Medicinal Chemistry Aware Database GDBMedChem.

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• The GDB17 database contains over 166 billion molecules, but lacks medicinal chemistry filtering.
• Existing drug databases may not fully represent chemical diversity or possess optimal synthetic accessibility.

Purpose of the Study:

• To create a curated subset of drug-like molecules from GDB17 with medicinal chemistry principles.
• To generate a diverse and synthetically accessible chemical library for drug design.

Main Methods:

• Applied medicinal chemistry rules to filter GDB17, removing problematic functional groups and complex structures.
• Sampled the filtered subset uniformly for molecular size, stereochemistry, and polarity.
• Developed GDBMedChem, a collection of 10 million small molecules.

Main Results:

• GDBMedChem exhibits reduced molecular complexity and enhanced synthetic accessibility compared to GDB17.
• The GDBMedChem collection shows higher sp3-carbon fraction and natural product likeness than current drugs.
• GDBMedChem offers greater molecular diversity and novel substructures compared to existing molecule databases.

Conclusions:

• GDBMedChem provides an unprecedented source of chemical diversity for drug discovery.
• The database facilitates exploration of novel chemical space for identifying new drug candidates.
• GDBMedChem is accessible for 3D visualization, similarity searching, and download.