How to download database?

[ryao-mdanderson]

Dear CellphoneDB authors,

The README "Installing CellphoneDB " section step 5. download the database, Follow this tutorial,
the link is 404 page not found. Please advise how to download database.

Thank you.

[datasome]

Hi ryao-mdanderson,

Thank you for spotting a typo - https://github.com/ventolab/CellphoneDB/blob/master/README.md is now fixed. The tutorial you need is in: https://github.com/ventolab/CellphoneDB/blob/master/notebooks/T0_DownloadDB.ipynb

Best,

Robert.

[ryao-mdanderson]

Hello Robert @datasome

Thank you for your fix. I am not familiar with Jupyter notebook, in order to use CellphoneDB, should I have to use the installation instruction 4. Set up the kernel for the Jupyter notebooks? Should I have to download database using notebooks/T0_DownloadDB.ipynb?

I am an HPC system admin, I plan to set up a module for cluster users. The Jupyter notebooks approach seems to be a personal space, I wonder if this is suitable for cluster setting? Thank you for your suggestion.

Best,
Rong

[datasome]

Hi Rong,

The answer to both questions is "yes". Personally, I don't have experience of running Jupyter notebooks in a HPC environment, but it's certainly possible and done frequently - here are a few example links I found:

• https://ask.cyberinfrastructure.org/t/how-can-i-run-jupyter-notebooks-on-my-institutions-hpc-cluster/74
• https://radcamp.github.io/Lisbon2020/Jupyter_Notebook_HPC.html
• https://ulhpc-tutorials.readthedocs.io/en/latest/python/advanced/jupyter/

In our institute we have something called JupyterHub, I understand it allows users to run Jupyter notebooks on a remote cluster/VM from the comfort of their browser. This to me seems to be an ideal set up for the user - that allows for easy tweaking of parameters and re-running analyses without having to worry about HPC job management layer.

@prete, is there any high level guidance you could provide on the use of Jupyter notebooks in an HPC environment?

Best,

Robert.

[ryao-mdanderson]

Hi Robert @datasome,

I appreciate your answers and providing me links. This is helpful.

We do have Jupyter notebook (multiple versions as module) set up on HPC; Our cluster users are able to launch Jupyter.

For CellphoneDB, I use a service account to install the conda environment 'cpdb-5.0.0'.

From instruction, 4. Set up the kernel for the Jupyter notebooks.
1)Install the ipython kernel: pip install -U ipykernel.
(cpdb-5.0.0) [ris_hpc_apps@ldragon4 ~]$ pip install -U ipykernel
2) Add the environment as a jupyter kernel: python -m ipykernel install --user --name 'cpdb'.
(cpdb-5.0.0) [ris_hpc_apps@ldragon4 ~]$ python -m ipykernel install --user --name 'cpdb'

I notice the command installed kernelspec cpdb in /risapps/ris_hpc_apps/.local/share/jupyter/kernels/cpdb
which is in the service account .local folder.

3. Open/Start Jupyter and select the created kernel.
   I am puzzled in this step, how cluster user to do this step with the cpdb jupyter kernel installed in other accoount .local?
   Is the application desired for a single user mode? i.e. install conda environment and Jupyter kernel in user space?

Thank you!
Rong

[datasome]

Hi Rong,

The default would be that the user installs the kernel for themselves, but I suspect there are ways of users using shared kernels - I just don't have experience of setting that up. Perhaps the following links might help:

• https://groups.google.com/g/jupyter/c/bOGfNogIzrM?pli=1
• https://www.carc.usc.edu/user-information/user-guides/software-and-programming/jupyter-kernels

Sorry I could not be of more help.

Best,

Robert.

[ryao-mdanderson]

Dear Robert @datasome ,

No worries, your help is already much appreciated!

Best,
Rong