be reproduced fairly well in the pooled samples. This was the case for the entire dataset and when limited to the top 100 CpG sites, consistent with a previous study using the 450K BeadChip array. However, the statistical results of the EWAS for the DMP by individual samples were not replicated in pooled samples. Qualitative analyses highlighting methylation within an arbitrary candidate gene were replicable. Focusing on chr 20, the statistical results of EWAS for DMR from individual samples showed replicability in the pooled samples as long as they were limited to regions with a sufficient effect size.

CONCLUSIONS

The pooled sample method replicated the methylation values well and can be used for EWAS in DMR. This method is sample amount-effective and cost-effective and can be utilized for screening by carefully understanding the effective features and disadvantages of the pooled sample method and combining it with candidate gene analyses.

Batch-effect correction with sample remeasurement in highly confounded case-control studies

Batch effects are pervasive in biomedical studies. One approach to address the batch effects is repeatedly measuring a subset of samples in each batch. These remeasured samples are used to estimate and correct the batch effects. However, rigorous statistical methods for batch-effect correction with remeasured samples are severely underdeveloped. Here we developed a framework for batch-effect correction using remeasured samples in highly confounded case-control studies. We provided theoretical analyses of the proposed procedure, evaluated its power characteristics and provided a power calculation tool to aid in the study design. We found that the number of samples that need to be remeasured depends strongly on the between-batch correlation. When the correlation is high, remeasuring a small subset of samples is possible to rescue most of the power.

Methodology to identify a gene expression signature by merging microarray datasets

A vast number of microarray datasets have been produced as a way to identify differentially expressed genes and gene expression signatures. A better understanding of these biological processes can help in the diagnosis and prognosis of diseases, as well as in the therapeutic response to drugs. However, most of the available datasets are composed of a reduced number of samples, leading to low statistical, predictive and generalization power. One way to overcome this problem is by merging several microarray datasets into a single dataset, which is typically a challenging task. Statistical methods or supervised machine learning algorithms are usually used to determine gene expression signatures. Nevertheless, statistical methods require an arbitrary threshold to be defined, and supervised machine learning methods can be ineffective when applied to high-dimensional datasets like microarrays. We propose a methodology to identify gene expression signatures by merging microarray datasets. This methodology uses statistical methods to obtain several sets of differentially expressed genes and uses supervised machine learning algorithms to select the gene expression signature. This methodology was validated using two distinct research applications: one using heart failure and the other using autism spectrum disorder microarray datasets. For the first, we obtained a gene expression signature composed of 117 genes, with a classification accuracy of approximately 98%. For the second use case, we obtained a gene expression signature composed of 79 genes, with a classification accuracy of approximately 82%. This methodology was implemented in R language and is available, under the MIT licence, at https://github.com/bioinformatics-ua/MicroGES.

Candidate biomarkers from the integration of methylation and gene expression in discordant autistic sibling pairs

While the genetics of autism spectrum disorders (ASD) has been intensively studied, resulting in the identification of over 100 putative risk genes, the epigenetics of ASD has received less attention, and results have been inconsistent across studies. We aimed to investigate the contribution of DNA methylation (DNAm) to the risk of ASD and identify candidate biomarkers arising from the interaction of epigenetic mechanisms with genotype, gene expression, and cellular proportions. We performed DNAm differential analysis using whole blood samples from 75 discordant sibling pairs of the Italian Autism Network collection and estimated their cellular composition. We studied the correlation between DNAm and gene expression accounting for the potential effects of different genotypes on DNAm. We showed that the proportion of NK cells was significantly reduced in ASD siblings suggesting an imbalance in their immune system. We identified differentially methylated regions (DMRs) involved in neurogenesis and synaptic organization. Among candidate loci for ASD, we detected a DMR mapping to CLEC11A (neighboring SHANK1) where DNAm and gene expression were significantly and negatively correlated, independently from genotype effects. As reported in previous studies, we confirmed the involvement of immune functions in the pathophysiology of ASD. Notwithstanding the complexity of the disorder, suitable biomarkers such as CLEC11A and its neighbor SHANK1 can be discovered using integrative analyses even with peripheral tissues.

Characterization of methylation profiles in spontaneous preterm birth placental villous tissue

Preterm birth is a global public health crisis which results in significant neonatal and maternal mortality. Yet little is known regarding the molecular mechanisms of idiopathic spontaneous preterm birth, and we have few diagnostic markers for adequate assessment of placental development and function. Previous studies of placental pathology and our transcriptomics studies suggest a role for placental maturity in idiopathic spontaneous preterm birth. It is known that placental DNA methylation changes over gestation. We hypothesized that if placental hypermaturity is present in our samples, we would observe a unique idiopathic spontaneous preterm birth DNA methylation profile potentially driving the gene expression differences we previously identified in our placental samples. Our results indicate the idiopathic spontaneous preterm birth DNA methylation pattern mimics the term birth methylation pattern suggesting hypermaturity. Only seven significant differentially methylated regions fitting the idiopathic spontaneous preterm birth specific (relative to the controls) profile were identified, indicating unusually high similarity in DNA methylation between idiopathic spontaneous preterm birth and term birth samples. We identified an additional 1,718 significantly methylated regions in our gestational age matched controls where the idiopathic spontaneous preterm birth DNA methylation pattern mimics the term birth methylation pattern, again indicating a striking level of similarity between the idiopathic spontaneous preterm birth and term birth samples. Pathway analysis of these regions revealed differences in genes within the WNT and Cadherin signaling pathways, both of which are essential in placental development and maturation. Taken together, these data demonstrate that the idiopathic spontaneous preterm birth samples display a hypermature methylation signature than expected given their respective gestational age which likely impacts birth timing.

