Three-dimensional compound comparison methods and their application in drug discovery.

Area of Science:

• Computational Chemistry
• Drug Discovery Informatics
• Medicinal Chemistry

Background:

• Virtual screening is a key process in drug discovery, with ligand-based virtual screening (LBVS) methods being widely adopted.
• LBVS methods do not require target receptor structural information and are computationally faster than structure-based approaches.
• LBVS methods are categorized into 1D, 2D, and 3D, with 3D methods offering potential performance enhancements by accounting for conformational flexibility.

Purpose of the Study:

• To review various three-dimensional (3D) ligand-based virtual screening (LBVS) methods.
• To benchmark the performance of four representative 3D LBVS methods in virtual screening applications.
• To evaluate the ability of these methods to identify dissimilar active compounds and assess their computational speed.

Main Methods:

• Review of existing three-dimensional (3D) ligand-based virtual screening (LBVS) methodologies.
• Benchmarking of four selected 3D LBVS methods using established virtual screening datasets.
• Performance evaluation focused on the identification of structurally diverse active compounds and computational efficiency.

Main Results:

• Three-dimensional (3D) LBVS methods demonstrate potential for improved virtual screening efficacy compared to lower-dimensional approaches.
• Performance variations were observed among the four benchmarked 3D LBVS methods regarding speed and the ability to retrieve diverse active compounds.
• Specific 3D LBVS methods showed advantages in balancing hit diversity and computational resource utilization.

Conclusions:

• Three-dimensional ligand-based virtual screening (LBVS) methods are valuable tools in drug discovery, particularly for exploring chemical space and identifying novel scaffolds.
• The choice of 3D LBVS method impacts screening efficiency and the diversity of identified active compounds.
• Further investigation into optimizing 3D LBVS parameters is warranted for maximizing drug discovery success rates.