A detailed description of methylomic profiling protocols can be found in Milnik et al.17 (link). Briefly, methylomic profiling was performed using the Illumina HumanMethylation450 array. Samples of non-European ancestry were identified using Hapmap references population genotypes and excluded from analysis (n = 35 in BASEL1 sample, yielding N = 533 remaining for analysis; none identified in BASEL2 sample). The β-values were calculated from SWAN normalized intensities20 (link). Subsequently, β-values were M-transformed and adjusted for processing plate effect (z-transformation), age, sex, and the main sources of technical variations inferred from principal components analysis17 (link). The β-values with detection p-value > 0.05 were considered as missing. Individual CpG sites were excluded based on the following criteria: non-CpG context, non-autosomal probes, probes with a SNP mapping to the target CpG site or with three or more SNPs within the 50-mer probe (minor allele frequency (MAF )> 0.01) (based on RnBeads package annotation), multi-mapping or polymorphic CpGs (MAF > 0.01 in European population) reported in refs.21 (link),22 (link), and probes with missing rate ≥5% in final samples. Prior to analysis, missing values were imputed using the R package impute (https://bioconductor.org/packages/release/bioc/html/impute.html) with k = 10.
Free full text: Click here