Chelombus Platform
A website where anyone can explore the billion-molecule clustering results. No login, no setup.
The algorithms behind Chelombus are interesting, but useless if nobody can access the results. So I built a public website where you can browse datasets, view interactive maps, compare molecular properties, and read the documentation. The site serves 180,000+ pre-generated interactive maps directly from disk and is deployed with Docker (Next.js for the app, Nginx for the static files).
Tech Stack
Next.jsReactTypeScriptTailwind CSSNginxDockerSphinxPython
Features
- Browse datasets with quick-facts cards and one-click access to maps.
- Interactive map viewer with property dashboards for side-by-side comparison.
- 180,000+ pre-generated maps served directly from disk via Nginx.
- Documentation with quickstart guide and large-scale examples.
- No login required. Anyone can explore the results immediately.
- Dockerized deployment: Next.js app + Nginx, easy to replicate.
Challenges
- Serving 180,000+ static HTML files without slowing down the app.
- Making a complex pipeline understandable through docs and tutorials.
- Deploying behind a reverse proxy while keeping static file delivery fast.
Learnings
- Keeping the static maps separate from the app keeps everything fast.
- No-login access is a big deal. Collaborators try it immediately instead of never.
- Good documentation is as important as the algorithm for scientific tools.