Biomedical Engineering Reference
In-Depth Information
(target) protein molecule. Within the constraints of these assumptions and limitations, the
advantages of heuristic methods over ab initio methods are significant, and include improved
accuracy and an ability to work with large protein molecules as opposed to protein fragments. In
addition, the potential time savings of heuristic over experimental methods is a driving force for
investment in heuristic methods from the pharmaceutical and private investment communities.
The main heuristic method of predicting protein structure from amino acid sequence data is
comparative modeling—that is, finding similarities in amino acid sequence, independent of the
molecule's lineage. Comparative modeling is sometimes confused with homology modeling. However,
homology implies ancestral relationships, and assumes that proteins from the same families share
folding motifs even if they don't share the same sequences. In contrast, comparative modeling
assumes that proteins with similar amino acid sequences share the same basic 3D structure.
The basis for comparative modeling is typically the PDB, which contains descriptions of 3D structures
of proteins and other molecules as determined by NMR and X-ray crystallography experiments.
Another source of modeling data is the Molecular Modeling Database (MMDB), which combines PDB
data with cross references to sequence, chemical, and structural data. It's important to note that the
protein structures defined within PDM, MMDB, and virtually every other protein structure database
are based on assumptions that may not be completely valid. For example, the common assumption
that similar amino acid sequences result in similar protein structures is known to have exceptions.
Comparative modeling is an iterative, multi-phase process. As outlined in
Figure 9-20
, given protein
sequence data, the main phases of the process are template selection, alignment, model building,
and evaluation. 3D visualization is often performed as part of the evaluation phase. The key activities
in each phase of the comparative modeling process are outlined here.
Figure 9-20. Comparative Modeling Process. Not illustrated is the optional
use of ab initio methods at the end of the model building phase to reduce
errors in the computed structure.
Template Selection
The first phase of the comparative modeling process, template selection, involves searching a
template database for the closest match or matches to the new (target) molecule, based on the
target's amino acid sequence. The goal of template selection is to discover a link between the target
