Integration of NMR and LC-MS metabolomics datasets

Datasets used in the protocol "Integration of NMR and LC-MS metabolomics datasets". These datasets are part of the AddNeuroMed study (Lovestone et al. 2009), a dementia research cohort which contains NMR and LC-MS data. The data provideded are: A metadata file, containing the patients age, gender and diagnostic; Two NMR datasets: 1) NMR dataset corresponding to 1H-NMR Carr-Purcell-Meiboom-Gill (CPMG) spectra obtained with water presaturation that have been preprocessed using nPYc Python package (Sands et al. 2019); and 2) reduced feature NMR dataset (227 features) was also prepared by peak-picking individual peaks of the 1H-NMR CPMG full resolution data followed by peak grouping, using the ‘speaq’ R package (Beirnaert et al. 2018). Three LC-MS serum assays consisting of reversed-phase chromatography tailored for the separation of lipophilic analytes to produce lipid positive (LPOS) and negative (LNEG) datasets; and a hydrophilic interaction liquid chromatography (HILIC) assay in positive mode for separation of hydrophilic analytes (HPOS). The LC-MS raw data were converted to mzML format, peaks detected using XCMS (Smith et al. 2006) and preprocessing and quality control steps performed using the nPYc toolbox software. Final datasets contained 681, 4886, and 2091 features for HPOS, LPOS, and LNEG respectively. Any possible missing values resulting from the XCMS processing were replaced by zero. A total of 573 samples were available: 158 from Alzheimer’s disease (AD) patients, 254 from mild-cognitive impairment (MCI) patients and 161 from healthy controls (HC). All the matrices contain samples from all patients sorted in the same order with no missing values. All 6 data files are saved as a .csv files and stored in a single .zip file.