ESE-ΔH-DNN – A dense Neural Network for evaluation of solvation free energy and enthalpy based on Generalized-Born terms

Introduction and User's Guide

ESE-ΔH-DNN is a method for evaluation of solvation free energy and enthalpy of neutral molecules in organic solvents [2]. It requires the solute molecular geometry only. To obtain the solvation free energy, first electronegativity-equalization atomic charges are calculated. Subsequently, the Born-type terms, atomic surfaces and volumes are evaluated and, along with five solvent features, are fed into a Dense Neural Network.

The present scheme is an adaptation of my previous ESE-GB-DNN method. In contrast to the previous ESE-GB-DNN scheme [3], the ESE-ΔH-DNN method yields both ΔG_solv and ΔH_solv. However, ESE-ΔH-DNN was trained on a database of neutral solutes in organic solvents. Therefore, if you need aqueous solutions or ions, please use ESE-GB-DNN

The present ESE-ΔH-DNN scheme is a continuation of my previous works [3–5] and is loosely based on the uESE [7] and xESE [8] that we developed jointly with Alexander Voityuk.

The supported elements are H, C, N, O, F, Si, P, S, Cl, Br, I.

ΔG_solv and ΔH_solv can be calculated by the program ESE-ΔH-DNN, which can be downloaded here free of charge:

ESE-DeltaH-DNN.exe – Windows version

ESE-DeltaH-DNN.x – Linux version.

The ESE-ΔH-DNN program can be run from the command line as follows:

ESE-DeltaH-DNN.exe xyz-file -solvent solvent

Warning: The xyz-file should contain atomic symbols (or numbers) and Cartesian coordinates (in Å) and an empty line at the end. It should not contain any header.

If your solvent is not in this list, you can use the following call format:

ESE-DeltaH-DNN.exe xyz-file -Eps dielectric_constant -BP boiling_point_°C -Nheavy number_of_non_hydrogen_atoms_in_solvent -MolVolume molar-volume -H-bond number_of_hydrogen_bon_centers>

Once you use results calculated by the ESE-ΔH-DNN program, you should include at least the following citations:

1. S. F. Vyboishchikov, ESE-ΔH-DNN program, Girona, 2024

2. S. F. Vyboishchikov, Liquids, 2024, 4, 525–538. DOI: 10.3390/liquids4030030

3. S. F. Vyboishchikov, J. Chem. Theory Comput., 2023, 19, 8340–8350. DOI: 10.1021/acs.jctc.3c00858

and preferably also cite some our previous related work:

4. S. F. Vyboishchikov, J. Chem. Inf. Model., 2023, 63, 6283–6292. DOI: 10.1021/acs.jcim.3c00922

5. S. F. Vyboishchikov, J. Comput. Chem., 2023, 44, 307–318. DOI: 10.1002/jcc.26894

6. S. F. Vyboishchikov, A. A. Voityuk, Chemical Reactivity, vol. 2: Approaches and applications, S. Kaya, L. von Szentpály, G. Serdaroğlu, K. Guo (Eds.), Elsevier, Amsterdam, 2023, 399–427. DOI: 10.1016/B978-0-32-390259-5.00021-4

7. S. F. Vyboishchikov, A. A. Voityuk, J. Chem. Inf. Model., 2021, 61, 4544–4553. DOI: 10.1021/acs.jcim.1c00885

8. S. F. Vyboishchikov, A. A. Voityuk, J. Comput. Chem., 2021, 42, 1184–1194. DOI: 10.1002/jcc.26531

9. A. A. Voityuk, S. F. Vyboishchikov, Phys. Chem. Chem. Phys. 2020, 22, 14591–14598. DOI: 10.1039/d0cp02667k

10. A. A. Voityuk, S. F. Vyboishchikov, Phys. Chem. Chem. Phys., 2019, 21, 875–874. DOI: 10.1039/c9cp03010g

Questions related to the ESE-ΔH-DNN method and program should be addressed to Sergei Vyboishchikov.