Experimental Versus Predicted Heats of Reaction for Some Common Reaction Types in Pharmaceutical Industry

Abstract

Common pharmaceutical reactions were modeled using the commercially available CHETAH program and the novel TCIT program to obtain heats of reaction values. The programs were used to predict thermochemical data for isolated molecules based on their functional groups. Both programs use group theory, CHETAH, commercially available through ASTM and TCIT using a more modern approach and under development at Purdue. Reactions such as amide coupling (via both uronium reagent and T3P), debenzylation, bromination, tert-butyloxycarbonyl deprotection, Suzuki couplings, and halogenation were analyzed. These values were then compared to experimental values to assess the accuracy of each program’s predictions. Experimental values were provided by several pharmaceutical companies, who participated and advised in this endeavor, as listed in Acknowledgements. Deviations were on average within ±20% of experimental values for both programs, but outlier behavior with respect to certain functional groups was distinct. CHETAH calculations were at times hampered by missing groups, which was not the case for TCIT, which is designed to be arbitrarily extensible. This project resulted in a broader molecular library for TCIT, exposure of shortcomings in CHETAH, and strategies to improve accuracy moving forward.