Additional file 2 of Identifying epigenetic biomarkers of established prognostic factors and survival in a clinical cohort of individuals with oropharyngeal cancer

Additional file 2: Supplementary Table 1. Genome-wide differentially-methylated CpG sites associated with smoking status below a multiple testing threshold of P < 2.4e-07. Results are adjusted for age, sex, surrogate variables obtained by SVA, alcohol consumption and HPV16E6 seropositivity. Supplementary Table 2. - Genome-wide differentially-methylated CpG sites associated with alcohol consumption below a multiple testing threshold of P < 2.4e-07. Results are adjusted for age, sex, surrogate variables obtained by SVA, smoking status and HPV16E6 seropositivity. Supplementary Table 3. - Genome-wide differentially-methylated CpG sites associated with ~3-year survival below a multiple testing threshold of P < 2.4e-07. Results are adjusted for age, sex and surrogate variables obtained by SVA. Supplementary Table 4. Genome-wide differentially-methylated CpG sites associated with ~3-year survival below a multiple testing threshold of P < 2.4e-07. Results are adjusted for age, sex, surrogate variables obtained by SVA, smoking status, alcohol consumption and HPV16E6 seropositivity. Supplementary Table 5. Genetic instrumental variables (IVs) used in Mendelian randomization analyses to assess epigenetic mediation between prognostic factors and ~3-year survival. The final # SNPs denotes genetic IVs which both proxy a CpG and where the same position is available in the genome-wide association study of 3-year survival survival. Supplementary Table 6. Lookup of CpG sites in the MRCIEU EWAS Catalog across all EWAS analyses below a Bonferroni p-value threshold of 5.7e-08. Betas for all studies reporting beta values are calculated as a weighted mean, weighted by sample size. Supplementary Table 7. CpG sites (P<2.4e-7) associated with ~3-year survival adjusted for age, sex and surrogate variables obtained by SVA, compared against betas, standard errors and p-values at the same sites when comorbidity and stage are included as additional covariates in the EWAS model.