The assessment of computational methods with sets of high-quality reliable reference data, also known as benchmarking, has become an indispensable tool in quantum chemical method development.[1-3] A common practice is to test the accuracy of newly developed methods (e.g. force fields, semi-empirical approaches, machine-learning approaches, electron density functionals, and empirical corrections) by comparing the results obtained by them to those obtained with more accurate, but expensive methods. This has led to the construction of several reference data sets of molecular properties that are generally obtained with accurate quantum mechanical (QM) methods. In this contributed poster, I will present the compilation of new diverse benchmark data sets of three different molecular properties that are absent from the literature. The molecular properties considered for the construction of the three new data sets are: (i) polypeptide conformational energies (PEPCONF database[4]), as they reflect the interplay between covalent and non-covalent interactions that are present in proteins, (ii) bond dissociation enthalpies (BDE49 database), as they are an important theoretical descriptor of the strength of chemical bonds, and (iii) reaction barrier height energies (BH10 database), as they are immensely helpful for computational investigation of chemical reaction mechanisms. This three data sets are expected to serve the immediate purpose of addressing the need of such sets for development and comparison of new computational methods that are applicable for modeling enzymatic catalysis.

References [1] Mata, R. A.; Suhm, M. A. Angew. Chemie Int. Ed. 2017, 56 (37), 11011–11018. [2] Mardirossian, N.; Head-Gordon, M. Mol. Phys. 2017, 115 (19), 2315–2372. [3] Goerigk, L.; Mehta, N. Aust. J. Chem. 2019, 72(8), 563-573. [4] Prasad, V. K.; Otero-de-la-Roza, A.; DiLabio, G. A. Sci. Data 2019, 9, 180310.