GDBChEMBL

GDBChEMBL provides a curated subset of 10 million molecules from GDB17 selected by a ChEMBL-likeness score to prioritize compounds with circular substructures common to ChEMBL version 24 for applications in drug discovery and chemical biology.


Key Features:

  • Library composition: A subset of 10 million molecules derived from GDB17, which enumerates 166.4 billion possible molecules up to 17 atoms of carbon (C), nitrogen (N), oxygen (O), sulfur (S), and halogens.
  • ChEMBL-likeness score (CLscore): A scoring metric that guides selection by evaluating the presence of circular substructures shared with ChEMBL version 24.
  • Circular substructure matching: Selection is based on identifying circular molecular fragments common to known bioactive compounds in ChEMBL v24.
  • Uniform sampling across molecular properties: The subset is sampled uniformly across molecular size, number of stereocenters, and heteroatom content to ensure broad chemical-space representation.
  • Comparison to other GDB17 subsets: Contrasts with FDB17 (fragment-likeness) and GDBMedChem (medicinal chemistry criteria) by emphasizing ChEMBL-derived structural similarity to enhance synthetic accessibility and potential bioactivity.

Scientific Applications:

  • Drug discovery: Prioritizing candidate molecules with higher likelihood of synthetic feasibility and bioactivity for compound screening.
  • Lead optimization: Providing diverse, ChEMBL-like scaffolds to support optimization of potency, selectivity, and ADMET properties.
  • Structure–activity relationship (SAR) analysis: Enabling analysis of the impact of circular substructures on bioactivity by comparison to ChEMBL-derived fragments.
  • Chemical space exploration: Studying molecular diversity across size, stereochemistry, and heteroatom composition within a tractable subset of GDB17.

Methodology:

Selection of 10 million molecules from GDB17 was guided by a ChEMBL-likeness score (CLscore) based on circular substructure matches to ChEMBL version 24, followed by uniform sampling across molecular size, stereocenters, and heteroatom composition.

Topics

Details

Tool Type:
web application
Programming Languages:
Python
Added:
1/18/2021
Last Updated:
1/22/2021

Operations

Publications

Bühlmann S, Reymond J. ChEMBL-Likeness Score and Database GDBChEMBL. Frontiers in Chemistry. 2020;8. doi:10.3389/fchem.2020.00046. PMID:32117874. PMCID:PMC7010641.

Links