Chemical Library Space: Definition and DNA-Encoded Library Comparison Study Case

Area of Science:

• Chemoinformatics
• Computational Chemistry
• Drug Discovery

Background:

• DNA-encoded library (DEL) technology presents unique challenges for chemical library analysis.
• Treating chemical libraries, especially inseparable mixtures like DELs, as individual chemoinformatic objects is crucial.
• Existing methods may not adequately capture the complexity of analyzing large chemical libraries.

Purpose of the Study:

• Introduce the concept of Chemical Library Space (CLS) for analyzing and comparing chemical libraries.
• Develop and evaluate vectorial representations of libraries within CLS for effective similarity analysis.
• Demonstrate the application of CLS for selecting DNA-encoded libraries that match reference collections.

Main Methods:

• Definition and comparison of four vectorial library representations using generative topographic mapping.
• Development of property-tuned CLS encodings for simultaneous comparison of property and chemotype distributions.
• Application of CLS encodings to the selection of DNA-encoded libraries against the ChEMBL28 reference collection.

Main Results:

• Vectorial CLS representations enable effective comparison and interpretation of chemical library similarities.
• Property-tuned CLS encodings allow fine-tuning of library matching criteria based on properties and chemotypes.
• The CLS approach facilitates the selection of DNA-encoded libraries that optimally overlap with reference datasets.

Conclusions:

• Chemical Library Space (CLS) offers an efficient method for the polyvalent analysis of thousands of chemical libraries.
• CLS enables tunable selection of compound collections for drug discovery, optimizing for screening or portfolio enrichment.
• This approach supports strategic library design and selection by considering chemical space coverage and property distributions.