DNA extraction and bisulfite conversion were performed as described previously27 (link). DNA samples were processed and run on HumanMethylation450 (HM450) arrays following the manufacturer’s instructions (Illumina Inc.). Raw intensity data was read, preprocessed and batch corrected in R (v3.2.2). After subsequent filtering, 426,718 probes were left for analysis. Additional HM450 array datasets include: Level 1 HM450 array data from 87 EAC and 11 adjacent normal tissue samples obtained from TCGA on 11/13/2015 (https://tcga-data.nci.nih.gov/tcga/), and 125 EAC samples from the Gene Expression Omnibus (study ID: GSE72872) 23 (link).