Integrative Approaches of DNA Methylation Patterns According to Age, Sex and Longitudinal Changes

Background

In humans, age-related DNA methylation has been studied in blood, tissues, buccal swabs, and fibroblasts, and changes in DNA methylation patterns according to age and sex have been detected. To date, approximately 137,000 samples have been analyzed from 14,000 studies, and the information has been uploaded to the NCBI GEO database.

Methods

A correlation between age and methylation level and longitudinal changes in methylation levels was revealed in both sexes. Here, 20 public datasets derived from whole blood were analyzed using the Illumina BeadChip. Batch effects with respect to the time differences were correlated. The overall change in the pattern was provided as the inverse of the coefficient of variation (COV).

Results

Of the 20 datasets, nine were from a longitudinal study. All data had age and sex as common variables. Comprehensive details of age-, sex-, and longitudinal change-based DNA methylation levels in the whole blood sample were elucidated in this study. ELOVL2 and FHL2 showed the maximum correlation between age and DNA methylation. The methylation patterns of genes related to mental health differed according to age. Age-correlated genes have been associated with malformations (anteverted nostril, craniofacial abnormalities, and depressed nasal bridge) and drug addiction (drug habituation and smoking).

Conclusion

Based on 20 public DNA methylation datasets, methylation levels according to age and longitudinal changes by sex were identified and visualized using an integrated approach. The results highlight the molecular mechanisms underlying the association of sex and biological age with changes in DNA methylation, and the importance of optimal genomic information management.

Data processing choices can affect findings in differential methylation analyses: an investigation using data from the LIMIT RCT

Objective

A wide array of methods exist for processing and analysing DNA methylation data. We aimed to perform a systematic comparison of the behaviour of these methods, using cord blood DNAm from the LIMIT RCT, in relation to detecting hypothesised effects of interest (intervention and pre-pregnancy maternal BMI) as well as effects known to be spurious, and known to be present.

Methods

DNAm data, from 645 cord blood samples analysed using Illumina 450K BeadChip arrays, were normalised using three different methods (with probe filtering undertaken pre- or post- normalisation). Batch effects were handled with a supervised algorithm, an unsupervised algorithm, or adjustment in the analysis model. Analysis was undertaken with and without adjustment for estimated cell type proportions. The effects estimated included intervention and BMI (effects of interest in the original study), infant sex and randomly assigned groups. Data processing and analysis methods were compared in relation to number and identity of differentially methylated probes, rankings of probes by p value and log-fold-change, and distributions of p values and log-fold-change estimates.

Results

There were differences corresponding to each of the processing and analysis choices. Importantly, some combinations of data processing choices resulted in a substantial number of spurious 'significant' findings. We recommend greater emphasis on replication and greater use of sensitivity analyses.

Who's afraid of the X? Incorporating the X and Y chromosomes into the analysis of DNA methylation array data

BACKGROUND

Many human disease phenotypes manifest differently by sex, making the development of methods for incorporating X and Y-chromosome data into analyses vital. Unfortunately, X and Y chromosome data are frequently excluded from large-scale analyses of the human genome and epigenome due to analytical complexity associated with sex chromosome dosage differences between XX and XY individuals, and the impact of X-chromosome inactivation (XCI) on the epigenome. As such, little attention has been given to considering the methods by which sex chromosome data may be included in analyses of DNA methylation (DNAme) array data.

RESULTS

With Illumina Infinium HumanMethylation450 DNAme array data from 634 placental samples, we investigated the effects of probe filtering, normalization, and batch correction on DNAme data from the X and Y chromosomes. Processing steps were evaluated in both mixed-sex and sex-stratified subsets of the analysis cohort to identify whether including both sexes impacted processing results. We found that identification of probes that have a high detection p-value, or that are non-variable, should be performed in sex-stratified data subsets to avoid over- and under-estimation of the quantity of probes eligible for removal, respectively. All normalization techniques investigated returned X and Y DNAme data that were highly correlated with the raw data from the same samples. We found no difference in batch correction results after application to mixed-sex or sex-stratified cohorts. Additionally, we identify two analytical methods suitable for XY chromosome data, the choice between which should be guided by the research question of interest, and we performed a proof-of-concept analysis studying differential DNAme on the X and Y chromosome in the context of placental acute chorioamnionitis. Finally, we provide an annotation of probe types that may be desirable to filter in X and Y chromosome analyses, including probes in repetitive elements, the X-transposed region, and cancer-testis gene promoters.

CONCLUSION

While there may be no single "best" approach for analyzing DNAme array data from the X and Y chromosome, analysts must consider key factors during processing and analysis of sex chromosome data to accommodate the underlying biology of these chromosomes, and the technical limitations of DNA methylation arrays.

Normalization of Large-Scale Transcriptome Data Using Heuristic Methods

In this study, we introduce an artificial intelligent method for addressing the batch effect of a transcriptome data. The method has several clear advantages in comparison with the alternative methods presently in use. Batch effect refers to the discrepancy in gene expression data series, measured under different conditions. While the data from the same batch (measurements performed under the same conditions) are compatible, combining various batches into 1 data set is problematic because of incompatible measurements. Therefore, it is necessary to perform correction of the combined data (normalization), before performing biological analysis. There are numerous methods attempting to correct data set for batch effect. These methods rely on various assumptions regarding the distribution of the measurements. Forcing the data elements into pre-supposed distribution can severely distort biological signals, thus leading to incorrect results and conclusions. As the discrepancy between the assumptions regarding the data distribution and the actual distribution is wider, the biases introduced by such “correction methods” are greater. We introduce a heuristic method to reduce batch effect. The method does not rely on any assumptions regarding the distribution and the behavior of data elements. Hence, it does not introduce any new biases in the process of correcting the batch effect. It strictly maintains the integrity of measurements within the original batches.

