Descriptor , molecular structure generation

A descriptor for the 3D arrangement of atoms in a molceulc can be derived in a similar manner. The Cartesian coordinates of the atoms in a molecule can be calculated by semi-empirical quantum mechanical or molecular mechanics (force field) methods, For larger data sets, fast 3D structure generators are available that combine data- and rule-driven methods to calculate Cartesian coordinates from the connection table of a molecule (e.g., CORINA [10]). 

Because of the large number of chemicals of actual and potential concern, the difficulties and cost of experimental determinations, and scientific interest in elucidating the fundamental molecular determinants of physical-chemical properties, considerable effort has been devoted to generating quantitative structure-property relationships (QSPRs). This concept of structure-property relationships or structure-activity relationships (QSARs) is based on observations of linear free-energy relationships, and usually takes the form of a plot or regression of the property of interest as a function of an appropriate molecular descriptor which can be calculated using only a knowledge of molecular structure or a readily accessible molecular property. 

Molecular descriptors and chemical spaces. The majority of chemoinformatics methods depend on the generation of chemical reference spaces into which molecular data sets are projected and where analysis or design is carried out. The definition of chemical spaces critically depends on the use of computational descriptors of molecular structure, physical or chemical properties, or pharmacophores. Essentially, any comparison of molecular characteristics that goes beyond simple structural comparison requires the calculation of property values and the application... 

We have evaluated three different techniques to generate QSAR models, namely Comparative Molecular Field Analysis (CoMFA), Comprehensive Descriptors for Structural and Statistical Analysis (CODESSA), and Hologram QSAR (HQSAR). More specifically they were evaluated for their utility (predictivity, speed, accuracy, and reproducibility) to predict ER binding activity quantitatively (Tong et al., 1998 Shi et al., 2001). Common to the three QSAR methods is the... 

Quantitative Structure Activity Relationship (QSAR) is a method that makes predictions by the quantitative description of molecular properties with the use of descriptors of the chemical structure (Dearden 2003). This means QSAR models describe the quantitative or calculated relationship between a chemical structure and their biological activity (e.g. toxicity) with the help of chemical descriptors that are generated from the molecular structure (Durham and Pearl 2001). This relationship is described in from of a mathematical equation (e.g. log 1/C = a tt + b a +. .. + const). QSAR models generally show better predictivity if all compounds of a dataset involved in the prediction are derived from a congeneric series of compounds, that means they should all act by the same mechanism of action, since the physico-chemical and structural descriptors used in the QSAR reflect the same mechanism of action. Sometimes it is difficult to determine the mechanism of action, so series of compounds involved in a QSAR model are often restricted to a given chemical class in the hope that this will ensure a single mechanism of action (Dearden 2003). 

Bauknecht et al. [48] and Sadowski et al. [49] have also developed molecular descriptors that are based on the physicochemical properties of the atoms in a molecule. They calculate several different electronic properties for each atom in a molecule and then use autocorrelation to generate a fixed-length vector that is independent of the size of the molecule. Autocorrelation was first applied to the topology of a molecular structure by Moreau and Broto [50] using the following function ... 

Generation and storage of molecular structure descriptors from the topological or geometrical representations of the structures. 

The fundamental idea of QSAR consists of the possibility of a relationship between a set of descriptors, which are derived from molecular structure, and a molecular response. Within this scope, several molecular descriptors, which discretely parameterize a given molecular set, have been devised. From the early work of Cross [1], where a relationship between toxicity and water solubility was observed, several other parameters have been proposed, such as Hammett s sigma [5], which accounts for electronic effects due to molecular backbone substitution, or the octanol/water partition coefficient [6], widely used to describe lipophilicity. Later on, other frameworks have been formulated in order to include molecular shape, size, polarizability, and many other structural features, also based on the three-dimensional molecular character. A number of reviews have been published [7-9], concerning the historical development, descriptor generation, and their application into the QSAR field. 

---