ErG: 2D pharmacophore descriptions for scaffold hopping

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Reduced graph (rG) methods and binding property pairs (BPP) are established techniques in molecular representation.
• Existing methods may have limitations in capturing comprehensive molecular properties for drug discovery.

Purpose of the Study:

• To introduce and evaluate an extended reduced graph (ErG) approach for molecular property encoding.
• To compare the performance of ErG against Daylight fingerprints (DFP) for virtual screening tasks.
• To highlight ErG's capability for scaffold hopping and efficient similarity searching.

Main Methods:

• Developed an extended reduced graph (ErG) approach integrating pharmacophore-type node descriptions.
• Hybridized concepts from reduced graphs and binding property pairs.
• Validated ErG using 11 activity classes from the MDL Drug Data Report database.

Main Results:

• ErG demonstrated stable and high performance, outperforming or matching DFP in 10 out of 11 activity classes.
• ErG effectively identified structurally diverse active compounds, showcasing scaffold hopping capabilities.
• The method achieves high molecular abstraction, resulting in low-dimensional descriptor vectors for rapid similarity searches.

Conclusions:

• ErG is a versatile and effective method for identifying diverse active molecules.
• The approach offers significant advantages in computational efficiency for similarity searches.
• ErG provides a robust framework for drug discovery by enabling exploration of diverse chemical spaces.