DNA methylation (DNAm) is highly tissue-specific, and neuropsychiatric epigenetics therefore relies most directly on DNAm profiles measured in the human brain. Yet, brain tissue is rarely accessible from living individuals, forcing most studies to rely on peripheral proxies such as blood, saliva, and buccal cells. A central unresolved question is which peripheral DNAm signals truly reflect DNAm variation in the brain. IMAGE-CpG2 provides an empirical resource to address this challenge by integrating three newly generated paired cohorts (DB6–DB8) into a single cross-tissue DNAm correspondence database. With 148 individuals and 599 EPIC-array samples across brain and matched peripheral tissues, IMAGE-CpG2 represents the world's largest brain–periphery DNAm correlation resource to date. The web tool enables users to query and visualize brain–periphery correspondence at single CpG sites and gene-based loci, supporting interpretation of peripheral DNAm findings in the context of the brain and prioritization of loci with stronger brain relevance.
Characteristics
Integrated cohorts: DB6 (Adult-Biopsy: 40), DB7 (Adult-Autopsy: 82), DB8 (Pediatric: 26)
Platform: Illumina Infinium MethylationEPIC BeadChip (EPIC v1/v2 = 89/59 arrays; probe annotation mapped to the hg38/GRCh38 reference genome)
Sample source: Biopsy/Autopsy = 62/86
Brain regions: Frontal cortex (98), Temporal cortex (37), Amygdala (61), Hippocampus (98), Other (9)
Peripheral tissue: Blood (135), Saliva (38), Buccal epithelial (123)
QC is conducted within each cohort (DB6/DB7/DB8). Data are normalized up to ssNoob; no batch effect removal is applied to avoid potential distortions of biological variation.
Cross-tissue correspondence is reported using Spearman’s rho (rank-based and robust under the above non-harmonized integration strategy)
An optional multi‑tissue plot mode is available, which visualizes DNAm levels across selected tissues for a queried CpG or gene.
Companion resources are also available: COMPASS (cross-database meta-analysis) and TRACE-CpG (a “Virtual Brain Biopsy” prediction model).
*: Includes data from the previous version of IMAGE-CpG