Eleven quick tips for data cleaning and feature engineering

Applying computational statistics or machine learning methods to data is a key component of many scientific studies, in any field, but alone might not be sufficient to generate robust and reliable outcomes and results. Before applying any discovery method, preprocessing steps are necessary to prepare the data to the computational analysis. In this framework, data cleaning and feature engineering are key pillars of any scientific study involving data analysis and that should be adequately designed and performed since the first phases of the project. We call "feature" a variable describing a particular trait of a person or an observation, recorded usually as a column in a dataset. Even if pivotal, these data cleaning and feature engineering steps sometimes are done poorly or inefficiently, especially by beginners and unexperienced researchers. For this reason, we propose here our quick tips for data cleaning and feature engineering on how to carry out these important preprocessing steps correctly avoiding common mistakes and pitfalls. Although we designed these guidelines with bioinformatics and health informatics scenarios in mind, we believe they can more in general be applied to any scientific area. We therefore target these guidelines to any researcher or practitioners wanting to perform data cleaning or feature engineering. We believe our simple recommendations can help researchers and scholars perform better computational analyses that can lead, in turn, to more solid outcomes and more reliable discoveries.

Disease classification for whole-blood DNA methylation: Meta-analysis, missing values imputation, and XAI

BACKGROUND

DNA methylation has a significant effect on gene expression and can be associated with various diseases. Meta-analysis of available DNA methylation datasets requires development of a specific workflow for joint data processing.

RESULTS

We propose a comprehensive approach of combined DNA methylation datasets to classify controls and patients. The solution includes data harmonization, construction of machine learning classification models, dimensionality reduction of models, imputation of missing values, and explanation of model predictions by explainable artificial intelligence (XAI) algorithms. We show that harmonization can improve classification accuracy by up to 20% when preprocessing methods of the training and test datasets are different. The best accuracy results were obtained with tree ensembles, reaching above 95% for Parkinson's disease. Dimensionality reduction can substantially decrease the number of features, without detriment to the classification accuracy. The best imputation methods achieve almost the same classification accuracy for data with missing values as for the original data. XAI approaches have allowed us to explain model predictions from both populational and individual perspectives.

CONCLUSIONS

We propose a methodologically valid and comprehensive approach to the classification of healthy individuals and patients with various diseases based on whole-blood DNA methylation data using Parkinson's disease and schizophrenia as examples. The proposed algorithm works better for the former pathology, characterized by a complex set of symptoms. It allows to solve data harmonization problems for meta-analysis of many different datasets, impute missing values, and build classification models of small dimensionality.

Longitudinal change in blood DNA epigenetic signature after smoking cessation

Cigarette smoking is associated with epigenetic changes that may be reversible following smoking cessation. Whole blood DNA methylation was evaluated in Framingham Heart Study Offspring (n = 169) and Third Generation (n = 30) cohort participants at two study visits 6 years apart and in Atherosclerosis Risk in Communities (ARIC) study (n = 222) participants at two study visits 20 years apart. Changes in DNA methylation (delta β values) at 483,565 cytosine-phosphate-guanine (CpG) sites and differentially methylated regions (DMRs) were compared between participants who were current, former, or never smokers at both visits (current-current, former-former, never-never, respectively), versus those who quit in the interim (current-former). Interim quitters had more hypermethylation at four CpGs annotated to AHRR, one CpG annotated to F2RL3, and one intergenic CpG (cg21566642) compared with current-current smokers (FDR < 0.02 for all), and two significant DMRs were identified. While there were no significant differentially methylated CpGs in the comparison of interim quitters and former-former smokers, 106 DMRs overlapping with small nucleolar RNA were identified. As compared with all non-smokers, current-current smokers additionally had more hypermethylation at two CpG sites annotated to HIVEP3 and TMEM126A, respectively, and another intergenic CpG (cg14339116). Gene transcripts associated with smoking cessation were implicated in immune responses, cell homoeostasis, and apoptosis. Smoking cessation is associated with early reversion of blood DNA methylation changes at CpG sites annotated to AHRR and F2RL3 towards those of never smokers. Associated gene expression suggests a role of longitudinal smoking-related DNA methylation changes in immune response processes.

Genome wide exploration of the methylome in aggressive B-cell lymphoma in Golden Retrievers reveals a conserved hypermethylome

Few recurrent DNA mutations are seen in aggressive canine B cell lymphomas (cBCL), suggesting other frequent drivers. The methylated island recovery assay (MIRA-seq) or methylated CpG-binding domain sequencing (MBD-seq) was used to define the genome-wide methylation profiles in aggressive cBCL in Golden Retrievers to determine if cBCL can be better defined by epigenetic changes than by DNA mutations. DNA hypermethylation patterns were relatively homogenous within cBCL samples in Golden Retrievers, in different breeds and in geographical regions. Aberrant hypermethylation is thus suspected to be a central and early event in cBCL lymphomagenesis. Distinct subgroups within cBCL in Golden Retrievers were not identified with DNA methylation profiles. In comparison, the methylome profile of human DLBCL (hDLBCL) is relatively heterogeneous. Only moderate similarity between hDLBCL and cBCL was seen and cBCL likely cannot be accurately classified into the subtypes seen in hDLBCL. Genes with hypermethylated regions in the promoter-TSS-first exon of cBCL compared to normal B cells often also had additional hyper- and hypomethylated regions distributed throughout the gene suggesting non-randomized repeat targeting of key genes by epigenetic mechanisms. The prevalence of hypermethylation in transcription factor families in aggressive cBCL may represent a fundamental step in lymphomagenesis.

Tissue- and ethnicity-independent hypervariable DNA methylation states show evidence of establishment in the early human embryo

We analysed DNA methylation data from 30 datasets comprising 3474 individuals, 19 tissues and 8 ethnicities at CpGs covered by the Illumina450K array. We identified 4143 hypervariable CpGs ('hvCpGs') with methylation in the top 5% most variable sites across multiple tissues and ethnicities. hvCpG methylation was influenced but not determined by genetic variation, and was not linked to probe reliability, epigenetic drift, age, sex or cell heterogeneity effects. hvCpG methylation tended to covary across tissues derived from different germ-layers and hvCpGs were enriched for proximity to ERV1 and ERVK retrovirus elements. hvCpGs were also enriched for loci previously associated with periconceptional environment, parent-of-origin-specific methylation, and distinctive methylation signatures in monozygotic twins. Together, these properties position hvCpGs as strong candidates for studying how stochastic and/or environmentally influenced DNA methylation states which are established in the early embryo and maintained stably thereafter can influence life-long health and disease.

