Biomedical Engineering Reference
In-Depth Information
protein and a known protein structure. This usually involves the use of a protein structure template
databases, such as the PDB. Selecting an appropriate group of database entries from the database to
serve as structure templates is typically based on some form of sequence comparison or threading.
Pairwise sequence comparison involves searching sections of the template candidate for amino acid
sequences that are similar to sequences in the target protein. A key decision in sequence comparison
is how similar is similar enough. Multiple sequence comparison relies on an iterative algorithm that
expands the template search to include all reasonable candidate templates from the template
databases. As a result, multiple sequence comparison is more sensitive and more likely to find
suitable templates in the template database.
Threading involves aligning the sequence of the target protein with the 3D structure of a template to
determine whether the amino acid sequence is spatially and chemically similar to the template.
Threading can be thought of as searching through a bin of factory-second gloves, looking for a glove
that fits, where the hand is the target protein and the gloves represent templates. Some gloves may
be able to accommodate only four fingers (no thumb), whereas others might have a channel for a
sixth finger. These gloves represent templates that don't match the target protein. Gloves that, on
visual inspection (the gloves aren't actually tried on—yet) can accommodate five fingers on the
proper hand—assuming the hand is "normal"—are retained as potential templates. Similarly,
templates that best fit the target protein are identified for use later in the comparative modeling
process.
There are various forms of threading. For example, in contact potential threading, which is based on
the analysis of the number and closeness of contacts between amino acids in the protein core, the
idea is to position amino acids and compute empirical energies from the observed associations of
amino acids. The most energetically stable conformation is the most likely protein structure. A more
complex form of threading involves modeling energies from first principles. This method is based on
dynamic programming techniques and is a recursive method of solving a problem that involves
saving intermediate results in a matrix or table so that they can be used for future calculations.
Regardless of the technology used, the goal is to find the template that best fits the target protein's
structure. Template selection is complicated because not only do different sequences adopt the same
fold, but many combinations of amino acids can fit into the same 3D conformation.
Alignment
The goal of the alignment phase of comparative modeling is to align the sequence of polypeptides in
the target sequence with that of the template structure in order to position the target and template in
the same 3D orientation. Continuing with the glove scenario, alignment involves placing the hand in
the glove so that all of the fingers fill the appropriate sections of the glove. Many of the alignment
procedures are based on dynamic programming techniques, often supplemented with manual
methods based on visual inspection of the molecule.
Model Building
Once the libraries of templates that match the target protein have been identified, the actual model
building or assembly can begin. Ideally, the structure of one of the templates will exactly fit the
definition of the target protein, suggesting that the structure of that target is identical to that of the
template. However, this almost never occurs. For any given target-template pair, there are likely to
be several bends or kinks in the template backbone that don't align with the sequence in the target
protein. Because a single change in only one bond angle can result in a major change in the
conformation of a protein molecule, there is a better chance of identifying a fit if parts of the
template are pieced together, one at a time. The issue in this approach is how big a piece of template
structure to use.
One approach, rigid body assembly, uses large segments of the template that are dissected at
natural folds and reassembled over the superimposed structure of the target molecule. The accuracy
of model building through rigid body assembly can be increased if parts from several templates are
available because there is an increased chance that a sub-assembly of the molecule—a rigid
