Thesis

196 Chapter 7 SUPPLEMENTARY MATERIAL Supplementary Methods Further detailed description of following methods: 1. Verification of methylation assay efficiency 2. Linear mixed-effects models 1. Verification of methylation assay efficacy Promoter hypermethylation detection of the CDO1, SOX17, and TAC1 genes was carried out by quantitative methylation-specific PCR. The efficiency of this assay was verified using 11 pairs of tumours and adjacent normal tissues from NSCLC patients of a previously published cohort. Differences in DNA methylation levels between cancerous and non-cancerous tissue was evaluated by comparing the square root cycle threshold (ct) ratios. Methylation levels were displayed in boxplots and tested for statistical significance using the nonparametric paired samples Wilcoxon test. 2. Linear mixed-effect models Linear mixed-effects modelling for the effect of the circadian rhythm on cfDNA concentration and methylation levels of CDO1, SOX17, and TAC1 Models were fitted by backward stepwise elimination (p ≥ 0.05 for removal) to select fixed and random parts of the linear mixed model using the ‘lmerTest’ package in R. Final models are displayed below: # Response variable Final model 1 cfDNA concentration lmer(DNAconc ~ time + day + gender + (1/Sample| subject), data = DAYTIME_R) 2 CDO1 methylation level lmer(CDO1sqrt ~ time + day + (1/Sample|subject), data = DAYTIME_R) 3 SOX17 methylation level lmer(SOX17sqrt ~ time + day + (1/Sample|subject), data = DAYTIME_R) 4 TAC1 methylation level lmer(TAC1sqrt ~ time + day + (1/Sample|subject), data = DAYTIME_R) Testing model assumptions The assumptions of linearity, normality of the residuals and random effects, and homoscedasticity (i.e. constant variance of the residuals) were checked visually. A series of diagnostic plots were computed using the ‘sjPlot’ package to check these assumptions.

RkJQdWJsaXNoZXIy MjY0ODMw