P-CASSO: A compartment-resolved pan-cancer spatial atlas of intercellular interactions and prognostic ligand-receptor axes

This record contains the processed single-cell and spatial transcriptomicsdata underlying P-CASSO, a compartment-resolved pan-cancer spatial atlas ofintercellular interactions and prognostic ligand–receptor axes. ## Files **_sc_ALL_260730_h5ad.tar.gz** Single-cell RNA-seq reference atlas.- 2,183,305 cells × 14,213 genes, from 686 samples of 491 patients- 13 cancer types: BRCA, LUCA, OVCA, STCA, PACA, THCA, HNCA, PRCA, LICA, COCA, KICA, SKCA, UECA- `X` — log-normalized expression; `layers['counts']` — raw UMI counts- `obs` — Cancer_type, Organ_orig, Patient, Sample, Platform, Kit_version, Subtype, Tissue_group, Tissue_site, global_anno_level1/level2, sub_anno, and per-cell QC metrics **_vis_ALL_260730_h5ad.tar.gz** 10x Visium spatial transcriptomics data.- 635,720 spots × 10,862 genes, from 268 Visium sections- `X` — log-normalized expression; `layers['counts']` — raw counts- `obs` — cancer_type, sample_id, array_row/array_col, cnv_score, and per-spot cell-type abundance scores * 7 major-level columns, suffixed `_enriched` (Epithelial, T cell, B cell, Myeloid, Endothelial, Fibroblast, Mural) * 93 fine-grained cell-type columns (e.g. CD8 Tex, SPP1+_Macro, mCAF) - `obsm` — spatial (spot coordinates), X_pca, X_umap, and cell2location deconvolution outputs (means/q05/q95/stds_cell_abundance_w_sf)- `uns['spatial']` — H&E histology images and scale factors for all 268 sections- `.obs['Location']` — spatial compartment per spot: `Mal` (malignant), `Bdy` (boundary), `Normal` **_vis_obs_proportion_csv.gz**Cell-type proportions per Visium spot, derived from the cell2locationdeconvolution results. - 635,720 spots (rows) × 100 cell types (columns)- Two annotation levels are stored side by side, each normalized to sum to 1 independently: * 7 major-level columns, suffixed `_enriched` (Epithelial, T cell, B cell, Myeloid, Endothelial, Fibroblast, Mural) * 93 fine-grained cell-type columns (e.g. CD8 Tex, SPP1+_Macro, mCAF) Each row therefore sums to 2.0 across all 100 columns — select one level ## Note Disk space required after extraction: ~114 GB (single-cell) and ~68 GB(spatial).