Computational Approaches in Peptide and Protein Design: An Overview - Peptide and Protein Design for Biopharmaceutical Applications - page 10

force fields, such as ECEPP, employ rigid valence geometry and therefore
do not include the first two terms of the potential expression, which
results in significant reduction of the computer time needed for energy
calculations. All of the other force fields include flexibility of valence
geometry. Options are to consider all hydrogens separately (slower cal-
culations) or consider aliphatic hydrogens united with their bonded
carbon atoms (united-atom assumption; speedier calculations). The
force fields with flexible valence geometry are utilized also for other
classes of biochemical compounds, such as nucleic acids, carbohydrates,
etc. Specifically, force fields for peptides and proteins have been reviewed
many times, emphasizing different aspects of their applications; the
reader is referred to the recent excellent reviews by Ponder and Case
[22] and Mackerell [23].
It is difficult to determine which force field is preferable for conforma-
tional calculations involving peptides and proteins. On the one hand,
obviously, the force fields with flexible valence geometry and those
calibrated to fit the results of quantum chemistry calculations are more
likely to yield an accurate value of energy for a given state of a peptide
system including solvent. But on the other hand, computational
approaches are especially valuable for problems involving conforma-
tional flexibility of peptide chains and orientations of ligands within the
binding site of the receptor. In both cases, estimations of energy for a
large number of possible states of the system are required in order to
select the most plausible states as fast as possible with as few as possible
false positives and false negatives, providing some justification for the
more computationally efficient approximations.
In this regard, a convenient test utilizes reconstruction of the
Ramachandran map for the simplest element of the peptide chain,
the acetyl-N-Methyl- L -alanine (Ac-Ala-OMe), by various force fields. The
Ramachandran map is a function of the two torsional angles, f(rotation
around the bond NH-C a )andc(C a -CO), adjacent to the a-carbon that
maps the potential surface of peptide backbone conformations. The
Ramachandran map can be roughly divided into four quadrants: the upper-
left, corresponding to (f < 0, c > 0); the lower-left (f < 0, c < 0);
the upper-right (f > 0, c > 0); and the lower-right (f > 0, c < 0). The
upper-left quadrant contains the (f, c) points corresponding to an extended
structure, such as a b-strand (f 140, c 140), and the lower-left and
upper-right quadrants contain points corresponding to the right- (f 60,
c 60) and left-handed (f 60, c 60) a-helices, respectively. The
upper-left quadrant also contains the (f, c) points ( 75,140)and( 80,
80), corresponding to conformations P II (polyproline II) and C 7 eq (the