starfish
starfish copied to clipboard
Tests for Starfish.data
We are doing a good job of testing pipelines that we are creating for the spaceTx and biological community by testing notebooks and scripts that are derived from the notebooks.
However, we have many datasets that do not yet have pipelines that may be useful as points of exploration for interested users, or that we want to expose in starfish.data to enable the groups that generated the data to build pipelines for themselves.
To ensure these datasets are updates as the format evolves, we should write tests and use our CI to ensure that all the datasets that we are supporting stay up to date.
A simple test that exercises the key API endpoints of Experiment and FieldOfView (e.g. to extract primary and auxiliary images for one or more fields of view of an experiment) and to verify the validity of the resulting data structure (an ImageStack) should be adequate for this.
cc @shanaxel42 @ttung
This is potentially quite expensive (and destructive to our travis caches). We should stand up an environment with the caches disabled and wire it up to @joshmoore's periodic tests.
That sounds great. Look forward to Josh returning from Japan!
@ttung: what type of environment are you talking about standing up here?
Sorry, not an environment, but just the setup you have for checking requirements.
Gotcha. Seems straight-forward enough. Happy to add something but I'll need the list of datasets for consideration.
This was recently discussed in #1741
Per @ttung's comment there, if we can get some lightweight tests that simply validate that the datasets (specifically, the experiment.json files) registered in starfish are available, that will be enough I think.