DNA-Encoded Chemical Library Screening with Target Titration Analysis: DELTA

Area of Science:

• Medicinal Chemistry
• Biochemistry
• Computational Chemistry

Background:

• DNA-encoded chemical libraries (DELs) are powerful tools for screening vast numbers of small molecules against biological targets.
• DEL screening data often suffer from noise due to variations in synthetic yields, complicating hit identification.
• Machine learning models benefit from DEL data for virtual screening, but data quality is a key challenge.

Purpose of the Study:

• To develop and validate a novel split-sample DNA-encoded chemical library screening strategy.
• To improve the accuracy of binding affinity estimation and hit compound ranking in DEL screening.
• To address the challenge of noisy data and enhance the detection of high-affinity binders.

Main Methods:

• A split-sample DEL screening approach was employed against Bruton's tyrosine kinase (BTK).
• Affinity selections were performed at varying target concentrations.
• A probabilistic model was used to estimate binding affinity and relative library member concentrations.
• Model predictions were validated against Surface Plasmon Resonance (SPR) measurements of resynthesized compounds.

Main Results:

• The developed methodology provided an improved ranking of library members by binding affinity compared to traditional enrichment metrics.
• The probabilistic model successfully estimated binding affinities and relative input concentrations.
• A highly potent binding compound, previously undetectable, was recovered using this enhanced method.
• Validation against SPR confirmed the improved accuracy of the model's predictions.

Conclusions:

• The split-sample DEL screening strategy with a probabilistic model significantly enhances the reliability of hit identification.
• This approach mitigates the impact of synthetic yield variations, leading to more accurate binding affinity assessments.
• The methodology offers a more sensitive and robust way to discover potent drug candidates from large chemical libraries.