HarmonizR enables data harmonization across independent proteomic datasets with appropriate handling of missing values

Dataset integration is common practice to overcome limitations in statistically underpowered omics datasets. Proteome datasets display high technical variability and frequent missing values. Sophisticated strategies for batch effect reduction are lacking or rely on error-prone data imputation. Here we introduce HarmonizR, a data harmonization tool with appropriate missing value handling. The method exploits the structure of available data and matrix dissection for minimal data loss, without data imputation. This strategy implements two common batch effect reduction methods—ComBat and limma (removeBatchEffect()). The HarmonizR strategy, evaluated on four exemplarily analyzed datasets with up to 23 batches, demonstrated successful data harmonization for different tissue preservation techniques, LC-MS/MS instrumentation setups, and quantification approaches. Compared to data imputation methods, HarmonizR was more efficient and performed superior regarding the detection of significant proteins. HarmonizR is an efficient tool for missing data tolerant experimental variance reduction and is easily adjustable for individual dataset properties and user preferences.

Dataset integration is common practice to overcome limitations in statistically underpowered omics datasets. Here the authors present “HarmonizR”, a tool for missing data tolerant experimental variance reduction in large, integrated but independently generated datasets without data imputation, adjustable for individual dataset modalities, correction algorithm, and user preferences.

Batch-effect detection, correction and characterisation in Illumina HumanMethylation450 and MethylationEPIC BeadChip array data

Background

Genomic technologies can be subject to significant batch-effects which are known to reduce experimental power and to potentially create false positive results. The Illumina Infinium Methylation BeadChip is a popular technology choice for epigenome-wide association studies (EWAS), but presently, little is known about the nature of batch-effects on these designs. Given the subtlety of biological phenotypes in many EWAS, control for batch-effects should be a consideration.

Results

Using the batch-effect removal approaches in the ComBat and Harman software, we examined two in-house datasets and compared results with three large publicly available datasets, (1214 HumanMethylation450 and 1094 MethylationEPIC BeadChips in total), and find that despite various forms of preprocessing, some batch-effects persist. This residual batch-effect is associated with the day of processing, the individual glass slide and the position of the array on the slide. Consistently across all datasets, 4649 probes required high amounts of correction. To understand the impact of this set to EWAS studies, we explored the literature and found three instances where persistently batch-effect prone probes have been reported in abstracts as key sites of differential methylation. As well as batch-effect susceptible probes, we also discover a set of probes which are erroneously corrected. We provide batch-effect workflows for Infinium Methylation data and provide reference matrices of batch-effect prone and erroneously corrected features across the five datasets spanning regionally diverse populations and three commonly collected biosamples (blood, buccal and saliva).

Conclusions

Batch-effects are ever present, even in high-quality data, and a strategy to deal with them should be part of experimental design, particularly for EWAS. Batch-effect removal tools are useful to reduce technical variance in Infinium Methylation data, but they need to be applied with care and make use of post hoc diagnostic measures.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13148-022-01277-9.

Proteomics of Human Biological Fluids for Biomarker Discoveries: Technical Advances and Recent Applications

INTRODUCTION

Biological fluids are routine samples for diagnostic testing and monitoring. Blood samples are typically measured because of their moderate collection invasiveness and high information content on health and disease. Several body fluids, such as cerebrospinal fluid (CSF), are also studied and suited to specific pathologies. Over the last two decades proteomics has quested to identify protein biomarkers but with limited success. Recent technologies and refined pipelines have accelerated the profiling of human biological fluids.

AREAS COVERED

We review proteomic technologies for the identification of biomarkers. Those are based on antibodies/aptamers arrays or mass spectrometry (MS), but new ones are emerging. Advances in scalability and throughput have allowed to better design studies and cope with the limited sample size that had until now prevailed due to technological constraints. With these enablers, plasma/serum, CSF, saliva, tears, urine, and milk proteomes have been further profiled; we provide a non-exhaustive picture of some recent highlights (mainly covering literature from last five years in the Scopus database) using MS-based proteomics.

EXPERT OPINION

While proteomics has been in the shadow of genomics for years, proteomic tools and methodologies have reached a certain maturity. They are better suited to discover innovative and robust biofluid biomarkers.

Accelerated biological aging in COVID-19 patients

Chronological age is a risk factor for SARS-CoV-2 infection and severe COVID-19. Previous findings indicate that epigenetic age could be altered in viral infection. However, the epigenetic aging in COVID-19 has not been well studied. In this study, DNA methylation of the blood samples from 232 healthy individuals and 413 COVID-19 patients is profiled using EPIC methylation array. Epigenetic ages of each individual are determined by applying epigenetic clocks and telomere length estimator to the methylation profile of the individual. Epigenetic age acceleration is calculated and compared between groups. We observe strong correlations between the epigenetic clocks and individual's chronological age (r > 0.8, p < 0.0001). We also find the increasing acceleration of epigenetic aging and telomere attrition in the sequential blood samples from healthy individuals and infected patients developing non-severe and severe COVID-19. In addition, the longitudinal DNA methylation profiling analysis find that the accumulation of epigenetic aging from COVID-19 syndrome could be partly reversed at late clinic phases in some patients. In conclusion, accelerated epigenetic aging is associated with the risk of SARS-CoV-2 infection and developing severe COVID-19. In addition, the accumulation of epigenetic aging from COVID-19 may contribute to the post-COVID-19 syndrome among survivors.

