Evaluating the Consistency and Accuracy of COSMO-RS Based Free Energy Predictions

Since its first publication in 1995, COSMO-RS[1] has found wide-spread use and has developed a reputation² as an efficient approach for the qualitative and quantitative prediction of many thermodynamic properties in liquid systems. The commonly predicted properties include activity coefficients, solubilities, phase separation, vapor pressures and two phase partitioning.

In this contribution, we evaluate the accuracy and consistency of COSMO-RS predictions on the basis of two different cases: a recent benchmark [3,4] comparing different methods for predicting the free energy of solvation of simple organic liquids and the 2016 SAMPL[5] blind challenge for water / cyclohexane distribution of drugs.

The presented data show that COSMO-RS provides the most accurate predictions in both cases. A good consistency over different properties and different classes of molecules is also observed as all predictions were conducted with the same parameters and without prior modifications or adjustments.

References:

1) A. Klamt, “Conductor-like screening model for real solvents: a new approach to the quantitative calculation of solvation phenomena,” The Journal of Physical Chemistry, vol. 99, no. 7, pp. 2224–2235, 1995.

2) P. Deglmann, A. Schäfer, and C. Lennartz, “Application of quantum calculations in the chemical industry - an overview,” International Journal of Quantum Chemistry, vol. 115, no. 3, pp. 107–136, 2015.

3) J. Zhang, D. Tuguldur, D. van der Spoel, “Force Field Benchmark of Organic Liquids. 2. Gibbs Energy of Solvation.” Journal of Chemical Information and Modeling, vol. 55, no. 6, pp. 1192-1201, 2015

4) J. Zhang, D. Tuguldur, D. van der Spoel, “Correction to Force Field Benchmark of Organic Liquids. 2. Gibbs Energy of Solvation” Journal of Chemical Information and Modeling, vol. 56, no. 5, pp. 819-820, 2016

5) https://drugdesigndata.org/about/sampl