I assume that DXJupyterLab is a custom containerised image of Jupyterlab. We would like to have available as part of the jupyterlab some internal proprietary packages, access to some internal data through those packages, and also access to some custom already pre-built environments so we don't waste time again and again when we open up a new jupyterlab. The solution I see is for us to be able to run our own custom jupyterlab images on DNANexus. Is that possible? Can someone point me in the right direction? Is there any other potential solution?
I know technically this is achievable by manually customising a DXJupyterLab instance and then saving a snapshot. However, our internal packages and environments evolve over time and we already have automated workflows in place to generate new images for our internal docker server which from engineering standpoint seems the right thing to do here.
I believe that JupyterLab app and its codebase on dnanexus/rap is not public and cannot be downloaded using "dx get" command.
My personal idea for customizing and rebuilding a JL app/let, would be to use the following github repo, developed by lindgrengroup: https://github.com/lindgrengroup/hail-on-dnanexus The team has developed the custom version of Spark based JL with latest Hail.
The repo could serve as a template to try building anything you want to do with Jupyter - of course, it will need some development and testing. Please note that this is a Spark version. For nonSpark, you will need to change the folder/file structure appropriately.
Comments
1 comment
Hi @Cristian Regep?,
I believe that JupyterLab app and its codebase on dnanexus/rap is not public and cannot be downloaded using "dx get" command.
My personal idea for customizing and rebuilding a JL app/let, would be to use the following github repo, developed by lindgrengroup: https://github.com/lindgrengroup/hail-on-dnanexus The team has developed the custom version of Spark based JL with latest Hail.
The repo could serve as a template to try building anything you want to do with Jupyter - of course, it will need some development and testing. Please note that this is a Spark version. For nonSpark, you will need to change the folder/file structure appropriately.
Please sign in to leave a comment.