CpG Sites Research Articles

CmirC update 2024: a multi-omics database for clustered miRNAs

Clustered miRNAs consist of two or more miRNAs transcribed together and may coordinately regulate gene expression. Differential expression of clustered miRNAs is found to be controlled by crosstalk of genetic or epigenetic mechanisms. It has been demonstrated that clustered miRNA expression patterns greatly impact cancer cell progression. With the CmirC initiative, we initially developed a comprehensive database to identify copy number variation (CNV) driven clustered miRNAs in cancer. Now, we extended the analysis and identified three miRNAs, mir-96, mir-183, and mir-21, were found to be significantly upregulated in 17 cancer types. Further, CmirC is now upgraded to determine the impact of changes in the DNA methylation status at clustered miRNAs by utilizing The Cancer Genomic Atlas (TCGA) cancer datasets. We examined specific methylation datasets from 9,639 samples, pinpointing 215,435 methylation sites and 27,949 CpG islands with miRNA cluster information. The integrated analysis identified 34 clusters exhibiting differentially methylated CpG sites across 14 cancer types. Furthermore, we determined that CpG islands in the promoter region of 20 miRNA clusters could play a regulatory role. Along with ensuring a straightforward and convenient user experience, CmirC has been updated with improved data browsing and analysis functionalities, as well as enabled hyperlinks to literature and miR-cancer databases. The enhanced version of CmirC is anticipated to play an important role in providing information on the regulation of clustered miRNA expression, and their targeted oncogenes and tumor suppressors. The newly updated version of CmirC is available at https://slsdb.manipal.edu/cmirclust/.

Comparing methylation levels assayed in GC-rich regions with current and emerging methods

DNA methylation is an epigenetic mechanism that regulates gene expression, and for mammals typically occurs on cytosines within CpG dinucleotides. A significant challenge for methylation detection methods is accurately measuring methylation levels within GC-rich regions such as gene promoters, as inaccuracies compromise downstream biological interpretation of the data. To address this challenge, we compared methylation levels assayed using four different Methods Enzymatic Methyl-seq (EM-seq), whole genome bisulphite sequencing (WGBS), Infinium arrays (Illumina MethylationEPIC, “EPIC”), and Oxford Nanopore Technologies nanopore sequencing (ONT) applied to human DNA. Overall, all methods produced comparable and consistent methylation readouts across the human genome. The flexibility offered by current gold standard WGBS in interrogating genome-wide cytosines is surpassed technically by both EM-seq and ONT, as their coverages and methylation readouts are less prone to GC bias. These advantages are tempered by increased laboratory time (EM-seq) and higher complexity (ONT). We further assess the strengths and weaknesses of each method, and provide recommendations in choosing the most appropriate methylation method for specific scientific questions or translational needs.

Quantitative trait loci concentrate in specific regions of the Mexican cavefish genome and reveal key candidate genes for cave-associated evolution.

A major goal of modern biology is connecting phenotype with its underlying genetic basis. The Mexican cavefish (Astyanax mexicanus), a characin fish species comprised of a surface ecotype and a cave-derived ecotype, is well suited as a model to study the genetic mechanisms underlying adaptation to extreme environments. Here we map 206 previously published quantitative trait loci (QTL) for cave-derived traits in A. mexicanus to the newest version of the surface fish genome assembly, AstMex3. These analyses revealed that QTL cluster in the genome more than expected by chance, and this clustering is not explained by the distribution of genes in the genome. To investigate whether certain characteristics of the genome facilitate phenotypic evolution, we tested whether genomic characteristics associated with increased opportunities for mutation, such as highly mutagenic CpG sites, are reliable predictors of the sites of trait evolution but did not find any significant trends. Finally, we combined the QTL map with previously collected expression and selection data to identify 36 candidate genes that may underlie the repeated evolution of cave phenotypes, including rgrb, which is predicted to be involved in phototransduction. We found this gene has disrupted exons in all non-hybrid cave populations but intact reading frames in surface fish. Overall, our results suggest specific regions of the genome may play significant roles in driving adaptation to the cave environment in Astyanax mexicanus and demonstrate how this compiled dataset can facilitate our understanding of the genetic basis of repeated evolution in the Mexican cavefish.

