UniFORM: Towards Universal ImmunoFluorescence Normalization for Multiplex Tissue Imaging

Dataset Title: PRAD-CyCIF: Prostate Cancer Single-Cell Mean Fluorescence Intensity Dataset from Cyclic Immunofluorescence (CyCIF) Description: This dataset, PRAD-CyCIF, is a curated subset of a larger, unpublished dataset. It is one of two datasets used in the manuscript "UniFORM: Towards Universal ImmunoFluorescence Normalization for Multiplex Tissue Imaging" to validate and evaluate the novel UniFORM normalization algorithm, benchmarking it against commonly used normalization methods. The dataset represents single-cell intensity feature data extracted from prostate cancer tissue samples imaged using the CyCIF platform. The dataset includes cell-level mean intensity features for 20 markers across 20 patient samples imaged in 7 different batches, along with normalized feature data (by UniFORM algorithm) done by us. Each patient sample consists of one or more imaging scenes, and individual cell-level data are annotated with metadata linking cells to their respective patient samples, scenes, and imaging batches. The data is structured as an AnnData object and contains the following attributes: 1. Expression Matrix (prad_adata.X): A matrix of shape (1,760,530 cells, 20 markers). Each row corresponds to a single cell, and each column corresponds to a marker. Contains raw feature data representing mean marker intensity for individual cells. 2. Normalized Data (prad_adata.layers["normalized"]): A layer containing normalized feature data for the same 20 markers. Allows for comparative analyses between raw and normalized data. 3. Cell Metadata (prad_adata.obs): cell_id: Unique identifier for each cell (for each scene). sample_id: Patient sample ID from which the cell was extracted (e.g., PRAD-01 to PRAD-20). scene_id: The specific imaging scene from which the cell was extracted (e.g., scene000). batch_id: Batch ID for the imaging experiment (e.g., batch1 to batch7). x, y: Coordinates of the cell within the imaging scene. 4. Marker Information (prad_adata.var): marker_name: Names of the 20 markers included in the dataset, such as DAPI, EPCAM, CD45, and Ki67. 5. Unstructured Metadata (prad_adata.uns): image_dimensions: A dictionary providing the height and width (in pixels) of the imaging scene for each patient sample and scene. Intended Use: This dataset is intended for researchers to replicate results shown in our manuscript and further benchmarking and method development. Acknowledgments: Please cite the dataset and related publications appropriately when using this data in your research.