recountmethylation

recountmethylation provides resources for cross-study analyses of public DNA methylation array data from the NCBI Gene Expression Omnibus repository, produced using Illumina's Infinium HumanMethylation450K (HM450K) and MethylationEPIC (EPIC) platforms.

Key Features:

• Data compilation access: Enables download, summary, and filtering of large compilation files containing uniformly processed DNAm array data.
• Blood sample compilation: Includes 12,537 uniformly processed EPIC and HM450K blood samples from GEO for cross-study and cross-platform analyses.
• Metadata processing: Learns controlled vocabulary of sample labels by applying regular expressions to metadata and predicts various sample properties including epigenetic age.
• Quality assessment integration: Incorporates Illumina's 17 prescribed quality assessments and supports evaluation of signal distributions across samples.
• Tissue-specific analysis support: Provides compiled DNAm array data and metadata for seven tissues including adipose, nasal, blood, brain, buccal, sperm, and liver.

Scientific Applications:

• Cross-study replication: Facilitates experiment planning, replication, and cross-study and cross-platform analyses of publicly available DNAm array samples.
• Study ID bias adjustment: Enables adjustment for study ID bias to increase variation explained by biological and demographic variables.
• Variation decomposition: Supports analysis of autosomal DNAm variation explained by genetic ancestry and cell type fractions such as CD4+ T-cell fractions.
• Power analysis: Supports evaluation of power to detect differential methylation across sample types including peripheral blood mononuclear cells (PBMC), whole blood, and umbilical cord blood.
• Independent validation: Enables independent validation of differentially methylated probes (DMPs) from epigenome-wide association studies (EWAS).
• Tissue comparison: Supports characterization of specific probes distinguishing seven different tissues.

Methodology:

The package compiles DNAm array data and metadata into database files accessible via R/Bioconductor, with functions for download, summary, and filtering. Sample properties are predicted using existing models, and sample labels are learned by applying regular expressions to metadata. Uniform processing of blood samples enables cross-study and cross-platform analyses.