(1305-B) High Throughput Reaction Screening at GSK

Abstract: GSK Process Chemistry uses high throughput screening to identify and optimize potential reaction conditions for a range of reaction types. Using a combination of first pass templates and bespoke screening, initial hits can be identified and followed up on within as few as 2-3 screens, saving scientist time and materials. An end to end setup has been put in place that starts with the use of Katalyst, a high throughput electronic lab notebook, automated solid dispensing using a Chronect XPR, a, standardized work up procedure that includes an internal standard and dilution, and automated analysis and visualizations of results using Agilent OpenLab software and TIBCO Spotfire. An emphasis on strategic plate design and minimal reactions in early plates keeps material requirements low while screening a range of chemical space.
Minimal amounts of material are required when using the Chronect XPR which keeps the cost of high expense metal catalysts and other reagents low, and makes screening useful at even the earliest stages with small amounts of intermediates. A variety of reaction types can be screened in this manner including gas transfer, metal catalysis, and base and solvent screening. Literature precedent and selection of substrates that span chemical space can be used to identify first pass screening plates while follow up bespoke screenings can be used to optimize and cover a broader range of chemical space. This saves scientists time and efficiently spans a range of chemical space, minimizing work and maximizing results. The standardization of results using Katalyst and Spotfire will allow for data mining to take place and larger visualizations can be made over time to track trends in screening and successful reagents which allows for better first pass plates to be designed over time.
One example of a successful use of the platform is observed in a carbonylation screening. A first pass plate was run followed by three additional screens and conditions identified were able to be used within the process immediately. The first screen determined which base and class of ligand should be used, the second screen looked a wider range of ligands and the ligand to metal ratio and the final screen optimized the loading levels of the various reaction components. The final results showed full conversion of starting material with >90% assay yield and required around 2 weeks, a total of 92 reactions, and less than 1 g of material. Final results were used to run a reaction at the 100 g scale which validated that this hit was scalable and could be used in the process.