Electron density fingerprints (EDprints): virtual screening using assembled information of electron density

Area of Science:

• Computational Chemistry
• Cheminformatics
• Drug Discovery

Background:

• Molecular fingerprints are crucial for virtual screening in drug discovery.
• Existing methods like MOLPRINT 2D and FCFP-4 have limitations in handling certain molecular complexities.
• Efficient similarity searching is vital for large-scale database analysis.

Purpose of the Study:

• To introduce and evaluate a new molecular fingerprint method, EDprints, based on molecular electron densities.
• To compare the virtual screening performance and speed of EDprints against established fingerprinting techniques.
• To identify the strengths and weaknesses of EDprints in various molecular screening scenarios.

Main Methods:

• Encoding molecular properties (NMR shifts, partial charges) into electron density-based fingerprints (EDprints).
• Retrospective virtual screening accuracy assessment using the Directory of Useful Decoys (DUD).
• Comparative analysis with MOLPRINT 2D and FCFP-4 using ligand-based similarity searches.
• Benchmarking computational speed for fingerprint comparison.

Main Results:

• EDprints demonstrated comparable overall virtual screening accuracy to MOLPRINT 2D and FCFP-4.
• EDprints showed increased robustness for molecules with fused ring systems and subtle group rearrangements.
• EDprints exhibited significant speed advantages (56-233x) in fingerprint comparisons.
• EDprints' similarity searches can be influenced by molecular protonation states.

Conclusions:

• EDprints offers a valuable alternative for virtual screening, particularly for large datasets and specific molecular architectures.
• The speed of EDprints makes it highly suitable for exhaustive similarity searches.
• Further development may be needed to mitigate biases related to molecular protonation states.