Multiomics Screening Identified CpG Sites and Genes That Mediate the Impact of Exposure to Environmental Chemicals on Cardiometabolic Traits.

An understanding of the molecular mechanism whereby an environmental chemical causes a disease is important for the purposes of future applications. In this study, a multiomics workflow was designed to combine several publicly available datasets in order to identify CpG sites and genes that mediate the impact of exposure to environmental chemicals on cardiometabolic traits. Organophosphate and prenatal lead exposure were previously reported to change methylation level at the cg23627948 site. The outcome of the analyses conducted in this study revealed that, as the cg23627948 site becomes methylated, the expression of the GNA12 gene decreases, which leads to a higher body fat percentage. Prenatal perfluorooctane sulfonate exposure was reported to increase the methylation level at the cg21153102 site. Findings of this study revealed that higher methylation at this site contributes to higher diastolic blood pressure by changing the expression of CHP1 and GCHFR genes. Moreover, HKR1 mediates the impact of B12 supplementation → cg05280698 hypermethylation on higher kidney function, while CTDNEP1 mediates the impact of air pollution → cg03186999 hypomethylation on higher systolic blood pressure. This study investigates CpG sites and genes that mediate the impact of environmental chemicals on cardiometabolic traits. Furthermore, the multiomics approach described in this study provides a convenient workflow with which to investigate the impact of an environmental factor on the body's biomarkers, and, consequently, on health conditions, using publicly available data.

Post-COVID-19 Pain Is Not Associated with DNA Methylation Levels of the ACE2 Promoter in COVID-19 Survivors Hospitalized Due to SARS-CoV-2 Infection.

One of theories explaining the development of long-lasting symptoms after an acute severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection include changes in the methylation pattern of the host. The current study aimed to investigate whether DNA methylation levels associated with the angiotensin-converting enzyme 2 (ACE2) promoter are different when comparing individuals previously hospitalized due to COVID-19 who then developed long-lasting post-COVID pain with those previously hospitalized due to COVID-19 who did not develop post-COVID-19 pain symptoms. Non-stimulated saliva samples were obtained from a cohort of 279 (mean age: 56.5, SD: 13.0 years old, 51.5% male) COVID-19 survivors who needed hospitalization. Clinical data were collected from hospital medical records. Participants were asked to disclose pain symptoms developed during the first three months after hospital admission due to COVID-19 and persisting at the time of the interview. Methylations of five CpG dinucleotides in the ACE2 promoter were quantified (as percentages). Participants were evaluated up to 17.8 (SD: 5.3) months after hospitalization. Thus, 39.1% of patients exhibited post-COVID-19 pain. Most patients (77.05%) in the cohort developed localized post-COVID-19 pain. Headache and pain in the lower extremity were experienced by 29.4% of the patients. Seven patients received a post-infection diagnosis of fibromyalgia based on the presence of widespread pain characteristics (11.6%) and other associated symptoms. No significant differences in methylation percentages at any CpG location of the ACE2 promoter were identified when comparing individuals with and without post-COVID-19 pain. The current study did not observe differences in methylation levels of the ACE2 promoter depending on the presence or absence of long-lasting post-COVID-19 pain symptoms in individuals who needed hospitalization due to COVID-19 during the first wave of the pandemic.

