CMRCNet

1. Download datasets from https://huggingface.co/datasets/MahmoodLab/hest.
   Please refer to the instructions in HEST-1k to download the HEST-1k benchmark datasets.
   Or you can download this data manually from the link: https://huggingface.co/datasets/MahmoodLab/hest
   We provide sample IDs for the six datasets we used:
   COAD: TENX111, TENX147, TENX148, TENX149
   IDC: NCBI783, NCBI785, TENX95, TENX99
   LYMPH_IDC: NCBI681, NCBI682, NCBI683, NCBI684
   READ: ZEN36, ZEN40, ZEN48, ZEN49
   SKCM: TENX115, TENX117
   PSC: NCBI672, NCBI673, NCBI674, NCBI675

##Dataset

CMRCNet/
├── data/
│   ├── COAD/
        ├──patches/
        ├──adata/
│   └── READ/

3. Prepare spatial transcriptomics data: python utils_data/pre_process.py
4. Split training sets and test sets: python utils_data/splits.py

##Dataset_prepared

CMRCNet/
├── data/
│   ├── COAD/
        ├──patches/
        ├──adata/
        ├──adata_HVGs/
        ├──splits/
│   └── READ/

6. train: python train.py
7. inference: python inference.py

This work is based on BLEEP, MCFN and HEST-1k.