[ssml] Slides to demonstrate conservation of protein structure?

[Mensur Dlakic]

This will be a shameless self-promotion, but at least the example is
completely worked out and fairly convincing.

I have recently made a model of a methyltransferase based on ~10% sequence
identity with the template. When the same two sequences are attempted to align
with Blast2seq, it finds nothing significant to align even with E-value
increased to 50000. ClustalW aligns them so wrong that it's laughable - the
real domain never even gets included in the alignment with the structure.

Anyway, the structure of my protein came out and it aligns with the model over
107 out of 120 residues with an RMSD of 2.4A (over 157 residues and an RMSD of
3.55A with the original template). I already have all the files on my computer
at work to support this example (structural and ClustalW alignments, figures
of superimposed models and structures) but they are not accessible to me until
Monday.

If you want to try your hand at this, I suggest you go to the SCOP site, pick
the fasta file of all sequences with less than 20% identity (or even lower, to
make sure ClustalW will fail), select two structures from it (TIM-barrel
proteins are known to have similar structures with very low sequence identity
- their SCOP codes are c.1.*), then align their sequences with ClustalW (or
your method of choice) and align them structurally using any of the available
public servers.

Cheers,

Mensur


On Fri, 8 Oct 2004, Dan Bolser wrote:

|>
|>Hi,
|>
|>I would like to demonstrate the well known principal that protein
|>structure is more conserved than protein sequence in evolutionary
|>time. I am presenting to a group of structural biologists, so I
|>want to emphasize the bioinformatic perspective.
|>
|>My idea for a series of slides is this
|>
|>1) show two clearly structurally similar (same topology) protein domains.
|>1.1) show that the two sequences are very different, and result in
|>     a 'bad' sequence alignment (somehow visually).
|>
|>2) Show a nieve (and hopefully bad) structural alignment of the two
|>   domains based (nievely) on the sequence alignment.
|>
|>3) Show a pure (and simple) structural alignment of the two domains.
|>3.1) Show the sequence alignment based on the good structural alignment.
|>
|>
|>My idea is to convey the importance of structure in recognizing distant
|>homology between domains.
|>
|>I want to use a real example using real tools, but I am not sure how /
|>which tools to use.Any suggetions? Good demo / bad demo? Naturally I need
|>this done by yesterday, so the simpler the tools the better.
|>
|>I will be happy to post the slides somewhere appropriate to share this
|>work with the world.
|>
|>All the best,
|>Dan.

==========================================================================
| Mensur Dlakic, Ph.D.              | Tel: (406) 994-6576                |
| Department of Microbiology        | Fax: (406) 994-4926                |
| Montana State University          |                                    |
| 109 Lewis Hall, P.O. Box 173520   | http://myprofile.cos.com/mensur    |
| Bozeman, MT 59717-3520            | E-mail: mdlakic at montana.edu        |
==========================================================================