Table 2. Significant associations between tobacco smoking and methylation of lung cancer related genes in validation panel.

CpG site	Gene	Mean β value (Standard deviation)		Effect size b	Estimate (se)	p-value	FDR
		Never smoker	Current smoker
cg00640087	MSH5	0.165 (0.036)	0.159 (0.035)	−0.006	−7.4 e-3 (3.1 e-3)	0.019	0.049
cg03281572	VTI1A	0.812 (0.028)	0.793 (0.036)	−0.019	−0.018 (3.0 e-3)	3.8 e-7	1.2 e-5
cg07269053	VTI1A	0.733 (0.039)	0.715 (0.052)	−0.018	−0.013 (5.0 e-3)	0.007	0.023
cg10163955	GATA3	0.669 (0.043)	0.640 (0.049)	−0.029	−0.024 (5.1 e-3)	5.1 e-4	6.6 e-5
cg11430077	GATA3	0.147 (0.032)	0.132 (0.029)	−0.015	−0.013 (4.0 e-3)	0.001	0.006
cg12324353	TERT	0.788 (0.032)	0.779 (0.032)	−0.009	−0.011 (3.5 e-3)	0.002	0.011
cg17928584	STK32A	0.156 (0.053)	0.161 (0.052)	0.005	0.012 (5.0 e-3)	0.020	0.049
cg19335412	ACTA2	0.461 (0.036)	0.451 (0.033)	−0.010	−0.011 (4.1 e-3)	0.009	0.026
cg19696491	CHRNA5	0.470 (0.058)	0.488 (0.060)	0.018	0.018 (6.9 e-3)	0.010	0.028
cg20640261	MSH5	0.443 (0.048)	0.424 (0.048)	−0.019	−0.015 (5.0 e-3)	0.003	0.013
cg22770911	GATA3	0.481 (0.033)	0.458 (0.042)	−0.023	−0.015 (4.5 e-3)	0.001	0.005
cg24287110	KLF6	0.365 (0.056)	0.349 (0.053)	−0.016	−0.022 (6.0 e-3)	6.2 e-4	0.005
cg24908166	TERT	0.926 (0.021)	0.916 (0.026)	−0.010	−0.010 (2.6 e-3)	0.0001	0.001

^aAdjusted for age (years), sex, random batch effects, leukocyte distribution (Houseman algorithm [27]), alcohol consumption (abstainer/ low/ intermediate/ high), body mass index (BMI, underweight/ normal weight/ overweight/ obese), physical activity (inactive/ low/ medium or high), prevalence of cardiovascular diseases (yes/no), prevalence of diabetes (yes/no) and prevalence of cancer (yes/no)

All 31 loci identified by discovery panel were validated by the three models, and the threshold of FDR is 0.05. A total of 13 CpG sites were validated as significant smoking-related CpG sites by validation

^bEffect size = Mean β_{current smoker} – Mean β_{never smoker}