Calculation, semi-empirical

Spectroscopic methods such as IR and Raman have proven to be exceptionally powerful methods for solving many chemistry problems . However, the vibrational assignment, as well as the understanding of the relationship between the observed spectral features and molecular structure or reactivity of the sample, can be very difficult. Theoretical methods can certainly assist to obtain a deeper understanding of the vibrational spectra of new compounds. These are the well-established force field calculations, semi-empirical and ab initio methods . 

The calculated VCD spectra were obtained using the NECO formalism (equation 15), and C=0 stretching vibrations for the guanine (1655 cm"1) and cytosine (1650 cm"1) groups, and an independent C = C-C stretching mode on cytosine (1612 cm1). The directions of the transition moments were calculated semi-empirically [25]. 

The authors125 propose that the correlation energy be calculated semi-empirically in terms of a functional of the Hartree-Fock density p, since HF wavefunctions are relatively easy to obtain. This type of approach, of which earlier examples have been given by Gombas,126 expressed Ec in the form (10). 

Andersson, R, Haglund, P, Rappe, C. and Tysklind, M. (1996). Ultraviolet Absorption Characteristic and Calculated Semi-empirical Parameters as Chemical Descriptors in Multivariate Modelling of Polychlorinated Biphenyls. J.Chemom., 10,171-185. 

As noted at the beginning of the chapter, the four main methods for calibrating fractionation factors are theoretical calculations, semi-empirical bond-strength models, natural sample data, and laboratory experiments. Theoretical methods have already been... 

Functional fonns based on the above ideas are used in the FIFD [127] and Tang-Toeimies models [129], where the repulsion tenn is obtained by fitting to Flartree-Fock calculations, and in the XC model [92] where the repulsion is modelled by an ab initio Coulomb tenn and a semi-empirical exchange-repulsion tenn Cunent versions of all these models employ an individually damped dispersion series for the attractive... 

To calculate N (E-Eq), the non-torsional transitional modes have been treated as vibrations as well as rotations [26]. The fomier approach is invalid when the transitional mode s barrier for rotation is low, while the latter is inappropriate when the transitional mode is a vibration. Hamionic frequencies for the transitional modes may be obtained from a semi-empirical model [23] or by perfomiing an appropriate nomial mode analysis as a fiinction of the reaction path for the reaction s potential energy surface [26]. Semiclassical quantization may be used to detemiine anliamionic energy levels for die transitional modes [27]. 

The first point to remark is that methods that are to be incorporated in MD, and thus require frequent updates, must be both accurate and efficient. It is likely that only semi-empirical and density functional (DFT) methods are suitable for embedding. Semi-empirical methods include MO (molecular orbital) [90] and valence-bond methods [89], both being dependent on suitable parametrizations that can be validated by high-level ab initio QM. The quality of DFT has improved recently by refinements of the exchange density functional to such an extent that its accuracy rivals that of the best ab initio calculations [91]. DFT is quite suitable for embedding into a classical environment [92]. Therefore DFT is expected to have the best potential for future incorporation in embedded QM/MD. 

Molecular orbitals were one of the first molecular features that could be visualized with simple graphical hardware. The reason for this early representation is found in the complex theory of quantum chemistry. Basically, a structure is more attractive and easier to understand when orbitals are displayed, rather than numerical orbital coefficients. The molecular orbitals, calculated by semi-empirical or ab initio quantum mechanical methods, are represented by isosurfaces, corresponding to the electron density surfeces Figure 2-125a). 

Whereas it is generally sufficient (at least for the pubhshed methods) to specify the semi-empirical MO technique used in order to define the exact method used for the calculations, ab-initio theory offers far more variations, so that the exact level of the calculation must be specified. The starting point of most ab-initio jobs is an SCF calculation analogous to those discussed above for semi-empirical MO calculations. In ab-initio theory, however, all necessary integrals are calculated correctly, so that the calculations are very much (by a factor of about 1000) more time-consuming than their semi-empirical counterparts. 

