Snakefile

include: "../common.smk"
configfile: "config.yml"

# May need to get new URLs from https://cellxgene.cziscience.com/collections/edb893ee-4066-4128-9aec-5eb2b03f8287

# The single-cell lung cancer atlas (LuCA) -- extended atlas
H5AD_URL = "https://datasets.cellxgene.cziscience.com/3ab47484-a3eb-4f6a-beea-670e1a8fc1e8.h5ad"

rule all:
    input:
        [ (PROCESSED_DIR / f) for f in config['output'] ]

rule convert_to_zarr:
    input:
        (RAW_DIR / "3ab47484-a3eb-4f6a-beea-670e1a8fc1e8.h5ad")
    output:
        directory(PROCESSED_DIR / "sikkema_2023_full.h5ad.zarr")
    params:
        script=(SRC_DIR / "convert_to_zarr.py")
    shell:
        '''
        python {params.script} \
            -i {input} \
            -o {output}
        '''

# Download raw h5ad file.
rule download_adata:
    output:
        (RAW_DIR / "3ab47484-a3eb-4f6a-beea-670e1a8fc1e8.h5ad")
    params:
        file_url=H5AD_URL
    shell:
        '''
        curl -L --retry 999 --retry-delay 3 -C - -o {output} "{params.file_url}"
        '''