The correlation between CpG methylation and gene expression is driven by sequence variants

Gene promoter and enhancer sequences are bound by transcription factors and are depleted of methylated CpG sites (cytosines preceding guanines in DNA). The absence of methylated CpGs in these sequences typically correlates with increased gene expression, indicating a regulatory role for methylation. We used nanopore sequencing to determine haplotype-specific methylation rates of 15.3 million CpG units in 7,179 whole-blood genomes. We identified 189,178 methylation depleted sequences where three or more proximal CpGs were unmethylated on at least one haplotype. A total of 77,789 methylation depleted sequences (~41%) associated with 80,503 cis-acting sequence variants, which we termed allele-specific methylation quantitative trait loci (ASM-QTLs). RNA sequencing of 896 samples from the same blood draws used to perform nanopore sequencing showed that the ASM-QTL, that is, DNA sequence variability, drives most of the correlation found between gene expression and CpG methylation. ASM-QTLs were enriched 40.2-fold (95% confidence interval 32.2, 49.9) among sequence variants associating with hematological traits, demonstrating that ASM-QTLs are important functional units in the noncoding genome.

Pyrimidine-Dependent UV-Mediated Cross-Linking Magnifies Minor Genetic or Epigenetic Changes in Clinical Samples.

Detection of minor DNA allele alterations is becoming increasingly important for early detection and monitoring of cancer. We describe a new method that uses ultraviolet light to eliminate wild-type DNA alleles and enables improved detection of minor genetic or epigenetic changes. Pyrimidine-dependent UV-based minor-allele enrichment (PD-UVME) employed oligonucleotide probes that incorporated a UVA-sensitive 3-cyanovinylcarbazole (CNVK), placed directly opposite interrogated pyrimidines, such as thymine (T) or cytosine (C) in wild-type (WT) DNA. Upon UVA-illumination, CNVK cross-linked with T/C, preventing subsequent amplification. Mutations that removed the T/C escaped cross-linking and were amplified and detected. Similarly, CNVK discriminated between methylated and unmethylated cytosine in CpG dinucleotides, enabling direct enrichment of unmethylated DNA targets. PD-UVME was combined with digital droplet PCR (ddPCR) to detect serine/threonine-protein kinase B-Raf (BRAF) V600E mutations in model systems, thyroid patient cancer tissue samples, and circulating DNA of tumor origin (ctDNA) from melanoma patients. One thyroid cancer sample out of 9, and 6 circulating-DNA samples out of 7 were found to be BRAF V600E-positive via PD-UVME while classified as negative by conventional ddPCR. Positive samples via conventional ddPCR were also found positive via PD-UVME. All 10 circulating cell-free DNA (cfDNA) samples obtained from normal volunteers were negative via both approaches. Furthermore, preferential enrichment of unmethylated alleles in MAGEA1 promoters using PD-UVME was demonstrated. PD-UVME mutation/methylation enrichment performed prior to ddPCR magnifies low-level mutations or epigenetic changes and increases sensitivity and confidence in the results. It can assist with clinical decisions that hinge on the presence of trace alterations like BRAF V600E.

Sex-specific epigenetic signatures of circulating urate and its increase after BCG vaccination.

Urate concentration and the physiological regulation of urate homeostasis exhibit clear sex differences. DNA methylation has been shown to explain a substantial proportion of serum urate variance, mediate the genetic effect on urate concentration, and co-regulate with cardiometabolic traits. However, whether urate concentration is associated with DNA methylation in a sex-dependent manner is unknown. Additionally, it is worth investigating if urate changes after perturbations, such as vaccination, are associated with DNA methylation in a sex-specific manner. We investigated the association between DNA methylation and serum urate concentrations in a Dutch cohort of 325 healthy individuals. Urate concentration and DNA methylation were measured before and after Bacillus Calmette-Guérin (BCG) vaccination, used as a perturbation associated with increased gout flares. The association analysis included united, interaction, and sex-stratified analysis. Validation of the identified CpG sites was conducted using three independent cohorts. 215 CpG sites were associated with serum urate in males, while 5 CpG sites were associated with serum urate in females, indicating sex-specific associations. Circulating urate concentrations significantly increased after BCG vaccination, and baseline DNA methylation was associated with differences in urate concentration before and after vaccination in a sex-specific manner. The CpG sites associated with urate concentration in males were enriched in neuro-protection pathways, whereas in females, the urate change-associated CpG sites were related to lipid and glucose metabolism. Our study enhances the understanding of how epigenetic factors contribute to regulating serum urate levels in a sex-specific manner. These insights have significant implications for the diagnosis, prevention, and treatment of various urate-related diseases and highlight the importance of personalized and sex-specific approaches in medicine.

