← work

Bioinformatics · exploratory research

Exploring protein relationships through embeddings

GitHub repository

An exploratory project for collecting protein information, building embeddings, and examining how proteins organize in representation space.

The repository includes UniProt data collection, multiple embedding experiments, and clustering utilities using K-means, DBSCAN, and HDBSCAN. Dimensionality-reduction plots provide a way to inspect the resulting groups.

Research notes outline comparisons against known protein-interaction resources. The current showcase focuses on the data and exploration pipeline: clustering similarity alone does not establish a biological interaction, and a validated interaction-prediction result is not claimed.