Metabolomics-POS

We converted the original data into mzXML format using ProteoWizard software. We used an internal program for processing, developed by R for peak extraction, calibration and integration based on XCMS. We further processed the data matrix by removing more than 50% of the peaks of missing values in the samples. Using half of the minimum of the simulation method to simulate and fill in some missing values of the original data. The new standardized data matrix was produced by the internal standard (IS). And an in-house MS2 database (BiotreeDB), KEGG(www.genome.jp/kegg) and online database of HMDB (www.hmdb.ca) were applied to metabolite annotation. The cutoff value for comments was 0.3. In order to compare the differences of metabolomic characteristics between healthy people and IgAN patients, we used multivariate statistical analysis, including orthogonal projections to latent structures-discriminant analysis (OPLS-DA) and the method of principal component analysis (PCA).