Abstract:Study: Identification of metabolomics signatures for patient stratification between type 1 versus type 2 diabetes mellitus. Dataset: plasma samples from 69 diabetic patients were analyzed by reversed phase (C18) ultra-high performance liquid chromatography (UPLC) coupled to high-resolution mass spectrometry (Orbitrap Exactive). Age and body mass index (BMI) of patients are also provided as sample metadata since type 2 patients from this cohort are significantly older and with a higher BMI. The peak table contains 5,501 features (whose m/z and retention time have been matched against an in-house database). The intensities from the peak table have been log10 transformed. Workflow: The workflow consists of: PCA visualization (with the two diabetic types coloured on the score plot), univariate Wilcoxon hypothesis testing of median differences between diabetic types, OPLS-DA modeling of the diabetic type response, and selection of the metabolite significant signature for gender classification by PLS-DA, Random Forest or SVM. Comments: The ‘diaplasma’ data set is also available in the biosigner R package from the Bioconductor repository.