Cellbender (https://cellbender.readthedocs.io/en/latest/index.html) can be accessed on the Hoffman2 Cluster in several ways:
Run as a container
Get an interactive session on a GPU node, with, for example:
qrsh -l gpu,RTX2080Ti,cuda=1,h_data=20G
Load apptainer in your environment and start the Cellbender container:
module load apptainer
apptainer shell --nv $H2_CONTAINER_LOC/cellbender.sif
To verify that cellbender works, at the Apptainer shell prompt (Apptainer>) enter:
cellbender --help
Note, if a newer than what available in /u/local/apps/apptainer/H2_containers is desired it can be downloaded via:
apptainer pull cellblender.sif docker://'us.gcr.io/broad-dsde-methods/cellbender:latest
Run as python environment
Get an interactive session on a GPU node, with, for example:
qrsh -l gpu,RTX2080Ti,cuda=1,h_data=20G
Load python in your environment and create a virtual environment:
module load python
module load gcc/11.3.0
module load cuda/10.7
python -m venv ~/.venv/cellbender
source ~/.venv/cellbender/bin/activate
python -m pip install --upgrade pip
pip install cellbender
# as of 2024JULY09 numpy needs to be reinstalled as:
pip install "numpy<2"
To verify that cellbender works, at the shell prompt enter:
cellbender --help
Note: the installation procedure via pip in not guaranteed to work, a requirements file that works as of 2024JULY09 can be found here:
/u/local/apps/VIRT_ENV_REQS/cellbender_requirements.txt
which can be used as follows:
module load python
module load gcc/11.3.0
module load cuda/11.7
python -m venv ~/.venv/cellbender
source ~/.venv/cellbender/bin/activate
python -m pip install --upgrade pip
python -m pip install -r cellbender_requirements.txt