Fig. 5

PLS-DA analysis of serum metabolomics. a Graphical representation of the goodness-of-fit. The three first components explained more than 90% and predicted more than 75% of the variance. b, c PLS-DA score plots representing the distribution of samples with component 1 vs component 2 (b), and component 1 vs component 3 (c). All infected recipient (R) fish clustered separated from control (C) fish. In addition, R fish with high intensity of infection (H) were more separated from C than R with low (L) and medium (M) infection levels. R fish with the highest infection levels are included in the rectangle. The contribution of the different metabolites to the group separation was determined by variable importance in projection (VIP) measurements after three components. d List of the metabolites increased (in red) or decreased (in green) during the infection, and their VIP (variable importance in projection) scores