Metabolomics Analysis of Tgfb3 Mutant Mice

Univariate (UVA) and multivariate (MVA) analyses were performed for LC-MS and GC-MS analyses, as previously described (Wang et al., 2011 (link)). In univariate analysis, a non-parametric method through Mann–Whitney U test was applied. P-values were obtained for each feature in UVA. MVA was performed with SIMCA-P 15.0 (Umetrics). An orthogonal partial least squares discriminant analysis (OPLS-DA) model was built between Tgfb3^+/+ and Tgfb3^+/−, then the variable influence of projection (VIP) used to select the significant features obtained in the comparison between groups was obtained for each feature. Metabolites with P≤0.05 and/or VIP>1 were selected as significant.