A Final Frontier in Environment-Genome Interactions? Integrated, Multi-Omic Approaches to Predictions of Non-Communicable Disease Risk

Epidemiological and associative research from humans and animals identifies correlations between the environment and health impacts. The environment—health inter-relationship is effected through an individual’s underlying genetic variation and mediated by mechanisms that include the changes to gene regulation that are associated with the diversity of phenotypes we exhibit. However, the causal relationships have yet to be established, in part because the associations are reduced to individual interactions and the combinatorial effects are rarely studied. This problem is exacerbated by the fact that our genomes are highly dynamic; they integrate information across multiple levels (from linear sequence, to structural organisation, to temporal variation) each of which is open to and responds to environmental influence. To unravel the complexities of the genomic basis of human disease, and in particular non-communicable diseases that are also influenced by the environment (e.g., obesity, type II diabetes, cancer, multiple sclerosis, some neurodegenerative diseases, inflammatory bowel disease, rheumatoid arthritis) it is imperative that we fully integrate multiple layers of genomic data. Here we review current progress in integrated genomic data analysis, and discuss cases where data integration would lead to significant advances in our ability to predict how the environment may impact on our health. We also outline limitations which should form the basis of future research questions. In so doing, this review will lay the foundations for future research into the impact of the environment on our health.

Evaluation of critical data processing steps for reliable prediction of gene co-expression from large collections of RNA-seq data

Motivation

Gene co-expression analysis is an attractive tool for leveraging enormous amounts of public RNA-seq datasets for the prediction of gene functions and regulatory mechanisms. However, the optimal data processing steps for the accurate prediction of gene co-expression from such large datasets remain unclear. Especially the importance of batch effect correction is understudied.

Results

We processed RNA-seq data of 68 human and 76 mouse cell types and tissues using 50 different workflows into 7,200 genome-wide gene co-expression networks. We then conducted a systematic analysis of the factors that result in high-quality co-expression predictions, focusing on normalization, batch effect correction, and measure of correlation. We confirmed the key importance of high sample counts for high-quality predictions. However, choosing a suitable normalization approach and applying batch effect correction can further improve the quality of co-expression estimates, equivalent to a >80% and >40% increase in samples. In larger datasets, batch effect removal was equivalent to a more than doubling of the sample size. Finally, Pearson correlation appears more suitable than Spearman correlation, except for smaller datasets.

Conclusion

A key point for accurate prediction of gene co-expression is the collection of many samples. However, paying attention to data normalization, batch effects, and the measure of correlation can significantly improve the quality of co-expression estimates.

Association of Epigenetic Differences Screened in a Few Cases of Monozygotic Twins Discordant for Attention-Deficit Hyperactivity Disorder With Brain Structures

The present study examined the relationship between DNA methylation differences and variations in brain structures involved in the development of attention-deficit hyperactivity disorder (ADHD). First, we used monozygotic (MZ) twins discordant (2 pairs of 4 individuals, 2 boys, mean age 12.5 years) for ADHD to identify candidate DNA methylation sites involved in the development of ADHD. Next, we tried to replicate these candidates in a case-control study (ADHD: N = 18, 15 boys, mean age 10.0 years; Controls: N = 62, 40 boys, mean age 13.9 years). Finally, we examined how methylation rates at those sites relate to the degree of local structural alterations where significant differences were observed between cases and controls. As a result, we identified 61 candidate DNA methylation sites involved in ADHD development in two pairs of discordant MZ twins, among which elevated methylation at a site in the sortilin-related Vps10p domain containing receptor 2 (SorCS2) gene was replicated in the case-control study. We also observed that the ADHD group had significantly reduced gray matter volume (GMV) in the precentral and posterior orbital gyri compared to the control group and that this volume reduction was positively associated with SorCS2 methylation. Furthermore, the reduced GMV regions in children with ADHD are involved in language processing and emotional control, while SorCS2 methylation is also negatively associated with emotional behavioral problems in children. These results indicate that SorCS2 methylation might mediate a reduced GMV in the precentral and posterior orbital gyri and therefore influence the pathology of children with ADHD.

Batch effect reduction of microarray data with dependent samples using an empirical Bayes approach (BRIDGE)

Batch-effects present challenges in the analysis of high-throughput molecular data and are particularly problematic in longitudinal studies when interest lies in identifying genes/features whose expression changes over time, but time is confounded with batch. While many methods to correct for batch-effects exist, most assume independence across samples; an assumption that is unlikely to hold in longitudinal microarray studies. We propose Batch effect Reduction of mIcroarray data with Dependent samples usinGEmpirical Bayes (BRIDGE), a three-step parametric empirical Bayes approach that leverages technical replicate samples profiled at multiple timepoints/batches, so-called "bridge samples", to inform batch-effect reduction/attenuation in longitudinal microarray studies. Extensive simulation studies and an analysis of a real biological data set were conducted to benchmark the performance of BRIDGE against both ComBat and longitudinalComBat. Our results demonstrate that while all methods perform well in facilitating accurate estimates of time effects, BRIDGE outperforms both ComBat and longitudinal ComBat in the removal of batch-effects in data sets with bridging samples, and perhaps as a result, was observed to have improved statistical power for detecting genes with a time effect. BRIDGE demonstrated competitive performance in batch effect reduction of confounded longitudinal microarray studies, both in simulated and a real data sets, and may serve as a useful preprocessing method for researchers conducting longitudinal microarray studies that include bridging samples.

