Predicting nicotine metabolism across ancestries using genotypes

BMC Genomics

Table 2 Correlations between predicted and observed NMRs by study, ancestry, and model. The correlations between predicted and observed NMRs were summarized overall and by genomic ancestry (ancestry proportion \(>0.5\)). The MEC was the largest and most diverse sample and was used for training using partial least squares (PLS), project pursuit (PPR), elastic net (ENet), support vector machine with a linear kernel (SVM-L), support vector machine with a radial basis function kernel (SVM-R), gradient boosting machine (GBM), and random forests (RF). Predictions from these seven models were averaged in an ensemble model. The MEC trained models were applied to CENIC, HSS, and METs for validation

	PLS	PPR	ENet	SVM-L	SVM-R	GBM	RF	Ensemble	N
MEC (Training)
African	0.60	0.67	0.58	0.75	0.58	0.76	0.97	0.76	342
Asian	0.71	0.76	0.70	0.79	0.69	0.81	0.97	0.82	995
European	0.50	0.56	0.49	0.63	0.48	0.64	0.97	0.67	902
Overall	0.71	0.75	0.70	0.78	0.69	0.79	0.97	0.81	2239
CENIC (Validation)
African	0.35	0.33	0.32	0.33	0.32	0.41	0.41	0.37	111
Asian	0.71	0.28	0.68	0.60	0.67	0.51	0.50	0.61	9
European	0.42	0.38	0.43	0.41	0.37	0.52	0.47	0.46	395
Overall	0.42	0.37	0.41	0.39	0.37	0.50	0.45	0.45	515
HSS (Validation)
African									1
Asian	0.51	0.29	0.55	0.59	0.53	0.55	0.58	0.56	308
European	0.42	0.33	0.42	0.37	0.36	0.42	0.39	0.43	271
Overall	0.53	0.37	0.56	0.56	0.53	0.55	0.56	0.56	580
METS (Validation)
African	0.43	0.30	0.50	0.55	0.41	0.46	0.45	0.52	48
Asian	0.37	0.03	0.39	0.47	0.41	0.47	0.47	0.43	51
European	0.38	0.10	0.40	0.45	0.42	0.44	0.41	0.43	216
Overall	0.36	0.09	0.42	0.42	0.39	0.45	0.45	0.40	315

ISSN: 1471-2164