JEDA: Joint entropy diversity analysis. An information-theoretic method for choosing diverse and representative...

Area of Science:

• Computational chemistry
• Cheminformatics
• Drug discovery

Background:

• Selecting representative compound subsets from large combinatorial libraries is crucial for efficient screening.
• Existing diversity analysis methods may not fully capture the density distribution of chemical space.

Purpose of the Study:

• Introduce a novel method, Joint Entropy-based Diversity Analysis (JEDA), for selecting representative compound subsets.
• Compare JEDA's performance against established methods like Principal Components Analysis and Median Partitioning.

Main Methods:

• Utilizes chemical descriptors to partition chemical space.
• Employs a Shannon-entropy based scoring function within a probabilistic search algorithm.
• Allows user-defined subset sizes for practical application.

Main Results:

• JEDA selects diverse compound subsets that accurately reflect the density of the original chemical library.
• Demonstrates effective subset generation for a combinatorial library from the Comprehensive Medical Chemistry Dataset.
• Offers a flexible approach adaptable to resource constraints.

Conclusions:

• JEDA provides a robust and adaptable method for generating representative compound subsets.
• This approach enhances the efficiency of exploring chemical space in drug discovery.
• JEDA offers advantages in representing chemical space density compared to traditional methods.