thelabbingproject / django_analyses Goto Github PK

View Code? Open in Web Editor NEW

1.0 1.0 4.0 667 KB

A reusable Django app to manage analyses.

Home Page: https://django-analyses.readthedocs.io/en/latest/

License: Apache License 2.0

Python 99.98% CSS 0.02%

django-app research-data-management workflow-engine

django_analyses's People

Contributors

Stargazers

Watchers

Forkers

zvibaratz aharonyn galkepler loreggiagiacomo

django_analyses's Issues

Fix Run model's delete to remove run directory

Output directories aren't deleted for some reason with the overridden delete() method.

Hard coded media root in migration

Migration 0006 uses a hard coded path for MEDIA_ROOT.

Refactor run configuration check method to be more efficient

Currently checks default and is_configuration values even if the configuration is the same.

Parallel runs stop on failure

When any run fails during parallel execution, a RuntimeError is raised and that execution thread is stopped.

AutoField warning

Django requires a default auto field setting.

Fix `index` attribute function for `ListInput`s

Index isn't considered for some reason.

filter_queryset() no applied to provided queryset in QuerySetRunner

Add progressbar for QuerySetRunner's existing runs evaluation

Refactor analysis configuration management outside the Run model

There are currently too many methods cluttering the Run model that actually have to do more with parsing/adjusting the full configuration out of a run instance's kwargs. It could be better if they were managed separately, perhaps by some ConfigurationParser class.

Tests failure

Tests failing due to django_mri import.

Add `ListOutputDefinition` class

Should be similar to the ListInputDefinition class in terms of element type options.

CI Integration

DirectoryInput with no value raises an error

Seems like the fix_output_path() method doesn't handle the default scenario (no user provided value) and raises a TypeError by trying to Path(None).

Tests failing in CI

The Node mode's check_run_configuration_sameness() method returns True for the False test case, which doesn't happen locally and didn't happen in the past.

Add parallel execution functionality

We need to find a way to allow for parallelization of pipeline execution (e.g. running preprocessing on multiple subjects concurrently).

Create base class for QuerySet processing automation

Previous iteration's runs returned by PipelineRunner when running with different inputs

Seems to be caused by not resetting the runs dictionary.

Integrate Celerey for task management

In order for the app to scale, Celery should be integrated for task management.

Create mutually exclusive parameters configuration

Mutually exclusive meaning that certain combinations of values can be configured to be invalid. This should probably be part of the InputSpecification model.

Update documentation to reflect new QuerySetRunner output

Run instances should be deleted on KeyboardInterrupt

Currently run instances are created with no output.

get_run_set() returns an empty queryset for empty configuration nodes

Should return nodes with the appropriate configuration (defaults and no other is_configuration inputs).

Integrate isort

Add hyperlinks in admin site tables

Allow calling `query_progress()` independently of `run()`

Makes QuerySetRunner more useful.

Default QuerySetRunner Filtering docs highlighting issue

Rename QuerySetRunner's MODEL attribute to DATA_MODEL

Makes it a bit clearer IMO

Add run download button in admin

Add support for running the same node multiple times in a single pipeline

There are scenarios where the same Node instance needs to be executed multiple times within the same pipeline. The way PipelineRunner currently works doesn't really take that into account, and Pipe also needs to be updated to indicate which run of the source or destination node is referenced.