DFT calculations offer a good compromise between speed and accuracy. They are well suited for problem molecules such as transition metal complexes. This feature has revolutionized computational inorganic chemistry. DFT often underestimates activation energies and many functionals reproduce hydrogen bonds poorly. Weak van der Waals interactions (dispersion) are not reproduced by DFT a weakness that is shared with current semi-empirical MO techniques. 

All the techniques described above can be used to calculate molecular structures and energies. Which other properties are important for chemoinformatics Most applications have used semi-empirical theory to calculate properties or descriptors, but ab-initio and DFT are equally applicable. In the following, we describe some typical properties and descriptors that have been used in quantitative structure-activity (QSAR) and structure-property (QSPR) relationships. 

Molecular dipole moments are often used as descriptors in QPSR models. They are calculated reliably by most quantum mechanical techniques, not least because they are part of the parameterization data for semi-empirical MO techniques. Higher multipole moments are especially easily available from semi-empirical calculations using the natural atomic orbital-point charge (NAO-PC) technique [40], but can also be calculated rehably using ab-initio or DFT methods. They have been used for some QSPR models. 

The molecular electronic polarizability is one of the most important descriptors used in QSPR models. Paradoxically, although it is an electronic property, it is often easier to calculate the polarizability by an additive method (see Section 7.1) than quantum mechanically. Ah-initio and DFT methods need very large basis sets before they give accurate polarizabilities. Accurate molecular polarizabilities are available from semi-empirical MO calculations very easily using a modified version of a simple variational technique proposed by Rivail and co-workers [41]. The molecular electronic polarizability correlates quite strongly with the molecular volume, although there are many cases where both descriptors are useful in QSPR models. 

The MEP at the molecular surface has been used for many QSAR and QSPR applications. Quantum mechanically calculated MEPs are more detailed and accurate at the important areas of the surface than those derived from net atomic charges and are therefore usually preferable [Ij. However, any of the techniques based on MEPs calculated from net atomic charges can be used for full quantum mechanical calculations, and vice versa. The best-known descriptors based on the statistics of the MEP at the molecular surface are those introduced by Murray and Politzer [44]. These were originally formulated for DFT calculations using an isodensity surface. They have also been used very extensively with semi-empirical MO techniques and solvent-accessible surfaces [1, 2]. The charged polar surface area (CPSA) descriptors proposed by Stanton and Jurs [45] are also based on charges derived from semi-empirical MO calculations. 

This quantity is found to be related to the local polarization energy and is complementary to the MEP at the same point in space, making it a potentially very useful descriptor. Reported studies on local ionization potentials have been based on HF ab-initio calculations. However, they could equally well use semi-empirical methods, especially because these are parameterized to give accurate Koopmans theorem ionization potentials. 

The problem with most quantum mechanical methods is that they scale badly. This means that, for instance, a calculation for twice as large a molecule does not require twice as much computer time and resources (this would be linear scaling), but rather 2" times as much, where n varies between about 3 for DFT calculations to 4 for Hartree-Fock and very large numbers for ab-initio techniques with explicit treatment of electron correlation. Thus, the size of the molecules that we can treat with conventional methods is limited. Linear scaling methods have been developed for ab-initio, DFT and semi-empirical methods, but only the latter are currently able to treat complete enzymes. There are two different approaches available. 

Breindl et. al. published a model based on semi-empirical quantum mechanical descriptors and back-propagation neural networks [14]. The training data set consisted of 1085 compounds, and 36 descriptors were derived from AMI and PM3 calculations describing electronic and spatial effects. The best results with a standard deviation of 0.41 were obtained with the AMl-based descriptors and a net architecture 16-25-1, corresponding to 451 adjustable parameters and a ratio of 2.17 to the number of input data. For a test data set a standard deviation of 0.53 was reported, which is quite close to the training model. 

---