Targeting DNA junction sites by bis-intercalators induces topological changes with potent antitumor effects.

Targeting inter-duplex junctions in catenated DNA with bidirectional bis-intercalators is a potential strategy for enhancing anticancer effects. In this study, we used d(CGTATACG)2, which forms a tetraplex base-pair junction that resembles the DNA-DNA contact structure, as a model target for two alkyl-linked diaminoacridine bis-intercalators, DA4 and DA5. Cross-linking of the junction site by the bis-intercalators induced substantial structural changes in the DNA, transforming it from a B-form helical end-to-end junction to an over-wounded side-by-side inter-duplex conformation with A-DNA characteristics and curvature. These structural perturbations facilitated the angled intercalation of DA4 and DA5 with propeller geometry into two adjacent duplexes. The addition of a single carbon to the DA5 linker caused a bend that aligned its chromophores with CpG sites, enabling continuous stacking and specific water-mediated interactions at the inter-duplex contacts. Furthermore, we have shown that the different topological changes induced by DA4 and DA5 lead to the inhibition of topoisomerase 2 activities, which may account for their antitumor effects. Thus, this study lays the foundations for bis-intercalators targeting biologically relevant DNA-DNA contact structures for anticancer drug development.

Tumor detection by analysis of both symmetric- and hemi-methylation of plasma cell-free DNA

Aberrant DNA methylation patterns have been used for cancer detection. However, DNA hemi-methylation, present at about 10% CpG dinucleotides, has been less well studied. Here we show that a majority of differentially hemi-methylated regions (DHMRs) in liver tumor DNA or plasma cells free (cf) DNA do not overlap with differentially methylated regions (DMRs) of the same samples, indicating that DHMRs could serve as independent biomarkers. Furthermore, we analyzed the cfDNA methylomes of 215 samples from individuals with liver or brain cancer and individuals without cancer (controls), and trained machine learning models using DMRs, DHMRs or both. The models incorporated with both DMRs and DHMRs show a superior performance compared to models trained with DMRs or DHMRs, with AUROC being 0.978, 0.990, and 0.983 in distinguishing control, liver and brain cancer, respectively, in a validation cohort. This study supports the potential of utilizing both DMRs and DHMRs for multi-cancer detection.

Immune gene regulation is associated with age and environmental adversity in a nonhuman primate.

Phenotypic aging is ubiquitous across mammalian species, suggesting shared underlying mechanisms of aging. Aging is linked to molecular changes to DNA methylation and gene expression, and environmental factors, such as severe external challenges or adversities, can moderate these age-related changes. Yet, it remains unclear whether environmental adversities affect gene regulation via the same molecular pathways as chronological, or 'primary', aging. Investigating molecular aging in naturalistic animal populations can fill this gap by providing insight into shared molecular mechanisms of aging and the effects of a greater diversity of environmental adversities - particularly those that can be challenging to study in humans or laboratory organisms. Here, we characterised molecular aging - specifically, CpG methylation - in a sample of free-ranging rhesus macaques living off the coast of Puerto Rico (n samples = 571, n individuals = 499), which endured a major hurricane during our study. Age was associated with methylation at 78,661 sites (31% of all sites tested). Age-associated hypermethylation occurred more frequently in areas of active gene regulation, while hypomethylation was enriched in regions that show less activity in immune cells, suggesting these regions may become de-repressed in older individuals. Age-associated hypomethylation also co-occurred with increased chromatin accessibility while hypermethylation showed the opposite trend, hinting at a coordinated, multi-level loss of epigenetic stability during aging. We detected 32,048 CpG sites significantly associated with exposure to a hurricane, and these sites overlapped age-associated sites, most strongly in regulatory regions and most weakly in quiescent regions. Together, our results suggest that environmental adversity may contribute to aging-related molecular phenotypes in regions of active gene transcription, but that primary aging has specific signatures in non-regulatory regions.

