Customizable Generation of Synthetically Accessible, Local Chemical Subspaces

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Synthetic Chemistry

Background:

• Screening large chemical libraries is key for discovering bioactive compounds, but library size is a limitation.
• Combining combinatorial and computational chemistry offers a time-efficient strategy to explore chemical space.
• Integrating tools for evaluating large, feasible chemical libraries is essential for drug discovery.

Purpose of the Study:

• To present Finders and React2D, integrated into the Virtual Chemist platform, to enhance virtual combinatorial chemistry.
• To automate the identification of synthetically accessible compounds from user-defined chemical transformations.
• To streamline the design of realistic chemical libraries for medicinal chemistry applications.

Main Methods:

• Developed a modular software suite, Virtual Chemist, integrating Finders and React2D.
• Implemented automated, black-box processes for identifying compatible chemicals and reaction outcomes.
• Considered synthetic feasibility, including chemoselectivity, stereoselectivity, protecting groups, and chemical availability.

Main Results:

• Generated approximately 40 million synthetically accessible, stereochemically accurate compounds.
• Utilized a starting library of 100,000 purchasable molecules and 56 chemical reactions.
• Demonstrated the utility of the automated platform for rapid library generation.

Conclusions:

• The Virtual Chemist platform, with Finders and React2D, significantly enhances virtual combinatorial chemistry.
• This approach enables the efficient generation of large, realistic, and synthetically viable chemical libraries.
• The software provides an intuitive interface crucial for adoption in medicinal chemistry laboratories.