Figure 8. Identification of the expression of sPLA₂-IIA using its fecal lipid signature.

Machine learning was used to generate a fecal lipid signature able to distinguish WT and sPLA₂-IIA^TGN mice independently of their housing facility and sex (n = 15–16 nonarthritic WT and sPLA₂-IIA^TGN mice housed in either the SPF or Elite animal facility for 8 or 14 months, respectively). (A) Visualization of the data distribution using the identified lipids by PCA with 99% confidence ellipses to confirm the discrimination between the groups. (B) Heatmap of the Z scores — i.e., the number of SD above or below the mean, calculated from the concentration of the lipids. (C) Concentration of the 8 identified lipid metabolites in fecal samples. DAG, diacylglycerol; TAG, triacylglycerol; FA, fatty acid. Data are presented as boxes representing the median and quartiles, with whiskers extending up to 1.5 interquartile range. Statistical analysis included unpaired t test. **P < 0.01, ***P < 0.001.