Bioactivity Dataset · Leiden Academic Centre for Drug Research

Papyrus

Curated, standardised ligand–protein bioactivity data assembled for QSAR and proteochemometric modelling.

Use it with the research agent Project page Paper

What Papyrus is

Public bioactivity data is spread across sources that disagree about units, stereochemistry and what counts as the same measurement, and most of the work in a bioactivity model is reconciling them. Papyrus does that reconciliation once: activities are standardised to a common scale, molecules are normalised with and without stereochemistry, protein targets carry sequence-derived descriptors, and each release is pinned to a ChEMBL version so a model trained on it can be reproduced. The latest release, 05.7 (Zenodo version 2024.2), is built on ChEMBL 34. Precomputed molecular and protein descriptor sets ship beside the activity tables, which is what makes it usable without rebuilding a featurisation pipeline. The authors' Python package downloads and caches a chosen release.

Get the data

Its authors publish Papyrus outside the Hugging Face Hub, so the dataset explorer has no table to read and there are no rows on this page. The files are linked below, and the research agent can fetch them onto its machine and work through them there.

Try asking

“Fetch the Papyrus protein-target table on the machine and tell me how many distinct human targets have more than 500 standardised activity records.”

Open in the research agent

Licence

Check the licence before commercial use · CC BY-SA 4.0 · code MIT

Read 2026-10-24: the Zenodo records for releases 05.5, 05.6 and 05.7 (Zenodo version 2024.2) all declare CC BY-SA 4.0, while the download and query scripts are MIT. CC BY-SA permits commercial use but requires attribution and places the same share-alike terms on anything adapted from the data — read it before building a proprietary model on top of these tables.

Questions

Can I browse Papyrus in the browser?

No. Its authors publish it outside the Hugging Face Hub, so the dataset explorer has no table to read and there are no rows to show here (format: Tab-separated archives (.tsv.xz) and SD files, plus precomputed descriptor sets). This page describes what it holds and links to the files; the research agent can fetch and read them on its machine.

What licence is Papyrus under?

CC BY-SA 4.0, with the accompanying code under MIT. Read 2026-10-24: the Zenodo records for releases 05.5, 05.6 and 05.7 (Zenodo version 2024.2) all declare CC BY-SA 4.0, while the download and query scripts are MIT. CC BY-SA permits commercial use but requires attribution and places the same share-alike terms on anything adapted from the data — read it before building a proprietary model on top of these tables.

How to cite it

Béquignon, O. J. M. et al. Papyrus: a large-scale curated dataset aimed at bioactivity predictions. Journal of Cheminformatics 15, 3 (2023).