Feverbase

An open platform to accelerate any and all research related to clinical trials (including but not limited to the efficacy of repurposed drugs) for fighting COVID-19.

Dependencies

Feverbase depends on:

• MeiliSearch to index search results
• MongoDB to persist data

However, Docker takes care of setting up all required services and connecting them together!

You may also optionally create a .env file with the following keys:

  SLACK_WEBHOOK_URL=  # for feedback form
  GOOGLE_MAPS_KEY=    # for fetching locations
  MEILI_KEY=          # for a protected MeiliSearch instance

Running the App

Feverbase has two main components: serve.py which serves the Flask app which contains the actual interface to search and filter clinical trials. This can be run with python serve.py with the optional argument of --port <port> to manually specify a port.

The second component is the web scraper. By default this will scrape trials from every clinicial trial registry we have added support for. This behavior can be changed in fetch/__init__.py, specifically with the DRIPPING_FAUCETS array. The scraper can be run with python fetch.py. This may take a while to run if you're scraping every search query from every registry. We run this as a cron job every hour.

With Docker installed, you can run everything in one line:

docker-compose up

If you want to fetch results, run the above command in one terminal session. From another terminal session in the same folder, run the following:

docker-compose exec app python fetch.py

Running tests

With docker running in one terminal session:

docker-compose up

Execute the following from another terminal session in the same folder:

docker-compose exec app pytest

This will execute the tests in the container which is already running!