Cheminformatics Approaches Aiding the Design and Selection of DNA-Encoded Libraries

Abstract

AbstractThe combinatorial nature of DNA-Encoded Libraries (DEL) creates a vast number of compounds covering a wide range of chemistry space, enabling DEL target selections for possible binders that generate enormous amounts of data. This makes advanced cheminformatics and data analysis approaches indispensable for advancing DEL technology for practical applications in the pharmaceutical industry. When designed DEL libraries can be biased to contain compounds with more drug like features through the enumeration of the potential virtual products, based on their combinatorial assembly from available building blocks, and the calculation of molecular descriptors. Computational approaches are also critical in analyzing selection data. In a DEL target selection, it is essential to apply stringent analysis methods to the raw sequencing data to separate real hits from noise and to mitigate the amount of resource required for off-DNA synthesis for hit confirmation. In this chapter we focus on the important role of cheminformatics and computational analysis at all phases of the DEL hit identification workflow by discussing in detail DEL product enumeration, designing DELs of diverse products with focused in specific property spaces, estimating data noise levels in DEL selections, and the process of DEL selection data triage and analysis that eventually identifies hits for on- and off-DNA synthesis. We also outline perspectives of emerging cheminformatics methods for DELs that could further strengthen the technology to facilitate hit identification processes and improve the hit confirmation rate. Emerging methods such as efficient structure searching, visualizing DEL chemistry space, and potential machine learning applications for DEL selection analysis are discussed.KeywordsBuilding-block selectionDELDNA-encoded libraryDEL designDEL hit identificationLibrary screeningNoise level estimateSelection data triage