Table 1.

Comparison of model performances of PLS with sevenfold cross-validation between full metabolome vs post-selection.

		Full metabolome					Post-selection
	nVar	nComp	R2Y	Q2	PCV-ANOVA	r	nVar	nComp	R2Y	Q2	PCV-ANOVA	r
Pancreatic fat	910	1	0.39	0.25	2.43E–04	0.63	56	3	0.81	0.69	2.12E–12	0.90
Liver fat	910	1	0.52	0.33	6.86E–06	0.72	64	3	0.80	0.66	6.11E–12	0.89
VAT/SAT	910	1	0.47	0.30	1.32E–05	0.69	31	2	0.70	0.62	1.88E–12	0.83

nVar: number of variables; nComp: number of components; R2Y: goodness of model fit; Q2: predictivity; Pcv-anova: statistical significance of the PLS model; r: Pearson correlation coefficient.