OpenGrowth: An Automated and Rational Algorithm for Finding New Protein Ligands

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Drug discovery

Background:

• De novo ligand design is crucial for identifying novel drug candidates.
• Existing methods may lack efficiency in generating molecules with desirable drug-like properties.
• Fragment-based and R-group strategies are common approaches in drug discovery.

Purpose of the Study:

• To introduce OpenGrowth, an open-source software for de novo ligand design.
• To enable the generation of novel ligands by connecting organic fragments within protein active sites.
• To bias molecule generation towards structures with favorable synthetic accessibility and pharmacokinetic profiles.

Main Methods:

• Utilizes a molecule growth strategy biased by a training database of existing drugs.
• Incorporates protein flexibility into the growth process.
• Supports seeding for R-group strategy and fragment-based drug discovery.
• Includes a graphical user interface for user-friendly fragment selection.

Main Results:

• OpenGrowth successfully generated novel inhibitors for HIV-1 protease.
• Identified inhibitors exhibited a predicted dissociation constant (Kd) as low as 18 nM.
• Molecules produced show superior synthetic accessibility and pharmacokinetic properties compared to random growth.

Conclusions:

• OpenGrowth is an effective tool for de novo ligand design, accelerating the identification of potential drug candidates.
• The software's ability to bias growth towards drug-like properties enhances the efficiency of early-stage drug discovery.
• OpenGrowth provides a user-friendly platform for exploring novel chemical space in drug development.