Pharmacophore-oriented 3D molecular generation toward efficient feature-customized drug discovery

Area of Science:

• Computational chemistry
• Medicinal chemistry
• Drug discovery

Background:

• Current molecular generation methods are impractical for real-world drug discovery.
• Intelligent drug discovery requires advanced molecular generation techniques.

Purpose of the Study:

• To introduce PhoreGen, an explicit pharmacophore-oriented 3D molecular generation method.
• To demonstrate PhoreGen's capability in generating feature-customized molecules for drug discovery.

Main Methods:

• PhoreGen utilizes asynchronous perturbations and updates on atomic and bond information.
• A message-passing mechanism incorporates ligand-pharmacophore mapping knowledge during diffusion-denoising.
• The method generates 3D molecules with pharmacophore alignment.

Main Results:

• PhoreGen generates chemically reasonable, diverse, and drug-like 3D molecules with good binding affinity.
• The method successfully identified new bicyclic boronate inhibitors for metallo-β-lactamase and serine-β-lactamase.
• Inhibitors for metallo-nicotinamidases, emerging insecticide targets, were also identified.

Conclusions:

• PhoreGen offers an effective, explicitly constrained approach for feature-customized molecular generation.
• This method shows significant potential in accelerating intelligent drug discovery and identifying novel therapeutic agents and insecticides.