Epigenetic modelling of former, current and never smokers

Clinical Epigenetics

Table 1 Initial and final numbers of CpGs for each DNAm model of smoking

Classes	Model name	Novel/literature	Number of supplied features to LASSO (CpGs)	Final number of features (CpGs)
“Ever” versus never	AHRR	Literature	NA	1
	Maas	Literature	NA	13
	Agnostic LASSO	Novel	450K	29
	Candidate CpG LASSO	Novel	14	9
Current versus former	AHRR	Literature	NA	1
	Maas	Literature	NA	13
	Agnostic LASSO	Novel	450K	20
	Candidate CpG LASSO	Novel	40	4

“Literature”-based models contain pre-specified CpG sites and betas and were therefore not supplied to LASSO models in this paper. “Novel” models denote models where we supplied sets of CpGs for feature selection via cross-validated LASSO. The “final number of features” are those used to create the various DNAm scores of smoking seen in this paper

ISSN: 1868-7083