Unveiling Bladder Cancer Prognostic Insights by Integrating Patient-Matched Sample and CpG Methylation Analysis.

Bladder cancer prognosis remains a pressing clinical challenge, necessitating the identification of novel biomarkers for precise survival prediction and improved quality of life outcomes. This study proposes a comprehensive strategy to uncover key prognostic biomarkers in bladder cancer using DNA methylation analysis and extreme survival pattern observations in matched pairs of cancer and adjacent normal cells. Unlike traditional approaches that overlook cancer heterogeneity by analyzing entire samples, our methodology leverages patient-matched samples to account for this variability. Specifically, DNA methylation profiles from adjacent normal bladder tissue and bladder cancer tissue collected from the same individuals were analyzed to pinpoint critical methylation changes specific to cancer cells while mitigating confounding effects from individual genetic differences. Utilizing differential threshold settings for methylation levels within cancer-associated pathways enabled the identification of biomarkers that significantly impact patient survival. Our analysis identified distinct survival patterns associated with specific CpG sites, underscoring these sites' pivotal roles in bladder cancer outcomes. By hypothesizing and testing the influence of methylation levels on survival, we pinpointed CpG biomarkers that profoundly affect the prognosis. Notably, CpG markers, such as cg16269144 (PRKCZ), cg16624272 (PTK2), cg11304234, and cg26534425 (IL18), exhibited critical methylation thresholds that correlate with patient mortality. This study emphasizes the importance of tailored approaches to enhancing prognostic accuracy and refining therapeutic strategies for bladder cancer patients. The identified biomarkers pave the way for personalized prognostication and targeted interventions, promising advancements in bladder cancer management and patient care.

Somatic CpG hypermutation is associated with mismatch repair deficiency in cancer

Somatic hypermutation in cancer has gained momentum with the increased use of tumour mutation burden as a biomarker for immune checkpoint inhibitors. Spontaneous deamination of 5-methylcytosine to thymine at CpG dinucleotides is one of the most ubiquitous endogenous mutational processes in normal and cancer cells. Here, we performed a systematic investigation of somatic CpG hypermutation at a pan-cancer level. We studied 30,191 cancer patients and 103 cancer types and developed an algorithm to identify somatic CpG hypermutation. Across cancer types, we observed the highest prevalence in paediatric leukaemia (3.5%), paediatric high-grade glioma (1.7%), and colorectal cancer (1%). We discovered germline variants and somatic mutations in the mismatch repair complex MutSα (MSH2-MSH6) as genetic drivers of somatic CpG hypermutation in cancer, which frequently converged on CpG sites and TP53 driver mutations. We further observe an association between somatic CpG hypermutation and response to immune checkpoint inhibitors. Overall, our study identified novel cancer types that display somatic CpG hypermutation, strong association with MutSα-deficiency, and potential utility in cancer immunotherapy.

Integrating muti-omics data to identify tissue-specific DNA methylation biomarkers for cancer risk

The relationship between tissue-specific DNA methylation and cancer risk remains inadequately elucidated. Leveraging resources from the Genotype-Tissue Expression consortium, here we develop genetic models to predict DNA methylation at CpG sites across the genome for seven tissues and apply these models to genome-wide association study data of corresponding cancers, namely breast, colorectal, renal cell, lung, ovarian, prostate, and testicular germ cell cancers. At Bonferroni-corrected P < 0.05, we identify 4248 CpGs that are significantly associated with cancer risk, of which 95.4% (4052) are specific to a particular cancer type. Notably, 92 CpGs within 55 putative novel loci retain significant associations with cancer risk after conditioning on proximal signals identified by genome-wide association studies. Integrative multi-omics analyses reveal 854 CpG-gene-cancer trios, suggesting that DNA methylation at 309 distinct CpGs might influence cancer risk through regulating the expression of 205 unique cis-genes. These findings substantially advance our understanding of the interplay between genetics, epigenetics, and gene expression in cancer etiology.

