Table 3.

Results for BE-PLS-DA: for each sampling site and each treatment time (3 h or 24 h) the global model (containing the data from all the omics sources) is compared to the model obtained for single sources. For each model, the number of LVs included in the final model (N° LVs) and the accuracy of classification both in fitting (Acc% Cal) and in cross-validation (Acc% cv) are provided. Numbers in parentheses in the second and third column represent the number of original and selected variables, respectively.

Sampling Site	Omics Sources and (Original Variables)	Omics Sources and (Selected Variables)	N° LVs	Acc% Cal	Acc% cv
Muscle 3 h	Lipid (921) + Metab (464) + Prot (66)	Lipid (45) + Metab (16) + Prot (1)	1	100	100
	Lipid (921)	Lipid (18)	1	100	100
	Metab (464)	Metab (14)	1	100	100
	Prot (66)	Prot (6)	1	100	100
Muscle 24 h	Lipid (921) + Metab (464) + Prot (66)	Lipid (11) + Metab (13) + Prot (3)	1	100	100
	Lipid (921)	Lipid (27)	1	100	100
	Metab (464)	Metab (15)	1	100	100
	Prot (66)	Prot (6)	1	100	100
Skin 3 h	Lipid (944) + Metab (384) + Prot (99)	Lipid (26) + Metab (26) + Prot (1)	1	100	100
	Lipid (944)	Lipid (18)	1	100	100
	Metab (384)	Metab (24)	1	100	100
	Prot (99)	Prot (12)	1	100	100
Skin 24 h	Lipid (944) + Metab (384) + Prot (99)	Lipid (64) + Metab (69) + Prot (14)	1	100	100
	Lipid (944)	Lipid (12)	1	100	100
	Metab (384)	Metab (34)	1	100	100
	Prot (99)	Prot (4)	1	100	100
Gills 3 h	Metab (380) + Prot (108)	Metab (59) + Prot (14)	1	100	100
	Metab (380)	Metab (69)	1	100	100
	Prot (108)	Prot (12)	1	100	100
Gills 24 h	Metab (380) + Prot (108)	Metab (7) + Prot (1)	1	100	100
	Metab (380)	Metab (14)	1	100	100
	Prot (108)	Prot (14)	1	100	100
Eye 3 h	Metab (340)	Metab (69)	1	100	100
Eye 24 h	Metab (340)	Metab (21)	1	100	100