Epigenome-wide association studies: current knowledge, strategies and recommendations

The aetiology and pathophysiology of complex diseases are driven by the interaction between genetic and environmental factors. The variability in risk and outcomes in these diseases are incompletely explained by genetics or environmental risk factors individually. Therefore, researchers are now exploring the epigenome, a biological interface at which genetics and the environment can interact. There is a growing body of evidence supporting the role of epigenetic mechanisms in complex disease pathophysiology. Epigenome-wide association studies (EWASes) investigate the association between a phenotype and epigenetic variants, most commonly DNA methylation. The decreasing cost of measuring epigenome-wide methylation and the increasing accessibility of bioinformatic pipelines have contributed to the rise in EWASes published in recent years. Here, we review the current literature on these EWASes and provide further recommendations and strategies for successfully conducting them. We have constrained our review to studies using methylation data as this is the most studied epigenetic mechanism; microarray-based data as whole-genome bisulphite sequencing remains prohibitively expensive for most laboratories; and blood-based studies due to the non-invasiveness of peripheral blood collection and availability of archived DNA, as well as the accessibility of publicly available blood-cell-based methylation data. Further, we address multiple novel areas of EWAS analysis that have not been covered in previous reviews: (1) longitudinal study designs, (2) the chip analysis methylation pipeline (ChAMP), (3) differentially methylated region (DMR) identification paradigms, (4) methylation quantitative trait loci (methQTL) analysis, (5) methylation age analysis and (6) identifying cell-specific differential methylation from mixed cell data using statistical deconvolution.

DNA methylation-based signature of CD8+ tumor-infiltrating lymphocytes enables evaluation of immune response and prognosis in colorectal cancer

Background

Tumor-infiltrating lymphocytes (TILs), especially CD8+ TILs, can be used for predicting immunotherapy responsiveness and survival outcome. However, the evaluation of CD8+ TILs currently relies on histopathological methodology with high variability. We therefore aimed to develop a DNA methylation signature for CD8+ TILs (CD8+ MeTIL) that could evaluate immune response and prognosis in colorectal cancer (CRC).

Methods

A CD8+ MeTIL signature score was constructed by using CD8+ T cell-specific differentially methylated positions (DMPs) that were identified from Illumina EPIC methylation arrays. Immune cells, colon epithelial cells, and two CRC cohorts (n=282 and 335) were used to develop a PCR-based assay for quantitative analysis of DNA methylation at single-base resolution (QASM) to determine CD8 + MeTIL signature score.

Results

Three CD8+ T cell-specific DMPs were identified to construct the CD8+ MeTIL signature score, which showed a dramatic discriminability between CD8+ T cells and other cells. The QASM assay we developed for CD8+ MeTIL markers could measure CD8+ TILs distributions in a fully quantitative, accurate, and simple manner. The CD8+ MeTIL score determined by QASM assay showed a strong association with histopathology-based CD8+ TIL counts and a gene expression-based immune marker. Furthermore, the low CD8+ MeTIL score (enriched CD8+ TILs) was associated with MSI-H tumors and predicted better survival in CRC cohorts.

Conclusions

This study developed a quantitative DNA methylation-based signature that was reliable to evaluate CD8+ TILs and prognosis in CRC. This approach has the potential to be a tool for investigations on CD8+ TILs and a biomarker for therapeutic approaches, including immunotherapy.

An effective processing pipeline for harmonizing DNA methylation data from Illumina's 450K and EPIC platforms for epidemiological studies

OBJECTIVE

Illumina BeadChip arrays are commonly used to generate DNA methylation data for large epidemiological studies. Updates in technology over time create challenges for data harmonization within and between studies, many of which obtained data from the older 450K and newer EPIC platforms. The pre-processing pipeline for DNA methylation is not trivial, and influences the downstream analyses. Incorporating different platforms adds a new level of technical variability that has not yet been taken into account by recommended pipelines. Our study evaluated the performance of various tools on different versions of platform data harmonization at each step of pre-processing pipeline, including quality control (QC), normalization, batch effect adjustment, and genomic inflation. We illustrate our novel approach using 450K and EPIC data from the Diabetes Autoimmunity Study in the Young (DAISY) prospective cohort.

RESULTS

We found normalization and probe filtering had the biggest effect on data harmonization. Employing a meta-analysis was an effective and easily executable method for accounting for platform variability. Correcting for genomic inflation also helped with harmonization. We present guidelines for studies seeking to harmonize data from the 450K and EPIC platforms, which includes the use of technical replicates for evaluating numerous pre-processing steps, and employing a meta-analysis.

REBET: a method to determine the number of cell clusters based on batch effect removal

In single-cell RNA-seq (scRNA-seq) data analysis, a fundamental problem is to determine the number of cell clusters based on the gene expression profiles. However, the performance of current methods is still far from satisfactory, presumably due to their limitations in capturing the expression variability among cell clusters. Batch effects represent the undesired variability between data measured in different batches. When data are obtained from different labs or protocols batch effects occur. Motivated by the practice of batch effect removal, we considered cell clusters as batches. We hypothesized that the number of cell clusters (i.e. batches) could be correctly determined if the variances among clusters (i.e. batch effects) were removed. We developed a new method, namely, removal of batch effect and testing (REBET), for determining the number of cell clusters. In this method, cells are first partitioned into k clusters. Second, the batch effects among these k clusters are then removed. Third, the quality of batch effect removal is evaluated with the average range of normalized mutual information (ARNMI), which measures how uniformly the cells with batch-effects-removal are mixed. By testing a range of k values, the k value that corresponds to the lowest ARNMI is determined to be the optimal number of clusters. We compared REBET with state-of-the-art methods on 32 simulated datasets and 14 published scRNA-seq datasets. The results show that REBET can accurately and robustly estimate the number of cell clusters and outperform existing methods. Contact: H.D.L. (hongdong@csu.edu.cn) or Q.S.X. (qsxu@csu.edu.cn).

A cross-cohort analysis of autosomal DNA methylation sex differences in the term placenta

BACKGROUND

Human placental DNA methylation (DNAme) data is a valuable resource for studying sex differences during gestation, as DNAme profiles after delivery reflect the cumulative effects of gene expression patterns and exposures across gestation. Here, we present an analysis of sex differences in autosomal DNAme in the uncomplicated term placenta (n = 343) using the Illumina 450K array.

