Systemic evolutionary chemical space exploration for drug discovery.

Area of Science:

• Computational chemistry
• Drug discovery
• cheminformatics

Background:

• Chemical space exploration is crucial for identifying novel chemical entities.
• Computational de novo design offers an advantage over limited chemical libraries.
• Fragment-based drug design principles inspire new approaches.

Purpose of the Study:

• To introduce the systemic evolutionary chemical space explorer (SECSE) platform for de novo drug design.
• To demonstrate SECSE's capability in generating novel and diverse small molecules.
• To present SECSE as an open-source tool for virtual hit generation.

Main Methods:

• SECSE employs a miniaturized "lego-building" approach within target protein pockets.
• Human intelligence and deep learning are integrated for enhanced search and optimization.
• The platform frames virtual hit generation as a computational search problem.

Main Results:

• SECSE successfully identified novel and diverse small molecules against phosphoglycerate dehydrogenase (PHGDH).
• The generated molecules serve as attractive starting points for further drug validation.
• The platform's efficacy in exploring chemical space for hit finding was demonstrated.

Conclusions:

• SECSE is a potent computational tool for de novo design and chemical space exploration.
• The integration of AI and fragment-based principles enhances virtual screening capabilities.
• SECSE provides an open-source solution for accelerating the discovery of novel chemical entities.