AMBER empirical potential describes the geometry and energy of noncovalent halogen interactions better than advanced semiempirical quantum mechanical method PM6-DH2X.

A recently published study on halogen bonding in casein kinase 2 (CK2)-inhibitor complexes claimed that the halogen-bond-corrected PM6 semiempirical method (PM6-DH2X) describes the halogen bonding properties correctly, whereas the AMBER empirical potential fails. The current study employs our positive extra-point (PEP) approach for halogen bonding, in which the σ-hole on the halogen atom is represented by an extra point of charge. The performance of the two methods in describing halogen bonding in halobenzene···Lewis base and CK2-inhibitor complexes was reassessed. Compared to basis set superposition error- (BSSE-) corrected MP2/aug-cc-pVDZ (with PP functions on the Br and I atoms) data, the AMBER force field described the halogen bonding in halobenzene···Lewis base complexes slightly better than the PM6-DH2X method. Fifteen polyhalogenated benzimidazole inhibitors (taken from a study of Dobeš et al.) complexed to CK2 protein were studied. The binding energies were calculated using the molecular mechanical-generalized Born surface area (MM-GBSA) approach. Compared to the corresponding experimental data, the AMBER force field yielded much better results than the PM6-DH2X method. Finally, the performance of both methods in describing C-X···π-system and C-X···H-O/C interactions was examined. A comparison with MP2 data revealed that the PM6-DH2X method failed to describe them, whereas the AMBER force field performed well.