Statistical analysis and metabolic pathway analysis
XCMS software (Scripps Research Institute, La Jolla, CA, USA) was used to extract and process the characteristic peaks of the original data obtained by LC-MS, the extracted substances were added, and the ions were annotated using a camera. Through principal component analysis, the dimension of the data was reduced, and the overall distribution trend was previewed. Orthogonal partial least squares discriminant analysis was used for modeling the data. Based on a minimum fold change (≥2), analysis of variance p-value (≤0.05), and variable projection importance value calculated in the orthogonal projections to latent structures discriminant analysis (OPLS-DA) model (≥1), the differential metabolites were screened. Candidate markers were annotated with metabolites using the HMDB, KEGG, and other databases. The pathways of different metabolites were further screened by enrichment analysis and topological analysis, and the key pathway with the highest correlation with the metabolite differences was found.