Enhanced Calculation of Property Distributions in Chemical Fragment Spaces

Area of Science:

• Computational Chemistry
• Medicinal Chemistry
• Drug Discovery

Background:

• Chemical fragment spaces offer vast potential for drug design but are too large for traditional enumeration-based algorithms.
• Existing methods struggle to analyze the immense size and complexity of these fragment spaces effectively.

Purpose of the Study:

• To introduce SpaceProp2, an enhanced algorithm for calculating exact property distributions of chemical fragment spaces without enumeration.
• To extend the SpaceProp algorithm's capabilities to include TPSA, rotatable bond counts, and user-defined SMARTS patterns.
• To generate example molecules for each property bin to facilitate detailed interpretation of fragment space composition.

Main Methods:

• Evolution of the SpaceProp algorithm to enable exact property distribution calculations for chemical fragment spaces.
• Incorporation of TPSA, rotatable bond counts, and SMARTS pattern matching capabilities.
• Demonstration on established make-on-demand fragment spaces and Boehringer Ingelheim's BICLAIM in-house space.

Main Results:

• SpaceProp2 successfully computes exact property distributions for large chemical fragment spaces without full enumeration.
• The algorithm provides distributions for TPSA, rotatable bonds, and user-defined SMARTS patterns.
• Generated example molecules offer clear interpretations of property bins and fragment space composition.

Conclusions:

• SpaceProp2 is a powerful tool for analyzing and designing chemical fragment spaces, overcoming the limitations of enumeration.
• Simultaneous multi-SMARTS searching and example molecule generation provide novel insights into vast combinatorial molecule collections.
• This approach significantly aids drug design projects by enabling better understanding and utilization of fragment space resources.