Computational Approaches in Peptide and Protein Design: An Overview - Peptide and Protein Design for Biopharmaceutical Applications - page 14

Atthesametime,theRamachandranmap in Figure 2.1(f), calculated
with the much older ECEPP force field using rigid valence geometry
without accounting for solvent, whether implicit or explicit, shows at
least as good consistency with experimental data in Figure 2.1(e) as the
GROMOSandOPLS-AAmaps.Indeed,theallowedareasofthismap
cover all three regions, determined experimentally with approximately
the same relative populations. The zone region was also populated; in
fact, conformation C 7 eq had the lowest relative energy. Even the diag-
onal shape of the experimental distribution was preserved in the alpha R
region (and, though just slightly, in the alpha L region) of Figure 2.1(f).
The main difference from the experimental data was in the extension of
thealphaRregionintheECEPPmaptoward(f, c) values around ( 60
- 30, 150-30); these are virtually unpopulated in Figure 2.1(e).
One reason for the good consistency of the ECEPP map with the experi-
mental distribution was that the parameters of the ECEPP force field
were calibrated specifically to reproduce the X-ray data on crystal
packing of amino acids. This made the ECEPP force field limited in
applications to molecules other than peptides and proteins, which pre-
cluded its acceptance by commercially available modelling packages,
such as SYBYL, INSIGHT or MacroModel. Nevertheless, comparison
of maps in Figure 2.1(e) and (f) clearly suggests that the ECEPP force
field, though less sophisticated than GROMOS or OPLS, may be suc-
cessfully used in the sampling of possible conformational states of
peptide and protein systems. At the same time, practical applications
of the ECEPP force field require significantly less computer resources,
with the additional advantage that it is able to perform sampling in
dihedral angle space, which is much less complex than Cartesian coor-
dinate space (see Section 2.2.2.3).

Further developments of molecular force fields The MM force fields
outlined above are routinely used in computational design of peptides
and proteins. However, as was concluded in a recent review, 'Currently,
force fields are not perfect (even the all-atom ones). It is possible to obtain
different results with different force fields. Therefore, improving force
fields (both the all-atom and reduced ones, and the water potential) is a
priority.' [26]. Further development of force fields has occurred recently,
offering several possible improvements.
One approach is to combine quantum mechanics with molecular
mechanics (QM/MM), whereby selected parts of the molecular system
under study, for instance the active sites of enzymes, are treated by QM
approximations, and the larger surrounding molecular areas are