RESULTS

At a false discovery rate < 0.05 and a mean sex difference in DNAme beta value of > 0.10, we identified 162 autosomal CpG sites that were differentially methylated by sex and replicated in an independent cohort of samples (n = 293). Several of these differentially methylated CpG sites were part of larger correlated regions of sex differential DNAme. Although global DNAme levels did not differ by sex, the majority of significantly differentially methylated CpGs were more highly methylated in male placentae, the opposite of what is seen in differential methylation analyses of somatic tissues. Patterns of autosomal DNAme at these 162 CpGs were significantly associated with maternal age (in males) and newborn birthweight standard deviation (in females).

CONCLUSIONS

Our results provide a comprehensive analysis of sex differences in autosomal DNAme in the term human placenta. We report a list of high-confidence autosomal sex-associated differentially methylated CpGs and identify several key features of these loci that suggest their relevance to sex differences observed in normative and complicated pregnancies.

Social epigenomics: are we at an impasse?

Aim: Social scientists have placed particularly high expectations on the study of epigenomics to explain how exposure to adverse social factors like poverty, child maltreatment and racism - particularly early in childhood - might contribute to complex diseases. However, progress has stalled, reflecting many of the same challenges faced in genomics, including overhype, lack of diversity in samples, limited replication and difficulty interpreting significance of findings. Materials & methods: This review focuses on the future of social epigenomics by discussing progress made, ongoing methodological and analytical challenges and suggestions for improvement. Results & conclusion: Recommendations include more diverse sample types, cross-cultural, longitudinal and multi-generational studies. True integration of social and epigenomic data will require increased access to both data types in publicly available databases, enhanced data integration frameworks, and more collaborative efforts between social scientists and geneticists.

Random rotation for identifying differentially expressed genes with linear models following batch effect correction

MOTIVATION

Data generated from high-throughput technologies such as sequencing, microarray and bead-chip technologies are unavoidably affected by batch effects. Large effort has been put into developing methods for correcting these effects. Often, batch effect correction and hypothesis testing cannot be done with one single model, but are done successively with separate models in data analysis pipelines. This potentially leads to biased p-values or false discovery rates due to the influence of batch effect correction on the data.

RESULTS

We present a novel approach for estimating null distributions of test statistics in data analysis pipelines where batch effect correction is followed by linear model analysis. The approach is based on generating simulated datasets by random rotation and thereby retains the dependence structure of genes adequately. This allows estimating null distributions of dependent test statistics and thus the calculation of resampling based p-values and false discovery rates following batch effect correction while maintaining the alpha level.

AVAILABILITY

The described methods are implemented as randRotation package on Bioconductor: https://bioconductor.org/packages/randRotation/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Transcriptome Profiling Analyses in Psoriasis: A Dynamic Contribution of Keratinocytes to the Pathogenesis

Psoriasis is an immune-mediated inflammatory skin disease with a complex etiology involving environmental and genetic factors. A better insight into related genomic alteration helps design precise therapies leading to better treatment outcome. Gene expression in psoriasis can provide relevant information about the altered expression of mRNA transcripts, thus giving new insights into the disease onset. Techniques for transcriptome analyses, such as microarray and RNA sequencing (RNA-seq), are relevant tools for the discovery of new biomarkers as well as new therapeutic targets. This review summarizes the findings related to the contribution of keratinocytes in the pathogenesis of psoriasis by an in-depth review of studies that have examined psoriatic transcriptomes in the past years. It also provides valuable information on reconstructed 3D psoriatic skin models using cells isolated from psoriatic patients for transcriptomic studies.

Microarray Normalization Revisited for Reproducible Breast Cancer Biomarkers

Precision medicine for breast cancer relies on biomarkers to select therapies. However, the reliability of biomarkers drawn from gene expression arrays has been questioned and calls for reassessment, in particular for large datasets. We revisit widely used data-normalization procedures and evaluate differences in outcome in order to pinpoint the most reliable reprocessing methods biomarkers can be based upon. We generated a database of 3753 breast cancer patients out of 38 studies by downloading and curating patient samples from NCBI-GEO. As gene-expression biomarkers, we select the assessment of receptor status and breast cancer subtype classification. Each normalization procedure is applied separately, and biomarkers are then evaluated for each patient. Differences between normalization pipelines are quantified as percentages of patients having outcomes different for each pipeline. Some normalization procedures lead to quite consistent biomarkers, differing only in 1-2% of patients. Other normalization procedures—some of them have been used in many clinical studies—end up with distrusting discrepancies (10% and more). A good deal of doubt regarding the reliability of microarrays may root in the haphazard application of inadequate preprocessing pipelines. Several modes of batch corrections are evaluated regarding a possible improvement of receptor prediction from gene expression versus the golden standard of immunohistochemistry. Finally, we nominate those normalization methods yielding consistent and trustable results. Adequate bioinformatics data preprocessing is key and crucial for any subsequent statistics to arrive at trustable results. We conclude with a suggestion for future bioinformatics development to further increase the reliability of cancer biomarkers.

Simulating ComBat: how batch correction can lead to the systematic introduction of false positive results in DNA methylation microarray studies

Background

Systematic technical effects—also called batch effects—are a considerable challenge when analyzing DNA methylation (DNAm) microarray data, because they can lead to false results when confounded with the variable of interest. Methods to correct these batch effects are error-prone, as previous findings have shown.