Genome-Wide DNA methylation profile analysis identifies differentially methylated loci associated with personal PM2.5 exposure in adults with asthma

Particulate matter with aerodynamic diameters ≤2.5 µm (PM2.5) is a major environmental risk factor for acute asthma exacerbation, and the underlying mechanism is not completely understood. Studies have indicated that DNA methylation is a potential mechanism linking PM2.5 to its health effects. We conducted a panel study involving 24 adult patients with asthma in Beijing,China between 2017 and 2019. PM2.5 and other atmospheric pollutant exposure data were repeatedly measured. Blood samples were collected for genome-wide DNA methylation analysis. A linear mixed-effects (LME) model was conducted to identify differentially methylated probes (DMPs) associated with PM2.5 exposure. After filtering out probes that did not meet the criteria through quality control, 811,001 CpG sites were included in the LME model, and 36 DMPs were strongly associated with personal PM2.5 exposure at false discovery rate (FDR) < 0.05, of which 22 and 14 DMPs negatively and positively correlated with personal PM2.5 exposure, respectively. Functional analysis revealed that DMPs affected smooth muscle cell contraction and development, extracellular matrix synthesis and secretion, T cell activation and differentiation, and inflammatory factor production. This study provides evidence linking personal PM2.5 exposure to genome-wide DNA methylation in adult patients with asthma. Identifying enrichment pathways can provide biological insights into the acute health effects of PM2.5.

Generations of epigenetic clocks and their links to socioeconomic status in the Health and Retirement Study.

Aim: This is a brief description of links between nine epigenetic clocks related to human aging and socioeconomic and behavioral characteristics as well as health outcomes.Materials& methods: We estimate frequently used and novel clocks from one data source, the Health and Retirement Study.Results: While all of these clocks are thought to reflect "aging," they use different CpG sites and do not strongly relate to each other. First and fourth generation clocks are not as linked to socioeconomic status or health outcomes as second and third generation clocks.Conclusion: Epigenetic clocks reflect exciting new tools and their continued evolution is likely to improve our understanding of how exposures get under the skin to accelerate aging.

Male infertility is associated with differential DNA methylation signatures of the imprinted gene GNAS and the non-imprinted gene CEP41.

To investigate whether the DNA methylation profiles of GNAS(20q13.32), MEST(7q32.2), MESTIT1(7q32.2), IGF2(11p15.5), H19 (7q32.2), and CEP41(7q32.2) genes are related to the transcriptomic and epigenomic etiology of male infertility. The DNA methylation levels of spermatozoa were obtained from fertile (n = 30), oligozoospermic (n = 30), and men with normal sperm count (n = 30). The methylation status of each CpG site was categorized as hypermethylated or hypomethylated. Expression levels of target gene transcripts were determined using real-time PCR. The oligozoospermia showed a higher frequency of hypermethylation at GNASAS 1st, 3rd, and 5th CpG dinucleotides (66.7%, 73.3%, 73.3%) compared to the fertile group (33.3%, 33.3%, 40%, respectively). The normal sperm count exhibited a higher frequency of hypermethylation at the 3rd CpG of CEP41 (46.7%) than the fertile group (16.7%). Normal sperm count was predicted by CEP41 hypermethylation (OR = 1.750, 95%CI 1.038-2.950) and hypermethylation of both CEP41 and GNASAS (OR = 2.389, 95%CI 1.137-5.021). Oligozoospermia was predicted solely by GNASAS hypermethylation (OR = 2.460, 95%CI 1.315-4.603). In sperms with decreased IGF2 expression in the fertile group, we observed hypomethylation in the 2nd CpG of IGF2 antisense (IFG2AS), and hypermethylation in the 1st, 2nd, and 4th CpGs of H19. No significant relationship was found between IGF2 expression and methylation status of IGF2AS and H19 in infertile groups. The disappearance of the relationship between IGF2 expression and IGF2AS and H19 methylations in the infertile group provides new information regarding the disruption of epigenetic programming during spermatogenesis. A better understanding of sperm GNASAS and CEP41 hypermethylation could advance innovative diagnostic markers for male infertility.

