|
|
|
|
|
|
|
structure-activity relationships [SAR or Q(uantitative) SAR], the chemist can more directly target the most fruitful directions for synthesis. This information generally comes from examining a series of molecules whose biological activities are known and whose structures and properties are known or can be calculated. Such evaluation is always done by medicinal chemists during decisions about the progress of their synthesis. Computational methods assist this process by providing quantification and enhanced accuracy of the molecular properties and structures and by utilizing methods to derive quantitative relationships between the features and the biological activities. Originally, this latter stage was done through least squares or regression analysis. Later, other statistical methods, pattern recognition, neural networks, and genetic algorithms, or other methods of multivariate analysis (MVA) were applied. The result is an equation relating the properties of a series of molecules to their biological activities. The biological activity of a new compound can be predicted by inserting its properties into the equation. A number of introductory references are available [46]. |
|
|
|
|
|
|
|
|
In addition to the structural properties mentioned below, properties that can be calculated include geometrical quantities such as molecular volume, shape, and surface area. Electronic properties include electrostatic field, electron density, polarizability, and molar refraction. Bulk properties include boiling point, solubility in a range of solvents, and similarly, partition constants. Programs and algorithms are available for their calculation from a wide range of academic, government, and commercial sources [7]. |
|
|
|
|
|
|
|
|
The 3-dimensional shape of a hormone or drug is essential to its interaction with the binding site of the proteins that regulate physiology. Molecular mechanics techniques can, in many cases, rapidly provide very accurate representations of the three dimensional structures and shapes of molecules [9,10]. Molecular mechanics model building, driven by highly engineered force fields (potential energy functions), can often provide structures as accurate as experiment, but within fractions of a second [11,12]. The development of these force fields often require considerable time and expertise. When they are not available for the molecules in question, quantum mechanics can often provide structures within experimental error with substantially increased computation times (perhaps tens to hundreds of thousands of times more time) [13]. |
|
|
|
|
|
|
|
|
Most molecules are flexible and can obtain many (often many millions, even for small molecules) different conformations (3-dimensional arrangement of atoms in space). The determination of the bioactive conformation, that one which actually causes the biological effect, is often a principal pursuit of a drug discovery program [14,15]. Conformational analysis can be extremely demanding computationally. Once obtained, the structure of a bioactive conformer is often linked to the QSAR methods mentioned above. |
|
|
|
|
|