Results

Here, we demonstrate how using the R function ComBat to correct simulated Infinium HumanMethylation450 BeadChip (450 K) and Infinium MethylationEPIC BeadChip Kit (EPIC) DNAm data can lead to a large number of false positive results under certain conditions. We further provide a detailed assessment of the consequences for the highly relevant problem of p-value inflation with subsequent false positive findings after application of the frequently used ComBat method. Using ComBat to correct for batch effects in randomly generated samples produced alarming numbers of false discovery rate (FDR) and Bonferroni-corrected (BF) false positive results in unbalanced as well as in balanced sample distributions in terms of the relation between the outcome of interest variable and the technical position of the sample during the probe measurement. Both sample size and number of batch factors (e.g. number of chips) were systematically simulated to assess the probability of false positive findings. The effect of sample size was simulated using n = 48 up to n = 768 randomly generated samples. Increasing the number of corrected factors led to an exponential increase in the number of false positive signals. Increasing the number of samples reduced, but did not completely prevent, this effect.

Conclusions

Using the approach described, we demonstrate, that using ComBat for batch correction in DNAm data can lead to false positive results under certain conditions and sample distributions. Our results are thus contrary to previous publications, considering a balanced sample distribution as unproblematic when using ComBat. We do not claim completeness in terms of reporting all technical conditions and possible solutions of the occurring problems as we approach the problem from a clinician’s perspective and not from that of a computer scientist. With our approach of simulating data, we provide readers with a simple method to assess the probability of false positive findings in DNAm microarray data analysis pipelines.

Gene Expression: the Key to Understanding HIV-1 Infection?

Gene expression profiling of the host response to HIV infection has promised to fill the gaps in our knowledge and provide new insights toward vaccine and cure. However, despite 20 years of research, the biggest questions remained unanswered. A literature review identified 62 studies examining gene expression dysregulation in samples from individuals living with HIV. Changes in gene expression were dependent on cell/tissue type, stage of infection, viremia, and treatment status. Some cell types, notably CD4+ T cells, exhibit upregulation of cell cycle, interferon-related, and apoptosis genes consistent with depletion. Others, including CD8+ T cells and natural killer cells, exhibit perturbed function in the absence of direct infection with HIV. Dysregulation is greatest during acute infection. Differences in study design and data reporting limit comparability of existing research and do not as yet provide a coherent overview of gene expression in HIV. This review outlines the extraordinarily complex host response to HIV and offers recommendations to realize the full potential of HIV host transcriptomics.

Genome-wide analysis of constitutional DNA methylation in familial melanoma

Background

Heritable epigenetic alterations have been proposed as an explanation for familial clustering of melanoma. Here we performed genome-wide DNA methylation analysis on affected family members not carrying pathogenic variants in established melanoma susceptibility genes, compared with healthy volunteers.

Results

All melanoma susceptibility genes showed the absence of epimutations in familial melanoma patients, and no loss of imprinting was detected. Unbiased genome-wide DNA methylation analysis revealed significantly different levels of methylation in single CpG sites. The methylation level differences were small and did not affect reported tumour predisposition genes.

Conclusion

Our results provide no support for heritable epimutations as a cause of familial melanoma.

Mendelian Randomization Identifies CpG Methylation Sites With Mediation Effects for Genetic Influences on BMD in Peripheral Blood Monocytes

Osteoporosis is mainly characterized by low bone mineral density (BMD) and is an increasingly serious public health concern. DNA methylation is a major epigenetic mechanism that may contribute to the variation in BMD and may mediate the effects of genetic and environmental factors of osteoporosis. In this study, we performed an epigenome-wide DNA methylation analysis in peripheral blood monocytes of 118 Caucasian women with extreme BMD values. Further, we developed and implemented a novel analytical framework that integrates Mendelian randomization with genetic fine mapping and colocalization to evaluate the causal relationships between DNA methylation and BMD phenotype. We identified 2,188 differentially methylated CpGs (DMCs) between the low and high BMD groups and distinguished 30 DMCs that may mediate the genetic effects on BMD. The causal relationship was further confirmed by eliminating the possibility of horizontal pleiotropy, linkage effect and reverse causality. The fine-mapping analysis determined 25 causal variants that are most likely to affect the methylation levels at these mediator DMCs. The majority of the causal methylation quantitative loci and DMCs reside within cell type-specific histone mark peaks, enhancers, promoters, promoter flanking regions and CTCF binding sites, supporting the regulatory potentials of these loci. The established causal pathways from genetic variant to BMD phenotype mediated by DNA methylation provide a gene list to aid in designing future functional studies and lead to a better understanding of the genetic and epigenetic mechanisms underlying the variation of BMD.

Machine learning workflows to estimate class probabilities for precision cancer diagnostics on DNA methylation microarray data

DNA methylation data-based precision cancer diagnostics is emerging as the state of the art for molecular tumor classification. Standards for choosing statistical methods with regard to well-calibrated probability estimates for these typically highly multiclass classification tasks are still lacking. To support this choice, we evaluated well-established machine learning (ML) classifiers including random forests (RFs), elastic net (ELNET), support vector machines (SVMs) and boosted trees in combination with post-processing algorithms and developed ML workflows that allow for unbiased class probability (CP) estimation. Calibrators included ridge-penalized multinomial logistic regression (MR) and Platt scaling by fitting logistic regression (LR) and Firth's penalized LR. We compared these workflows on a recently published brain tumor 450k DNA methylation cohort of 2,801 samples with 91 diagnostic categories using a 5 × 5-fold nested cross-validation scheme and demonstrated their generalizability on external data from The Cancer Genome Atlas. ELNET was the top stand-alone classifier with the best calibration profiles. The best overall two-stage workflow was MR-calibrated SVM with linear kernels closely followed by ridge-calibrated tuned RF. For calibration, MR was the most effective regardless of the primary classifier. The protocols developed as a result of these comparisons provide valuable guidance on choosing ML workflows and their tuning to generate well-calibrated CP estimates for precision diagnostics using DNA methylation data. Computation times vary depending on the ML algorithm from <15 min to 5 d using multi-core desktop PCs. Detailed scripts in the open-source R language are freely available on GitHub, targeting users with intermediate experience in bioinformatics and statistics and using R with Bioconductor extensions.