Biology Reference
In-Depth Information
extensive research to develop models including hydrogen bonding,
solvent, and electrostatic effects [8-11]. These functional additions to
the design models are especially important for full sequence design
since packing interactions no longer dominate for noncore residues
(e.g., surface and intermediate residues). The incorporation of these
additional noncore residues increases the potential for diversity, and
therefore enhances the probability for improving functionality when
compared to the parent system. An additional complication is the need
to account for changes in amino acid compositions and inherent propen-
sities through the appropriate definition of a reference state [9,12-13].
TEMPLATE FLEXIBILITY
Many computational protein design efforts were based on the premise
that the three-dimensional coordinates of the template or backbone were
fixed. This assumption was first proposed by Ponder and Richards [4],
and was appealing because it greatly reduced the search space, and
the time required to converge to a solution for the minimum energy
sequence, regardless of the kind of search method employed. However,
the assumption was also highly questionable. Protein backbones had
been observed to allow residues that would not have been permissible
had the backbone been fixed [14]. In the Protein Data Bank, there exist
numerous examples of proteins that exhibit multiple NMR structures.
Though commonly assumed as rigid bodies as a first approximation,
the secondary structures of a-helices and b-sheets actually display some
twisting and bending in the protein fold, and Emberly et al. [15,16] had
applied principal component analysis of database protein structures to
quantify the degree and modes of their flexibility.
Su and Mayo [17] claimed that their ORBIT (Optimization of
Rotamers By Iterative Techniques) computational protein design process
was robust against 15% change in the backbone. Nevertheless, they
found out on a later case study on T4 lysosome that core repacking to
stabilize the fold was difficult to achieve without considering a flexible
template [18]. Therefore, to ensure that good sequence solutions are not
rejected, it is more desirable to assume backbone flexibility in de novo
protein design.
Researchers have formulated several methods to incorporate template
variability. First, backbone flexibility can simply be modeled by using
smaller atomic radii in the van der Waals potential. One common practice
has been to scale down the radii by 5-10% [6,19], thus permitting slight
overlaps between atoms due to backbone movements. Key disadvan-
tages of this simple approach include overestimation of the attractive
forces and also the possibility of hydrophobic core overpacking.
Another way to allow for backbone flexibility is through consid-
ering a discrete set of templates by using genetic algorithms and
Search WWH ::
Custom Search