Schedule a Notebook Job

A common scenario encountered by computational analysts is that a particular step in the analysis might be computationally intensive while the other steps are not. For instance, this might be a clustering algorithm or some other scientifically consuming procedure.

With Notebook Jobs, you use a regular notebook to execute the non-intense task interactively, and you schedule an asynchronous job to execute the computationally demanding tasks. Once this job completes, you can return to the notebook to look at the results produced by the job and derive further insights. For instance, you may look at the results of a clustering task executed as a job, and generate visualizations. 

To start a notebook job, look at one of the examples in the Jupyter Notebook examples folder. At present there are examples for scheduling an h-sne clustering run as well as running methods from the Seurat library.

Once you have scheduled a notebook job, you can check the status of the job in the Notebook Jobs view. If a job failed, a report will give detailed status. For a successful job, you can also find the output files in the view and use the file reference to load the data in a different notebook, if desired. You can also Clone the original IDE to get back to the original state, if so desired. 

Note that you can also use the Notebook Jobs view to cancel any jobs after you have scheduled them.