container_nnunet.def

# For more information on container definition files,
# visit https://sylabs.io/guides/3.7/user-guide/definition_files.html

Bootstrap: docker
# this basically provides pytorch and OS
From: nvcr.io/nvidia/pytorch:20.08-py3

%files
    # copy prediction script
    ./predict_nnunet.py /code/predict.py

    # Copy model files (including model weights and metadata files, test script)
    # When trying this example, please make sure you installed nnunet (https://github.com/MIC-DKFZ/nnUNet#installation)
    # Pretrained models can be downloaded via nnunet (https://github.com/MIC-DKFZ/nnUNet#how-to-run-inference-with-pretrained-models);
    # the BraTS 2020 models are only available when installing nnunet from the master branch.
    $RESULTS_FOLDER/nnUNet/3d_fullres/Task082_BraTS2020/nnUNetTrainerV2BraTSRegions_DA4_BN__nnUNetPlansv2.1_bs5 /params/nnUNetTrainerV2BraTSRegions_DA4_BN__nnUNetPlansv2.1_bs5
    $RESULTS_FOLDER/nnUNet/3d_fullres/Task082_BraTS2020/nnUNetTrainerV2BraTSRegions_DA4_BN_BD__nnUNetPlansv2.1_bs5 /params/nnUNetTrainerV2BraTSRegions_DA4_BN_BD__nnUNetPlansv2.1_bs5
    $RESULTS_FOLDER/nnUNet/3d_fullres/Task082_BraTS2020/nnUNetTrainerV2BraTSRegions_DA3_BN_BD__nnUNetPlansv2.1_bs5 /params/nnUNetTrainerV2BraTSRegions_DA3_BN_BD__nnUNetPlansv2.1_bs5

%post
    # Commands in this block are executed after the files have been copied in %files
    # Install ubuntu packages
    apt update && apt upgrade -y
    apt install -y git
    # Install your python requirements
    python -m pip install git+https://github.com/MIC-DKFZ/nnUNet.git

%runscript
    # This block is executed inside the container when it is run
    # The following arguments will be passed:
    # -i: path to input folder
    # -o: path to output folder

    # special for nnunet
    # [ignore; if container is run with access to host environment, this causes an error]
    unset nnUNet_raw_data_base
    unset nnUNet_preprocessed
    unset RESULTS_FOLDER

    # run the inference script
    echo "Starting prediction..."
    python /code/predict.py "$@" -p /params

%labels
    # Please add these fields
    Author "Maximilian Zenk"
    Team "Organizer"
    Institution "German cancer research center (DKFZ)"
    Email "m.zenk@dkfz-heidelberg.de"

%help
    # optional
    This singularity container predicts segmentations on the images in the input folder (first argument)
    and saves them to the output folder (second argument). It uses a pretrained nnUNet model.