Building a biological database with Python
summary
A community-driven database for high-throughput protein variant experiments stores sequence-to-function scores, uses Postgres with extensive JSON fields, and runs background jobs on RabbitMQ/Celery; the first Django-backed version reaches production within a year, attracts unexpected clinical use, and forces heavy curation. Limited funding and a tiny team prompt a reimplementation — a pragmatic Django to FastAPI migration in Docker Compose — to deliver customizable APIs, preserve the same SQL backend, and make automated Python SDK uploads and downloads practical.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 7:24Biological database need
- 10:13Starting with Django
- 13:40Technical challenges setup
- 15:46First deployment success
- 17:10Clinical data focus
- 18:34Django 1.11 end-of-life
- 20:18Rebuilding with FastAPI
- 22:29Lessons learned future
- 24:53Q&A
related talks
30mBuilding modern Python web apps with PostgreSQL
31mSpace Django: Migrating and Redesigning a Database while Hunting for Satellites.
02h 46mData analysis with SQLite and Python
25mProcessing Biggish Data with DuckDB and Python
24mBuild-a-Database with Python
48mAuto-generating an API using PostgreSQL, Django, and Django REST Framework
37mAccelerating Python Analytics by In-Database Processing
39mPython and SQL: Better Together, Powered by DuckDB