DNA methylation measurements and statistical analyses
DNA was extracted from 72 whole blood and 103 nasal brushing samples. Genome-wide DNA methylation was determined using Illumina Infinium HumanMethylation450 BeadChips. After quality control, 72 whole blood samples and 97 nasal epithelium samples (of in total 103 unique subjects) with 436,824 CpG sites remained for following steps.
We used robust linear regression to determine the differential methylation between persistent asthma and asthma remission: 1) PersA versus ClinR, and 2) PersA versus ComR, in whole blood and nasal brushing samples, with adjustment for covariates that are known to affect DNA methylation (age, sex, smoking status, pack years, and batch). For whole blood samples, we performed adjustment on the percentage of monocytes, B cells, NK cells, CD4+ T cells, CD8+ T cells, neutrophils and eosinophils, which were predicted by the Houseman13 algorithm using minfi14package. For nasal brushing samples, we applied the R package sva15 to estimate significant surrogate variables (SVs), representing unknown latent factors that capture heterogeneity in data, such as cell type composition. Differentially methylated regions (DMR) were identified using comb-p v0.4816.