Biomedical Engineering Reference
In-Depth Information
inheritance, Mendel focused on readily visible, obvious traits of pea plants that he could definitively
recognize and categorize. He and his attendants could unequivocally determine whether the peas
were round or wrinkled, if the plants were tall or short, and whether the flowers were white or purple.
He avoided measuring non-visual parameters such as weight days to flowering. Perhaps Mendel's
findings would have been noticed by his contemporaries if he had included graphics in his publication
similar to the type currently used in textbooks to describe his experiments. Similarly, Thomas Morgan
decided to use
Drosophilae melanogaster
to understand genetics, evolution, and development, in part
because he could easily observe visual changes in the flies, such as eye color. It also helped that he
could house thousands of experimental subjects in a few jars.
In bioinformatics, the majority of data is in an abstract form that needs visualization technologies to
enhance user understanding. This need is most pronounced in the areas of sequence visualization,
user interface development, protein structure visualization, and as a complement to numerical
analyses, especially statistical analysis. In each application area, the rationale for using graphics
instead of tables or strings of data is to shift the user's mental processing from reading and
mathematical, logical interpretation to faster pattern recognition.
A common activity in protein structure prediction is comparing the predicted structure with one
experimentally determined by X-ray crystallography and the same Nuclear Magnetic Resonance
Imaging (NMR, also referred to as MRI or Magnetic Resonance Imaging) technology used in clinical
medicine. The degree of similarity is often expressed as a Root Mean Squared Deviation (RMSD)
figure, which represents the distance between the corresponding atoms in each molecule. Similar
structures typically have an RMSD in the 1-3 Angstrom range, with larger RMSD values
corresponding to greater deviations in similarity. However, as the size of the protein increases, the
minimum RMSD to qualify for what is considered a good fit increases. Whereas an RMSD of 10
Angstroms would be considered a poor fit for a small protein, it might be considered excellent for a
longer protein with several hundred amino acids.
Consider the challenge of comparing the protein structures depicted in
Figure 5-2
. Although the
RMSD value provides a quantitative measure of closeness of fit, visualizing the overlap of structure
pairs is more intuitive. In addition to being more intuitive than simple RMSD values, the visualization
provides additional information—just as the graphics in
Figure 5-1
add value to a simple tabular
listing of clinical data. Even though the RMSD values for the four pairs of structures is identical, there
is clearly a difference in what the value represents in each case.
Figure 5-2. The Challenge of Structure Comparison. Each pair of protein
backbones has the same RMSD value, but different relative amounts of
structure similarity. Visualization, together with the RMSD value, provides
the best indicator of structure similarity. A—Uniformly Distributed
Difference; B—Localized Difference; C—Significant Difference with Few
Atoms; D—Small Difference with Many Atoms.
