NicheDeSig: Niche-aware Deconvolution and Adaptive Signature Analysis for Spatial Transcriptomics

This directory is the data package for the NicheDeSig. This data package contains the following data: - `source_data/simulated_benchmark_32/`: the 32 simulated benchmark datasets used for the deconvolution comparison. Each dataset contains `Spatial.h5ad`, `scRNA.h5ad`, `sort_scRNA.h5ad`, and `pro_real.csv` when provided by the upstream benchmark. Only `dataset1` through `dataset32` are included, matching the revised benchmark. - `source_data/simulated_colon_pseudost/`: simulated colon pseudo-ST data (`ST-colon1` through `ST-colon4`) and the colon scRNA reference used for the colon spatial-fidelity benchmark and ablation analysis. - `source_data/dlpfc_maynard_8_slices/`: DLPFC Visium slices 151669-151676, layer annotations, and `DLPFC_scRNA.h5ad`, used for the annotated-niche DLPFC analysis. - `source_data/brca_10x_visium/`: 10x Visium human breast cancer spatial input, spatial images/coordinates, metadata, and scRNA reference used for the unannotated/global adaptive-signature BRCA analysis. - `source_data/gse111672_pdac/`: raw and processed GSE111672 PDAC files used for the public PDAC-A validation and region-AUC analysis. - `source_data/gse225857_crlm/`: raw and processed GSE225857 CRLM files, H&E overlay QC outputs, effective-neighborhood labels, and the selected `selected_v1_T1p4` NicheDeSig analysis outputs.