|
|
|
Examples
This section presents some examples that show what you can do with
DomainFinder. It is rather small at the moment - check again later
to see more examples!
The images on this page are kept small to speed up downloading.
By clicking on the images, you can get a full-size version.
Lactoferrin
Lactoferrin is a protein with a particularly clear domain structure;
you can see the domains just by looking at the protein. It is thus
a good example for showing the ideal outcome of a domain analysis.
The example uses the open form of lactoferrin, PDB entry 1LFH.
 |
With 691 residues, lactoferrin is a mid-size protein. To find
the domain structure, 100 normal modes would be sufficient, but
200 have been chosen because the computational cost is negligible.
Note that DomainFinder has slightly increased this number to 222
for implementational reasons.
Four modes have been selected for the analysis. Two or three
would have been reasonable choices as well, but there is little
reason to use just the first and not the second, because the
frequencies are very similar.
The deformation threshold has been set to 600, about the average
deformation energy of the modes. The domain coarseness of 15
was found after some experimentation.
|
 |
The deformation energy distribution already shows three particularly
rigid regions (the blue ones), which are the cores of the dynamical
domains. As expected, the hinge region shows the highest flexibility.
There is a continuous transition between the low-deformation cores
and the high-deformation hinge, which shows that clean-cut domain
delimitations do not exist.
The deformation histogram is typical for a protein with well-defined
domains: a peak at low deformations corresponding to the mostly
rigid domains, and a fast decrease for higher deformation energy
values.
|
 |
The three domains come out as expected. The motion parameter plot shows
that the motions of the three domains are indeed well separated, and
that the parameters within one domain are very similar, at least for
the first two modes. In the third and fourth mode, the bands corresponding
to the domains widen, but do not split into subbands. This means
that the domains do not split into subdomains at higher frequencies,
but move with a higher overall deformation.
|
 |
The red domain in more detail. The plot at the bottom shows the similarity
of each part of the domain relative to the two parts which are most
similar to each other. This is the ratio used for defining the
domain coarseness. In this domain, the most dissimilar part requires
a coarseness of 13 to be included in the domain, i.e. if the coarseness
were lowered to a value below 13, one part would fall out of the
domain. The remaining ones are much more similar; the next highest
coarseness values is five.
The residue list on the right shows the residues that are colored red.
This is not a precise definition of the domain, but rather a list of
sufficiently many residues to describe the domain. Other residues
might have similar motion parameters and hence belong to the domain,
but are not included because of the arbitrary division of the protein
into cubic parts. This is not a problem because domains are large and
not precisely defined, whereas the wrongly eliminated residues are
always few in number.
|
|
