Known issues: preprocessing hiccups when run via jupyterhub

[zktuong]

Resurfacing this just in case

Discussed in https://github.com/zktuong/dandelion/discussions/146

^{Originally posted by zktuong April 29, 2021}
So there's a couple of issues that may happen when running via a notebook initialized by jupyterhub if following the default tutorial:

python scripts like AssignGenes.py may complain that the file cannot be found and environmental variables set in ~/.bash_profile are not seen.

This can be fixed by doing this:

Open a terminal and use jupyter kernelspec list to find the folder for your kernel.

Available kernels:
  dandelion    /home/jovyan/.local/share/jupyter/kernels/dandelion

Go there and edit kernel.json, it should look something like this:

cd /home/jovyan/.local/share/jupyter/kernels/dandelion
vi kernel.json

{
 "argv": [
  "/home/jovyan/my-conda-envs/dandelion/bin/python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "{connection_file}"
 ],
 "display_name": "Python (dandelion)",
 "language": "python"
}

Now edit that file and add an "env" key to that json. and add a dict with a "PATH" key to the binaries, as well as the other environmental variables, in your conda env like this:

{
 "argv": [
  "/home/jovyan/my-conda-envs/dandelion/bin/python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "{connection_file}"
 ],
 "env": {
     "PATH": "/home/jovyan/my-conda-envs/dandelion/bin:$PATH",
     "GERMLINE": "/home/jovyan/Softwares/database/germlines/",
     "IGDATA": "/home/jovyan/Softwares/database/igblast/",
     "BLASTDB": "/home/jovyan/Softwares/database/blast/"
 } ,
 "display_name": "Python (dandelion)",
 "language": "python"
}

Restart the instance/kernel and it should start working.