What is EWAS Atlas?

EWAS Atlas is used as a knowledgebase of epigenome-wide association studies (EWAS), which not only provides data from published EWAS, but also provides a user-friendly filter for effective retrieval of traits of interest. Moreover, based off of data from EWAS Atlas, a traits enrichment tool was developed to explore the relationshipw between traits or between traits and DNA methylation.

Eligibility criteria

Genome-wide methylation profiles are eligible for integration into EWAS if they meet the following criteria:
  • Based on 850K, 450K, 27K platforms
  • With available differentially methylated position (DMP)
  • From studies on human traits
  • The p-value is less than 1.0e-4 or q-value less than 0.05.
    (If a study perform genome-wide analysis for initial cohorts and then validate the identified DMP through pyrosequencing, the p-value from initial stage will be reported. For the meta-analysis, the combined p-value was taken as the standard. If a study does not report a combined p-value, the p-value from the largest sample size will be reported as long as the initial and replication samples each show an association of p < 1.0e-4.)
  • If a study has a p value cutoff less than 1.0e-4 and only provides associations corresponded to their cutoff, all provided association are recorded.
  • For studies without p-values, all significantly DMP reported in those studies are recorded.

Data structure

  • Basic information: The basic information of publication is provided, including PMID, headlines, summaries, etc. The citations were extracted from Europe PMC.
  • Dataset: Datasets and their links are sourced from the article.
  • One publication may correspond to multiple studies.
  • Reported trait: The traits reported in the studies.
  • Case/Control group: For studies with adoption of case/control's research strategy, case/control is the description of the case group and the control group respectively. For studies with adoption of regression strategy, case is used for recording the independent variableļ¼Œand control for the dependent variable.
  • Other description: Details of research under special circumstances.
  • One study may correspond to multiple associations and multiple cohorts.
  • Genome coordinates: For 450K and 850K, the mapping position of probes was obtained from the Illumina website. For 27K, probes were remapped by LiftOver to the hg19 reference genome using probe sequence information provided by Illumina.
  • Rank: p value is susceptible to the influence of sample size and statistical methods. Therefore, rank is put forward as an alternative indicator of correlation effectiveness. In general, rank is the order of the p value of the loci in the studies. For research without p values in their methods, rank depends on the indicators given by statistic methods.
  • Correlation: For study of the regression analysis, the change in the case group relative to the control group, indicate the variation trend of the methylation (hypermethylation and hypomethylation) as the independent variable value increased.
  • The biological problem-oriented strategy is adopted for the classification of cohorts, which are recorded according to the article.
  • Age information: Maximum, minimum, average and variance of age.
  • Ancestry: Ancestry information of the samples is recorded and further divided into 17 groups on the basis of the article by Morales et al.
  • Platform: Sequencing platforms, such as 850K, 450K, 27K, etc.
  • Stage: The stage of sample for research, including the initial and the replication.
  • Tissue: Tissue or cell sample source
  • Description: Other descriptions of samples, such as disease information, etc.


Gene annotation
For each probe on 850K/450K/27K, the associated transcripts from GENCODE (Release 28), the associated CpG island (CGI) from UCSC genome browser and the CpG sites' distance from each of these features were identified.
GWAS data
The GWAS data includes SNPs, and related traits were download from GWAS Catalog. SNPs located 5000bp away from the probe both ways were retained.
The reported traits were first mapped to terms from the Experimental Factor Ontology (EFO) automatically by using API provided by ZOOMA and further checked by curators.

Trait enrichment tool

Input: Probe ID of 850K/450K/27K
  • Global traits mode
    In this mode, a modified Fisher's exact test is used to examine the significance of association between input probes and traits-associated probe lists in EWAS Atlas.
  • Custom traits mode
    Users can customize a set of phenotypic traits, and use modified Fisher's exact test to perform an association analysis between a set of input probes and custom traits associated probes, detecting whether the input probe is associated with the custom phenotypic traits.
  • Trait: The detected phenotype
  • -log10 (p): The converted p-value of modified fisher test
  • Count: Probes involved in the term
  • Percentage: Involved probes/total probes associated with this trait


The information on this website is curated from publications and not intended for direct diagnostic use or medical decision-making without review by a genetics professional. Individuals should not change their health behavior solely on the basis of information contained on this website. If you have any question about the information contained on this website, please see a health care professional.