🎉 Featured in the marimo official gallery! 🚀
Interactive chemical space viewer built with marimo.
Explore chemical space on a t-SNE scatter plot, tune embedding and clustering parameters, and inspect selected molecules in a reactive table.
- 2000 NCI molecules from RDKit built-in dataset
- Morgan fingerprints (ECFP4) + Tanimoto distance matrix
- scikit-learn TSNE 2D embedding with precomputed distances
- HDBSCAN density-based clustering
- Reactive parameter controls for t-SNE and HDBSCAN (
mo.ui.number) - mo.ui.matplotlib interactive box / lasso selection
- Inline SVG structures in the property table via
format_mapping - Noise filtering (
cluster = -1) for selected rows
No install needed — just run with PEP 723 inline metadata:
uvx marimo edit ./app.py --sandboxOr with a local virtual environment:
uv sync
uv run marimo edit app.py