SI cloud app

Easy submission of spike sorting jobs locally or to the Cloud:

Keep track of your sorting jobs:

Running the app locally

Set ENV variables:

export AWS_DEFAULT_REGION=
export AWS_ACCESS_KEY_ID=
export AWS_SECRET_ACCESS_KEY=
export AWS_BATCH_JOB_QUEUE=
export AWS_BATCH_JOB_DEFINITION=
export DANDI_API_KEY=

Running with docker compose pulling images from github packages:

docker compose up

Running with docker compose building images locally (for dev, with hot reaload):

docker compose -f docker-compose-dev.yml up

If you did any changes in requirements.txt, package.json or Dockerfile, you should stop the containers and run again with an extra --build flag:

docker compose down
docker compose -f docker-compose-dev.yml up --build

Run rest api standalone (dev):

cd rest
python main.py

Run frontend standalone (dev):

cd frontend
yarn start

App components

The app is composed of four components:

• rest - the rest api, which is a FastAPI app
• frontend - the frontend, which is a React app
• db - the database, which is a Postgres database
• worker - the worker, which is a sorter container with a Flask app

Building images separately

DOCKER_BUILDKIT=1 docker build -t ghcr.io/catalystneuro/si-sorting-worker:latest -f Dockerfile.combined .
docker push ghcr.io/catalystneuro/si-sorting-worker:latest