Published February 18, 2026 | Version 1.0.2

Data Packages for the Paper "Scaffold-Based Evaluation Metrics for Fair Comparison of Molecular Generators"

  • 1. ROR icon University of Chemistry and Technology, Prague
  • 2. LACDR, Leiden University

Description

This dataset accompanies the paper “Scaffold-Based Evaluation Metrics for Fair Comparison of Molecular Generators.”

It contains all input data, output sets, and results used in the study:

  • data/input_recall_sets/ – Input Sets and Recall Sets for Glucocorticoid receptor and Leukocyte elastase

  • data/nuclear_receptor/ – Original ChEMBL31 data for the Glucocorticoid receptor, used in preprocessing

  • data/protease/ – Original ChEMBL31 data for the Leukocyte elastase, used in preprocessing

  • data/information_about_clusters/ – Statistics for five clusters across two different splits

  • data/results/ – Results for five different generators (some with different settings), structured by target (Glucocorticoid receptor and Leukocyte elastase), scaffold type (CSK, Murcko), and data split (dis/sim)

All analysis scripts are available in the associated GitHub repository: GitHub link.

The dataset is provided as a single data/ folder. To reproduce the results with the provided scripts, place this folder into the root directory of the GitHub repository.

Files

data.zip

Files (8.9 GB)

Name Size
md5:9ffe53da90a6fa513225413685a2ee55
8.9 GB Preview Download

Additional details

Related works

References
Journal article: 10.1186/s13321-026-01213-6 (DOI)