Biomedical Engineering Reference
In-Depth Information
Given a library of 3D structure candidates, the most promising structures are filtered from the less
promising structures. A common method of filtering is to use the Metropolis Algorithm to identify the
most stable molecular conformations. This method is based on the assumption that the native
conformation of a protein is the conformation with the lowest free energy. This method works for
many proteins, but not others. For example, prions (proteinaceous infectious particles) are
energetically unstable and are more stable in their disease-inducing state. Similarly, not all proteins
go to a lowest energy state in nature. Some proteins fold independently, whereas other proteins
require chaperones—molecules that act as catalysts—to fold.
Once the top protein structure candidate is identified, it is validated and visualized. Validation
typically refers to comparing the predicted protein structure with a structure derived from NMR and X-
ray crystallography experiments. That is, ab initio methods are still being perfected. Eventually, ab
initio methods may provide enough accuracy and handle molecules large enough to supplant
experimental methods. However, for now, validation involves assigning a figure of merit to the
predicted structure, based on comparison to the gold standard. The most often-used figure of merit
in protein structure comparison is the root mean squared deviation (RMSD). The calculation for
RMSD, expressed in Angstroms, is shown in
Figure 9-18
.
Figure 9-18. Root Mean Squared Deviation (RMSD) Calculation.
N
= number
of atoms.
D
= the distance in Angstroms between corresponding atoms in
the experimental and predicted protein structures.
