CellPick: strategic cell selection and spatial annotation for spatial proteomics — processed datasets

This record contains the processed imaging and proteomics datasets used in the three case studies presented in Miranda*, Pellizzoni* et al., "CellPick: strategic cell selection and spatial annotation for spatial proteomics". Three datasets are included: Ovarian tissue (STIC). Processed multiplexed immunofluorescence images of FFPE fallopian tube tissue sections, including regions of serous tubal intraepithelial carcinoma (STIC) and adjacent normal epithelium, originally generated by Makhmut et al. (Molecular Systems Biology, 2025). Staining: p53, Ki67, PAX8. Cell segmentation masks generated using Cellpose with default parameters. Pancreatic tissue. Multiplexed immunofluorescence images of FFPE human pancreatic tissue sections generated for this study. Staining: EpCAM, Insulin (Alexa Fluor 488), Glucagon, Somatostatin, Hoechst. Five-channel whole-slide images acquired on a Zeiss AxioScan Z7 at 20× magnification. Cell segmentation and classification into alpha, beta, delta, and unclassified cells performed using Cellpose and a pre-trained random forest classifier via scPortrait. Liver tissue. Processed multiplexed immunofluorescence images of frozen human liver tissue sections from a cohort of 18 individuals, originally generated by Weiss et al. (Nature Metabolism, 2026). Staining: phalloidin, DAPI, glutamine synthetase (GS), argininosuccinate synthetase 1 (ASS1). Cell segmentation masks generated using a custom-trained Cellpose model. The datasets are designed to cover CellPick's most common use cases and input types. The ovary and liver datasets are provided as raw images with annotated masks in either PNG or XML format. The Pancreatic dataset is provided as a SpatialData object, and includes cell segmentation masks, channel images, and cell type labels, to ensure reproducibility of the analyses presented in the manuscript. Data structure: Figure_2_1_Confined/└── Ovary/ ├── image0_ch0_crop2.tif # Channel 0 (DAPI), cropped region of interest ├── image0_ch1_crop2.tif # Channel 1 (PAX8), cropped region of interest ├── image0_ch2_crop2.tif # Channel 2 (p53), cropped region of interest ├── image0_ch3_crop2.tif # Channel 3 (Ki67), cropped region of interest └── image0_crop2_cp_masks.png # Cellpose segmentation masks Figure_2_2_Fairness/└── Pancreas/ └── CellPick_2_2_pancreas.sdata.zip # Full SpatialData object including # five-channel images, segmentation masks, # and cell type classifications Figure_2_3_Gradient/└── Liver/ ├── CellPick_2_3_ASS1.tif # ASS1 channel (periportal marker) ├── CellPick_2_3_DAPI.tif # DAPI channel (nuclei) ├── CellPick_2_3_GS.tif # GS channel (pericentral marker) ├── CellPick_2_3_Membrane.tif # Phalloidin channel (cell borders) ├── CellPick_2_3_shapes.meta # Cell shape metadata └── CellPick_2_3_shapes.xml # Cell segmentation shapes (LMD-compatible XML)