Run setup.sh to set default paths
source setup.sh
To obtain container locally (if not already downloaded) (Note for covert lab: we already have it on OAK)
singularity pull docker://scr.svc.stanford.edu/khoang99/containers/scrna:latest
# set RUN environment variable to execute singularity run
export RUN="singularity run scrna_latest.sif"
To run Jupyter Notebook with container, first run jupyter server on your compute node (Sherlock), ideally in a new screen (e.g., screen -S notebook)
# port can be set to any number from 20000 to 30000, pick a random one to not overlap with other users
$RUN jupyter notebook --port=23232 --no-browser --ip=127.0.0.1
# If you want jupyter lab instead
$RUN jupyter lab --port=23232 --no-browser --ip=127.0.0.1
Now, on your local computer, login to Sherlock with ssh forwarding
# modify port number according to what you chose above
ssh [USER]@login.sherlock.stanford.edu -L 23232:localhost:23232
# When inside login node, ssh port forward to compute node (modify sh02-02n26 according to compute node you start the jupyter server)
ssh -N -L 23232:localhost:23232 sh02-02n26
On local machine, paste http://localhost:23232/ (adjust port accordingly) into your browser, enter Sherlock password or the token string (if the displayed URL is https://127.0.0.1:23232/?token=185779e2c2af5e1a4b33380e657e9ada64e45d6c70f05da9 then enter 185779e2c2af5e1a4b33380e657e9ada64e45d6c70f05da9) to login Jupyter notebook