anthonyofseattle/ginkgo_challenge

Web application for Ginkgo Bioworks' coding challenge.

Ginkgo Bioworks' Backend Coding Challenge

Welcome to the repository for Seqsleuth, a sequence search application and my submission for Ginkgo Bioworks' Backend Coding Challenge. The application is comprised of a React frontend with a Django server to handle submission of searches and querying results.

Setup

Development setup can be ran completely through docker compose with modern versions of docker. Clone the repository and run the following command to start the development server at, http://localhost:8000/.

docker compose up -d --build

This will build the images, deploy, and then build a test repository of sequences to search. Building the test repository for the first time can take a minute or so, so please run the following command to see progress. When you see 'Starting development server...', you are good to go.

docker compose logs app

The project can then be torn down (including volumes) with the command.

docker compose down -v

Overview

If you would like to learn about how this application was implemented, please read below for brief overviews of inidividual components.

1) React Frontend

The frontend application for Seqsleuth is implemented entirely as a React application which is webpacked and served to the user. The SearchApp is comprised of two components, the SearchInput and the SearchHistory. Users interact directly with the SearchInput, which validates DNA sequences and POSTs to the backend server. The SearchHistory is continuously updated to query job statuses and results as they become available. Currently, sequence input is limited to well formated DNA sequences less than or equal to 200 characters long, the latter parameter being specified via the props.

While searches in the development setup are much to fast to see the job tracking features, the SearchApp does allow as many searches to be inputed as the user desires before any of them are finished. The SearchHistory will monitor these searches and inform the user whether they have started and how long they have been running.

2) Django Backend

The backend functionality of Seqsleuth is entirely handled through REST endpoints under the /searches/ resource. When a user POSTs a sequence to /searches/, a new search is created and associated with the user session, and then a celery task is started to handle the search. Search information is returned to the user while the search runs. A GET request to /searches/ returns all searches associated with the user session, along with any results.

The actual searching is handled by the genomes module. A GenomeBank class allows querying a string through every cDNA sequence in its cache. The cache is seeded by issuing the following management command.

python manage.py buildgenomebank

Currently, the class looks for the environment variable GENOME_LIST to know which genomes to download. This variable is specified in the .env.* files.