ПІДБІР ОПТИМАЛЬНИХ НАПІВЕМПІРИЧНИХ МЕТОДІВ ДЛЯ МОДЕЛЮВАННЯ ДОВЖИН ЗВ’ЯЗКІВ В АЗОЛЬНИХ СИСТЕМАХ

Abstract

Theoretical quantum-chemical methods are intensively used in studies and prediction of different properties of newly synthesized compounds. Thus, QC methods can be used for modeling and determination of the most stable conformations in solutions, spectrophotometric characteristics of dyes and polymers, complexation ability, determination of reactivity centers and regio-selectivity. Moreover, quantum-chemical methods are widely used for study and prediction of potential biological activity. However, given the large size of target biological molecules, to save machine time and resources, it is useful to use much cheaper semiempirical methods. Definitely, it is often necessary to make calculations at ab initio, DFT or even post Hartree-Fock methods, but at least in the geometry optimization phase, it is wise to use fast semiempirical NDDO (neglect of diatomic differential overlap) methods. Taking into account the above, in this paper we have validated a set of semiempirical methods from the point of view of their accuracy for prediction of equilibrium bond lengths in various azole systems. Such an examination is currently relevant. Among the methods we have considered: AM1, PM3, RM1, PM6, PM7. The choice of azole systems is given by a wide range of biological activity. In particular, derivatives of [1,2,4]triazolo[1,5-a]pyrimidine exhibit anti-tumor, antiviral, herbicidal and fungicidal activity. Moreover, substituted [1,3]thiazolo[1,2,4]triazoles have bactericidal and fungicidal activity, exhibit anti-inflammatory effects and can be used to treat gout. The considered semiempirical methods show different accuracy about calculated bond lengths in three selected azole systems. We must note that the newest PM7 method is generally the most accurate and reliable. Corresponding mean absolute error (MAE) is 0.0272 Å and root-mean-square deviation (RMSD) is 0.0334 Å. As an alternative, we advise to use the RM1 method. Despite that AM1 is often used to simulate large systems, it has shown to be the worst method with the highest statistical errors. In general, we have recommend to use PM7 Hamiltonian for reliable modeling of geometric parameters of azole systems.