Descriptors and Physicochemical Properties

Chemoinformatics is characterized by the use of large amounts of information. Specific ways to represent the molecules, and to organize and analyze the data have been and continue to be developed. There are different ways to represent the molecules, and they can be classified according to the information that they encode. The most basic level corresponds to representations that depend on or are associated with one dimensional (ID) representation, such as molecular weight. The next level corresponds to the 2D representations associated with the connectivity of the molecules without the consideration of the stereochemistry. In 3D methods the incorporation of stereochemistry conveys not only the specification of the chirality of stereogenic centers but also the possible conformation or conformations. In this regard, the research fields dedicated to conformational analysis have an important impact in reactivity prediction, molecular design, and stability. Examples of ID, 2D, and 

2D Fingerprints Atom types and substructures in a binary representation 

3D Quantum chemical descriptors Change distribution in the molecules 

3D descriptors along with the property represented are summarized in Table 2.1. 

A number of methods have been developed to store and share information. It is desirable that such information is stored in an inexpensive way and, depending on the task, retaining as much structural detail as possible. Based on the molecular representation used, the molecules are stored in suitable formatted files. For instance, for 2D representations, the molecules can be represented by fingerprints, SMILES, or SMARTS strings such files use a small amount of memory. For 3D representations, typical formats include. sdf,. pdb,. mol, and. mol2 in all these cases the files contain the 3D coordinates of the structures, thus the conformation can be defined. 

---

Wame, M., St. J., Connell, D. W., Hawker, D. W. (1990) Prediction of aqueous solubility and the octanol-water partition coefficient for lipophilic organic compounds using molecular descriptors and physicochemical properties. Chemosphere 16, 109-116. 

The virtual library was then characterized using the Cerius2 default topological descriptors and physicochemical properties (35). The 50 default descriptors were reduced to three principal components using principal components analysis, and this defined a 3D chemistry space into which the virtual library could be plotted. The chemistry space consisted of 1134 cells and, when the virtual library was mapped into the space, it was found to occupy 364 of the cells thus, this represents the maximum cell coverage that is achievable. 

Contents I. Introduction 34 II. Molecular Descriptors and Physicochemical Properties 36 III. Molecular Databases and Chemical Space 37 IV. Chemoinformatics in Food Chemistry 40 V. Examples of Molecular Similarity, Pharmacophore Modeling, Molecular Docking, and QSAR in Food or Food-Related Components 43 A. Molecular similarity 43 B. Pharmacophore model 47 C. QSAR and QSPR 48 D. Molecular docking 49 VI. Concluding Remarks and Perspectives 52 Acknowledgments 53 References 53... 

Abstract The aim of the present chapter is to present the current research and potential applications of chemoinformatics tools in food chemistry. First, the importance and variety of molecular descriptors and physicochemical properties is delineated, and then a survey and chemical space analysis of representative databases with emphasis on food-related ones is presented. A brief description of methods commonly used in molecular design, followed by examples in food chemistry are presented, such methods include similarity searching, pharmacophore modeling, quantitative... 

Wame, M. St.J., D.W. Connell, D.W. Hawker, and G. Schtiurmarm. Prediction of Aqueous Solubility and the Octanol-Water Partition Coefficient for Lipophilic Organic Compounds Using Molecular Descriptors and Physicochemical Properties, Chemosphere, 21(7) 877-888 (1990). 

Chemical reactivity and biological activity can be related to molecular structure and physicochemical properties. QSAR models can be established among hydrophobic-lipophilic, electronic, and steric properties, between quantum-mechanics-related parameters and toxicity and between environmental fate parameters such as sorption and tendency for bioaccumulation. The main objective of a QSAR study is to develop quantitative relationships between given properties of a set of chemicals and their molecular descriptors. To develop a valid QSAR model, the following steps are essential ... 

Quantitative structure-activity relationships (QSARs) are important for predicting the oxidation potential of chemicals in Fenton s reaction system. To describe reactivity and physicochemical properties of the chemicals, five different molecular descriptors were applied. The dipole moment represents the polarity of a molecule and its effect on the reaction rates HOMo and LUMO approximate the ionization potential and electron affinities, respectively and the log P coefficient correlates the hydrophobicity, which can be an important factor relative to reactivity of substrates in aqueous media. Finally, the effect of the substituents on the reaction rates could be correlated with Hammett constants by Hammett s equation. 

