Descriptor/Property correlation

Structural Interpretation — It is helpful if a descriptor can be interpreted in a way that allows features of the chemical structure or even the entire chemical structure to be derived. This is particularly interesting, if a molecular descriptor is applied to the task like a spectrum in this case we could talk about an artificial spectrum. Several approaches exist that use such artificial spectra to compare them with experimental spectra — for instance, to derive experimental spectra that do not exist in a database. Descriptor/Property Correlation — This is one of the basic tasks for a descriptor descriptors are usually helpful for predicting properties or comparing properties of chemical compounds. 

Descriptor/Property Correlation is a requirement for a molecular descriptor to correlate with at least one property of a molecule. 

These first components of the autocorrelation coefficient of the seven physicochemical properties were put together with the other 15 descriptors, providing 22 descriptors. Pairwise correlation analysis was then performed a descriptor was eliminated if the correlation coefficient was equal or higher than 0.90, and four descriptors (molecular weight, the number of carbon atoms, and the first component of the 2D autocorrelation coefficient for the atomic polarizability and n-charge) were removed. This left 18 descriptors. 

It is also important to check for correlations between the descriptors. Highly correlated descriptors could lead to the information that they encode being over-represented. A straightforward way to determine the degree of correlation between two properties is to calculate a correlation coefficient. Pearson s correlation coefficient is given by ... 

PSA is also used as H-bond descriptor to predict various properties of chemicals and drugs. PSA is defined as that part of a molecular surface that arises from oxygen and nitrogen atoms, and also the hydrogens attached to them. Applications of PSA as a QSAR descriptor in correlations with permeability and absorption were carried out first by Van de Waterbeemd et al. [30] and Palm et al. [31]. Clark [32-34] developed this further. Chapter 5 in this book is completely devoted to... 

Many groups have discussed the correlation between solubility and molecular properties [14-19], and the octanol/water partition coefficient, the molecular volume and surface area, the boiling point and charge distribution in the molecules are well-documented molecular descriptors that correlate strongly with experimental solubility. 

For a QSAR analysis a training set of compounds with known descriptor properties (e. g. pKa-values, surface areas, dipole moments etc.), including the property of interest, is required. The Hansch Analysis1461 is a statistical method to analyze and correlate these data in order to determine the magnitude of the target property (Eq. 2.15). 

The IV is defined as an average total number of double bonds per mole in a mixture of fatty materials. It is not a good descriptor for correlating physical and chemical properties with fatty acid composition in biodiesel. For example, the IV does not provide any information on structural factors such as number of allylic or bis-allylic methylene groups or location of double bond(s) within the hydrocarbon chain. Knothe (2002) recommended alternative indices termed allylic position equivalents (APE) and Ws-allylic position equivalent (BAPE) based on the total number of allylic and bis-allylic positions present in the fatty acid chains. 

Descriptors based on the 2D structure or simply on the connectivity matrix of a structure have long been used for chemical similarity and for property correlations. Because they often lack any relationship to mechanism, these descriptors are best used within a congeneric series or at least a set of similar structures. They may be empirically useful for cluster analysis and chemical library design, because they are effective at representing structure differences and similarities. A few programs and providers of topological descriptors include the following ... 

Although several molecular quantities were defined from the beginiming of quantum chemistry and graph theory, the term molecular descriptor has become popular with the development of structure-property correlation models. The - Platt number [Platt, 1947] and - Wiener index [Wiener, 1947c], defined in 1947, are sometimes referred to as the first molecular descriptors. 

It is not easy to find successful structure-activity/ property correlations, but the rapid growth of publications dealing with QSAR/QSPR studies clearly demonstrates the progress in this area. To obtain a significant correlation, it is crucial that appropriate descriptors be employed, whether they are theoretical, empirical, or derived from readily available experimental characteristics of the structures. Many descriptors reflect simple molecular properties and thus can provide insight into the physicochemical nature of the activity/ property under consideration. 

Van de Waterbeemd, H., dementi, S., Costantino, G., Carrupt, P.-A., Testa, B. CoMEA derived substituent descriptors for structure-property correlations. In 3D QSAR in drug design theory, methods, and applications (Kubinyi, H., Ed.). ESCOM Leiden, 1993, pp. 697-707. 

QSPR models (quantitative structure-property relationship) are derived from simple-descriptors and correlated to a set of experimental data. Examples are estimation of physico-chemical properties, ADME or toxicity properties. 

Major reasons for drug candidates to fail in early clinical phases are unpropitious pharmaco-kinetic properties (such as a lack of bioavailability) or toxicity. Therefore, estimation of these properties is an important part of a combinatorial library profile. An increasing number of publications [6, 42-45] demonstrate the importance of ADME parameters (ADME = absorption, distribution, metaboHsm, and ehmina-tion). Several commercial software packages are available. The ways in which ADME parameters are derived are similar in most available software products. In a first step, simple descriptors are calculated and in a second step, these descriptors are correlated with experimental data [46]. The parameters must describe properties that are important for pharmaco-kinetics (lipophilicity, size, and polarity of molecules, etc.). Standard correlation methods can be used, because the type of correlation has only secondary effects on the results. Generally, the predictivity of classification methods (the output from which can only be good or bad ) is slightly better compared to a quantitative correlation. 

Other topological indices can be obtained by using suitable functions applied to local vertex invariants the most common functions are atom and/or bond additive, resulting into descriptors, which correlate well physico-chemical properties, that are atom and/or bond additive themselves. Zagreb indices and ID numbers are derived according to this approach. 

Ren, B. (2003c) Atomic-level-based AI topological descriptors for structure-property correlations. 

We will use a topological formalism to develop most of our structure-property correlations for polymers. This formalism utilizes connectivity indices defined via graph theoretical concepts as its main structural and topological descriptors. Connectivity indices have been widely used for simple molecules. A review is provided in this section to familiarize the reader with these indices before discussing their extension to polymers. The information in this section is summarized from two books by Kier and Hall [1,2], to which the reader is referred for additional details. The first book [1] is more detailed, while the second book [2] includes the results of the research over the decade after the publication of the first book. 

As can be seen from equations 2.4-2.7, from Table 2.2, and most dramatically from figures 2.5-2.8, the % values are also extensive properties. They are sums over all vertices or edges of the hydrogen-suppressed graph. The number of terms in each summation increases in direct proportion to the size of the molecule or the polymeric repeat unit. This is the reason why the % values are proportional to N to a good approximation. They are, therefore, logical choices of topological descriptors to correlate with extensive properties. 

Fig. 2.1 Commonly employed descriptors and fingerprints in structure-property correlation studies...

The original CODESSA program was realized as an interactive menu system for the MS Windows environment and enabled to generate and use four different types of data sets stmctures, properties, descriptors, and correlations. The program allowed both interactive and file-based input of molecular stmctures and data. The output included textual and numerical data on stmctures, properties and descriptors, and graphical representation of the results of the statistical treatment. 

The number of occupied orbitals, the orbitals energies, and the differences between the energy of orbitals are descriptors. For example, expression of the ionization potential and the electron affinity through the energies of the HOMO and LUMO orbitals (Delchev et al. 2006) allows definition of the chemical hardness as (Elumo Ehomo)/2. The value of these descriptors often correlates with certain macroscopic properties such as reactivity, magnetic, electrical, and optical properties. 

---