Human Brain Barcodes.

Dynamic CpG methylation "barcodes" were read from 15,000 to 21,000 single cells from three human male brains. To overcome sparse sequencing coverage, the barcode had ~31,000 rapidly fluctuating X-chromosome CpG sites (fCpGs), with at least 500 covered sites per cell and at least 30 common sites between cell pairs (average of ~48). Barcodes appear to start methylated and record mitotic ages because excitatory neurons and glial cells that emerge later in development were less methylated. Barcodes are different between most cells, with average pairwise differences (PWDs) of ~0.5 between cells. About 10 cell pairs per million were more closely related with PWDs < 0.05. Barcodes appear to record ancestry and reconstruct trees where more related cells had similar phenotypes, albeit some pairs had phenotypic differences. Inhibitory and excitatory neurons both showed evidence of tangential migration with related cells in different cortical regions. fCpG barcodes become polymorphic during development and can distinguish between thousands of human cells.

Molecular subtypes explain lupus epigenomic heterogeneity unveiling new regulatory genetic risk variants

The heterogeneity of systemic lupus erythematosus (SLE) can be explained by epigenetic alterations that disrupt transcriptional programs mediating environmental and genetic risk. This study evaluated the epigenetic contribution to SLE heterogeneity considering molecular and serological subtypes, genetics and transcriptional status, followed by drug target discovery. We performed a stratified epigenome-wide association studies of whole blood DNA methylation from 213 SLE patients and 221 controls. Methylation quantitative trait loci analyses, cytokine and transcription factor activity - epigenetic associations and methylation-expression correlations were conducted. New drug targets were searched for based on differentially methylated genes. In a stratified approach, a total of 974 differential methylation CpG sites with dependency on molecular subtypes and autoantibody profiles were found. Mediation analyses suggested that SLE-associated SNPs in the HLA region exert their risk through DNA methylation changes. Novel genetic variants regulating DNAm in disease or in specific molecular contexts were identified. The epigenetic landscapes showed strong association with transcription factor activity and cytokine levels, conditioned by the molecular context. Epigenetic signals were enriched in known and novel drug targets for SLE. This study reveals possible genetic drivers and consequences of epigenetic variability on SLE heterogeneity and disentangles the DNAm mediation role on SLE genetic risk and novel disease-specific meQTLs. Finally, novel targets for drug development were discovered.

Decoding depression: a comprehensive multi-cohort exploration of blood DNA methylation using machine learning and deep learning approaches

The causes of depression are complex, and the current diagnosis methods rely solely on psychiatric evaluations with no incorporation of laboratory biomarkers in clinical practices. We investigated the stability of blood DNA methylation depression signatures in six different populations using six public and two domestic cohorts (n = 1942) conducting mega-analysis and meta-analysis of the individual studies. We evaluated 12 machine learning and deep learning strategies for depression classification both in cross-validation (CV) and in hold-out tests using merged data from 8 separate batches, constructing models with both biased and unbiased feature selection. We found 1987 CpG sites related to depression in both mega- and meta-analysis at the nominal level, and the associated genes were nominally related to axon guidance and immune pathways based on enrichment analysis and eQTM data. Random forest classifiers achieved the highest performance (AUC 0.73 and 0.76) in CV and hold-out tests respectively on the batch-level processed data. In contrast, the methylation showed low predictive power (all AUCs < 0.57) for all classifiers in CV and no predictive power in hold-out tests when used with harmonized data. All models achieved significantly better performance (>14% gain in AUCs) with pre-selected features (selection bias), with some of the models (joint autoencoder-classifier) reaching AUCs of up to 0.91 in the final testing regardless of data preparation. Different algorithmic feature selection approaches may outperform limma, however, random forest models perform well regardless of the strategy. The results provide an overview over potential future biomarkers for depression and highlight many important methodological aspects for DNA methylation-based depression profiling including the use of machine learning strategies.