The docker file uses postgres:latest and builds the pgvector extension in place.
The extension is automatically activated using an init.sql script.

Use docker compose to run the image.

This project uses poetry to manage python dependencies. Run poetry install and poetry shell to get a local virtual environment and activate it.

Several examples of using pgvector are in the examples folder.

Assume we have a db of molecules in a file molecules.csv - one SMILES per row.

To run the example:

docker compose up
poetry install
poetry shell
cd examples
python chem-feature.py <query_molecules> molecules.csv

The query molecule is a SMILES string, e.g. c1ccco1.