interatomicdistances), but also on the conformation of the molecule (as defined by
rotations around torsional angles). If the drug molecule under study is “large and floppy”
(i.e., it is conformationally labile and exists in a family of low energy conformers), it is
difficult to identify the lowest energy conformer using quantum pharmacology calcula-
tions. For example, if the putative drug being studied is a hexapeptide, it will exist in a
multiplicity of low energy shapes; the hexapeptide’s potential energy surface will have
many, many low energy wells, and trying to identify the global energy minimum (the
lowest energy well) is a challenging task. Such energy surfaces may have billions of low
energy wells, and trying to identify the single lowest energy well is a computationally
demanding problem. This problem is sometimes referred to as the multiple minima prob-
lem. The multiple minima problem also explains our inability to predict protein folding
when our only starting information is the primary amino acid sequence of a protein.
There exist a number of techniques for addressing the multiple minima problem
when trying to identify the lowest energy conformer for a flexible drug or for a recep-
tor protein. These techniques are computational chemistry methods that enable one to
“search the conformational space” of the floppy drug molecule or protein under study.
TheMonte Carlo methodwas one of the first methods used to search conformational
space, having been adapted from classical statistical mechanics. Using this method,
random moves are made to the rotatable bonds of an isolated molecule. Then, using a
Metropolis sampling procedure, it is possible to generate a large number of suitable
conformations. The spectrum of acceptable conformations is then energy minimized
(using a quantum mechanics or molecular mechanics approach, as discussed above),
and ranked by energy. Although it is necessary to generate a large number of confor-
mations, in principle it is possible, within a user-defined timeframe, to achieve a repre-
sentative sample from low-energy conformational space.
A second, widely used method for searching conformational space is through mole-
cular dynamicscalculations. A simple definition of molecular dynamics is that it sim-
ulates the motions of a system of atoms with respect to the forces that are present and
acting on the molecule. This collection of forces causes the system to change, but by
collective motion of atoms over time, in a way that is described by integrating Newton’s
second law of motion (F=ma, where Fis the force acting on an atom,mis its mass,
andais its acceleration). If one can calculate the next configuration of the collection of
atoms, it is possible to follow the evolution of the atomic movements within the mole-
cule over time. This is different from the Monte Carlo method, which requires outside
intervention to produce change by a random move; in molecular dynamics, all changes
result without external intervention and arise from within the system itself. In a molec-
ular dynamics calculation, the molecule is “heated” by assigning velocities randomly to
the atoms for a given temperature. Once the first velocities have been assigned, the mol-
ecular dynamics simulation is self perpetuating. As the simulation of the atomic move-
ments progresses, the new atomic positions are calculated. By “heating” the molecule
and permitting it to cool, it is possible to explore the conformational space of the
molecule, thereby identifying low energy shapes.
Thegenetic algorithmmethod is a technique that has very recently gained attention
for searching conformational space. Genetic algorithms may be applied to the multiple
minima problem of molecular conformational analysis via a variety of methods. In one
such method, the torsional angles within a given molecule are designated as “genes.”
50 MEDICINAL CHEMISTRY