A dataset of spatially resolved metabolomics on human kidney anatomic regions
Description
Kidney cortex, medulla and papilla were dissected from the same donor and one 10-µm section for each tissue was selected for MALDI-IMS analysis. Data acquisition was performed with the positive ion mode at a pixel size of 10 × 10 µm2 over m/z range 50 – 1500. The matrix used was 2,5-DHB in 90% acetonitrile with 0.1% trifluoroacetic acid. Matrix application was performed with the HTX M5 Sprayer and data acquisition was performed on the timsTOF fleX MALDI-2 instrument. Data were uploaded to METASPACE for annotation with the CoreMetabolome database (https://metaspace2020.eu/project/human_kidney_region). Downstream analysis was performed with our recently described package MALDIpy. For each sample, the raw matrix of spatially resolved metabolomics data, with false discovery rate (FDR) < 20% and with total ion count (TIC) normalization, is presented (compressed in a .zip file), where each row indicates a feature (annotation with the CoreMetabolome database v3) and each column indicates a x/y coordinate pair. Researchers can also export customized data matrices (e.g., choosing different FDR thresholds or no TIC normalization) through the METASPACE browser with the link provided above. Feature annotations are presented in a spreadsheet (.xlsx file), which describes metabolite species detected in each sample, including their m/z ratios, chemical formulas, ion adducts and molecule identities. Table 1 summarizes the number of detected features in each MALDI-IMS sample. A raw Anndata (.h5ad file) for merged three MALDI-IMS data matrices is presented, containing a total of 370,662 metabolomes (i.e., 10 × 10 µm2 pixels) and 562 features. Cortex, medulla and papilla samples contain 177,870, 92,120 and 100,672 metabolomes, respectively. This raw Anndata contains significant matrix background and low-quality metabolomes which can be removed through standard quality control pipeline. A processed Anndata (.h5ad file) is presented, containing a total of 170,459 metabolomes and 562 features. Cortex, medulla and papilla samples have 79,814, 49,771 and 40,874 metabolomes included in this processed data, respectively. All metabolomes in this processed file have a minimum count number of 40,000 and a minimum feature number of 30 detected. Other quality control procedures, including removal of matrix background observations and low-quality metabolomes, were performed. MALDIpy-based downstream data processing, multi-sample integration and dimensional reduction were also performed to generate this data. A total of 16 clusters were identified in Leiden clustering analysis. Cluster identities and anatomic region origins are included in the Anndata.obs columns.