Characterizing and understanding the reaction condition space for DNA-encoded chemistry with tailored lab automats and advanced statistics

• Applicants: Professor Dr. Andreas Brunschweiger; Professorin Dr. Katja Ickstadt; Professor Dr.-Ing. Norbert Kockmann
• Subject Area: Chemical and Thermal Process Engineering ; Epidemiology and Medical Biometry/Statistics ; Organic Molecular Chemistry - Synthesis and Characterisation
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 566029859

DNA-encoded libraries (DELs) of drug-like molecules are a widely-used technology for the identification of compounds binding to a biological target. DELs are based on the covalent linkage of compounds to barcoding DNA oligomers and are synthesized by combinatorial solution phase chemistry. The DEL synthesis consists of iterative, combinatorial cycles of building block coupling, solvent exchange, and enzymatic ligation of DNA barcode oligonucleotides, hence, vast compound libraries are accessed for screening on targets. The chemical nature of the DNA-tagged molecule dictates the choice of solvents ranging from polar for the DNA part to unipolar for the chemical building blocks. Furthermore, additives and co-solvents can assist to accommodate substrates and reagents. Laboratory automation of microfluidic equipment for DEL synthesis and integrated analysis will assist the screening efforts to increase the reproducibility and accuracy of the synthesis and analysis steps. Although automated microreactor systems are already established for reaction screening, setting up a complete automated process is still in its infancy. Integrated sensors on a fluidic platform, chemical analytics, robotics, and control feedback loops based on advanced statistical analysis and design of experiments are even more challenging. Hence, statistical methods will contribute crucially in designing and redesigning experiments, in analyzing the data sets, and in interpreting results. We will design automated laboratory equipment for batch and flow DEL chemistry and use it to explore solvent and reaction conditions space in terms of DNA solubility and DNA stability assisted by modern statistical methods. Cellulose ethers will be studied as solvent additives to promote reactions on DNA-encoded substrates in aqueous co-solvents, and to dissolve DNA oligonucleotides in (dry) organic solvents such as MeCN, dichloroethane, and alcohols for a given reaction. Our aim is to understand the combinatorial solvent and additive (cellulose ether) space for solubilizing DNA in aqueous and non-aqueous (co-)solvents. To understand DNA integrity under prospective reaction conditions, a DoE approach will guide experiments for incubation of DNA tags with catalysts and reagents under different conditions to obtain a matrix of DNA-compatible reaction conditions. This matrix will guide DNA reaction design and experimental investigations. With statistical regression approaches as Machine Learning methods experimental data is analyzed to predict reaction conditions for selected reactions, aiming at high DNA-encoded compound yields, broad substrate scope, and low DNA damage. The toolbox of learning approaches, datasets and readme files will be used as a data infrastructure in future projects for development of DNA-encoded chemistry methods. A long-term goal is a lab equipment pool and related ML model infrastructure for data-enabled design of DNA-encoded chemistry.

• DFG Programme: Research Grants