Dataset - Papyrus 2024 - A large scale curated dataset aimed at bioactivity predictions

This update of release 2024.1 fixes the following:

• Metadata in the columns type_IC50, type_EC50, type_KD, type_Ki, and type_other did not contain multiple values when multiple pChEMBL values where available but reported only a single value. This fix ensures all values are reported.
• Molecules were incorrectly standardized and mixtures were included in the dataset. Standardization (using the papyrus_structure_pipeline) is now correctly enforced and mixtures have been removed.

Changes since version 05.6

• ChEMBL data was updated to ChEMBL version 34
• data from the IUPHAR/BPS Guide to PHARMACOLOGY has been included
• data from Pickett et al.'s publication on MMP-12 has been included (ACS Med Chem Lett. 2011 Jan 13; 2(1): 28–33. DOI: 10.1021/ml100191f)

Papyrus++:

Previous versions mistakenly considered a deviation of 2 log units around compound-target pairs to determine the reproducibility of assays (see published article for more details). This has been fixed to 0.5 log units to ensure data points fall within a maximum range of 1 log unit. As a result, the number of entries in the Papyrus++ set from this release has drastically reduced compared to previous releases.