Then, two randomly selected starting conformations for this molecule are generated;
one conformation is termed “mother,” the other is “father.” One-half of the genes (torsional
angles) is selected from each of the parents and combined to produce an offspring confor-
mation. If this offspring has a lower energy than its parents (as determined using either
molecular mechanics or quantum mechanics calculations), the conformation is said to
have “fitness” and is permitted to survive. The “most fit” conformations are permitted to
propagate by exchanging their genes with their sibling conformers. A mathematical pro-
cedure, termed a “mutation operator,” is used to incorporate greater diversity amongst
the genes as successive generations are created. Genetic algorithm calculations permit
families of low energy conformers to be identified.
Monte Carlo methods, molecular dynamics calculations, and genetic algorithm meth-
ods are all techniques for searching conformational space; each has strengths and weak-
nesses. The techniques are complementary rather than competitive, and may be used
together in a concerted attempt to identify low energy conformers of drug molecules.
Since these methods are simply techniques for skipping across the conformational
space of a molecule, they must be used in conjunction with a mechanics method (e.g.,
quantum mechanics or molecular mechanics) to provide values of energy for the vary-
ing conformations that they generate.
One final issue, which confounds the use of these methods for identifying the elusive
global energy minimum, concerns the biological relevance of this lowest energy con-
formation once it has been identified. Just because a detailed quantum mechanics cal-
culation has identified a given conformation as the lowest energy shape for a drug
molecule, this does not mean that this is the bioactive conformation. The interaction of
a drug with its receptor is a dynamic process in which each molecule flexes to fit the
other. It is entirely possible that the drug molecule may assume a higher energy con-
formation (by several kcal/mol) in order to achieve this fit, thereby rendering the search
for a global energy minimum somewhat irrelevant.
For someone who has never taken a course in quantum mechanics, this discussion of
quantum pharmacology may have been somewhat confusing. However, understanding
these basic principles is important because of the important and ever-increasing role of
molecular modeling in drug design and discovery. The diverse concepts of section 1.6
presented thus far may be summarized as follows. The “mechanics” methods (quantum
mechanics [section 1.6.1.1], molecular mechanics [section 1.6.1.2]) provide a single
value of energy for a single shape or conformation of a drug molecule. Since a mole-
cule may have an almost infinite variety of shapes, the infinite number of single energy
values corresponding to these shapes define a surface (termed the potential energy
hypersurface). The lowest point on this surface (global minimum) is assumed to repre-
sent the most probable shape of the molecule. However, finding the lowest point on the
surface is difficult. Methods such as Monte Carlo, molecular dynamics, and genetic
algorithms (section 1.6.2) permit one to “hop and skip” across the potential energy
hypersurface to sample it in a point-by-point fashion which may identify a point (i.e., a
single conformation of the molecule) which lies in a low energy region of the surface.
Once a low energy region of the surface has been so identified, then energy minimiza-
tion algorithms (e.g., Newton-Raphson, section 1.6.1.4) may be used to “fine tune” the
geometry and conformation of the molecule to ensure that the lowest energy structure
has been identified. These concepts are diagrammatically illustrated in figure 1.13.
DRUG MOLECULES: STRUCTURE AND PROPERTIES 51