Finally, some authors (Lu et al., 2000 Sabljic, 2001) suggest that due to the significant correlation between the topological indices and physicochemical properties, such as log Kow the former alone or in principal components can replace the latter in the QSPR models. The strongest support of this point of view is probably the fact that the TIs do not bring experimental or calculation error into the models. For the successful replacement of physicochemical properties such as log Kow, more than one structural descriptor is often required, which reflects the stability and interpretability of the resulting models. 

Several papers have also been published in which a correlation has been sought between permeation across Caco-2 cells and physicochemical properties of the compounds. The review article by Ekins et al. (2000) discusses several studies to predict Caco-2 cell permeation. Correlations have been found with polar surface area, hydrogen bond descriptors, VolSurf, and other parameters. Van de Waterbeemd et al. (2001a) also discuss models for predicting oral absorption of compounds, including the use of Caco-2 cell lines. This paper also provides much useful information on the optimization of pharmacokinetic parameters in drug development. 

The so-called solvatochromic or linear solvation energy relationship (LSER) descriptors developed by Abraham and coworkers (Kamlet et al., 1983) have proved valuable in correlating a wide variety of biological endpoints and physicochemical properties, and two studies have utilized them to model BCF. Park and Lee (1993) found the following QSAR for the fish BCF values of a set of diverse chemicals ... 

Analog-filtering procedure was applied, based on the molecular physicochemical descriptors and optimum property ranges calculated for the set of known HIV PR inhibitors (Table 4.1), which permitted the selection of 100 most diverse cyclic urea analogs with suitable molecular properties. 

Benchmarking studies on various biological and physicochemical properties 307,312 QSAR/QSPR models for involving fragment descriptors... 

Theoretical (computational) calculations can also offer quantitative descriptors of physicochemical properties of the molecular structures, molecular interactions, and thermodynamics of interactions. Principally, extensive studies on the catalytic site of GP have been exploited in theoretical QSAR studies [4]. The techniques engaged correlate biochemical behaviors with the known crystallographic structures, and map regions around the inhibitor molecule and added water molecules to improve the in silico prediction [106-110]. 

Only a few compounds screened in early lead identification phases are synthesized in-house. More flexible and cost effective is to purchase chemicals from external suppliers. Most vendors provide lists of some ten to himdred thousand chemicals on compact discs and guarantee delivery within days to weeks. To explore this huge amount of data with the aid of computers, chemical information is transformed to computer-readable strings, e.g., smiles code, and different descriptors are determined. 1-dimensional (1-D) descriptors encode chemical composition and physicochemical properties, e.g., molecular weight, stoichiometry (C O Hj,), hydrophobicity, etc. 2-D descriptors reflect chemical topology, e.g., connectivity indices, degree of branching, number of aromatic bonds, etc. 3-D descriptors consider 3-D shape, volume or surface area. 

Undoubtedly, the energetic descriptors of molecules are very valuable. In many cases they contribute in a decisive way to the determination and interpretation of the chemical and physicochemical properties of molecules. However, quantum chemical methods are still less useful if one wishes to know the energetic descriptors solely of some part of a molecule. In this matter, approximate models are much more helpful than the methods originated from quantum chemistry, although the latter are more sophisticated and better grounded theoretically. 

With the growing number of proteins sequenced, there is a necessity for novel techniques to analyze protein sequences in order to determine their structure and function. The most commonly used protein sequence descriptors are based on evolutionary information and physicochemical properties. Even though these methods have proven to be efficient in most cases, in cases of transmembrane proteins, they may fall short. As the vast field of transmembrane proteins largely remains unexplored with many transmembrane proteins yet to be sequenced, it is possible to obtain new protein sequences without any known homolog. In such cases, traditional sequence analysis methods based on alignment profiles would not be sufficient. The evolutionary information-based descriptors appear inadequate, and indices based on physicochemical property can cause ambiguities. Therefore, it is of considerable interest to develop novel methods based on sequence information alone to represent protein sequences. 

The QSPR/QSAR methodology can also be applied to materials and mixtures where no structural information is available. Instead of descriptors derived from the compound s structure, various physicochemical properties, including spectra, can be used. In particular, spectra are valuable in this context as they reflect the structure in a sensitive way. 

An extensive series of studies for the prediction of aqueous solubility has been reported in the literature, as summarized by Lipinski et al. [15] and jorgensen and Duffy [16]. These methods can be categorized into three types 1 correlation of solubility with experimentally determined physicochemical properties such as melting point and molecular volume 2) estimation of solubility by group contribution methods and 3) correlation of solubility with descriptors derived from the molecular structure by computational methods. The third approach has been proven to be particularly successful for the prediction of solubility because it does not need experimental descriptors and can therefore be applied to collections of virtual compounds also. 

---