[Chimera-users] structure based sequence alignment
pett at cgl.ucsf.edu
Thu Jun 24 13:50:26 PDT 2010
On Jun 24, 2010, at 7:39 AM, Sona Vasudevan wrote:
> Hello Eric,
> Thanks for all the great features you have added in the last couple
> of years since I was last in touch with you. This is a fantastic tool!
Thanks! Glad you like it.
> I need some help with the alignment editor.
> I am trying to import an alignment in fasta format (fasta aligned)
> (see attached). The alignment as you can see has three structures. I
> use Load structures from the sequence
> window and it brings in all the chains corresponding to teh
It does, but there's a subtlety here that I should point out. The
first two sequences in the alignment are for 1XVA, chain A and chain
B. Chimera opens two copies of 1XVA and associates one copy with each
sequence, but it associates chain A in both, since the sequences of
the two chains are identical. It doesn't really look at the sequence
name to try to figure out which chain to associate, just how well the
sequences match and in this case they both match perfectly. By
hovering the mouse over the sequence name you can see which structure/
chain it is associated with.
To change the association of the second sequence to use chain B, you
need to go to the alignment's Structure menu and choose
"Associations...". Thus will bring up a panel showing the association
of each structure to a sequence. Change the association of "1xva
(#1)" from "chain A" to "chain B" and then click OK.
> I then use Matchmaker to align them. I have a few questions:
> 1. Does the match maker use the info from the imported alignment
> since the structures are already aligned? In other words is the
> structural alignment I am seeing the saem as that was imported.
No. MatchMaker is its own tool and doesn't use any open alignments.
After all, you could have several alignments involving those
structures open. If you want to superimpose the structures based on
your alignment, use the alignment's own "Structure->Match" menu item.
That brings up a dialog offering a variety of options as to how to
carry out the match. I like using the "Iterate" option, but it's
totally up to you.
> 2. Is there a way to just output the superposed co-ordinates without
> the chains that are not being used?
There are probably several ways. One that comes to mind is to click
on one of the alignment's "Consensus" letters to select all the
residues in that column. Then click into the graphics window and hit
the up-arrow key to expand the selection to the entire chain. Then
either choose Select->Invert or hit right-arrow to invert the
selection -- now the unused chains are selected. Actions->Atoms/Bonds-
>delete will delete them. You can then write out what's left.
Keep in mind that if you want to get model 2's coordinates as
superimposed with model 0's original coordinates (for instance) then
make sure to use the "relative 0" option in the "write" command or
check the "Save relative to" option in the Write PDB dialog and choose
model 0. Or you could just write out all models current coordinates.
They will all differ from their originals, but they will be
> 3. Looks like I can add sequences to the alignement. Can I add a
> fasta file with many sequences?
No. It is on my to-do list but I haven't gotten to it yet. I'll open
a feature-request ticket in our Trac database with you on the
recipient list so you'll know when it gets implemented. Sorry!
> Thanks so much! I have alignment files for 100 families which are
> high-qualityas they were manually aligned with Cn3d and I would like
> to import them into chimera for further analysis.
> Thanks so much for your time!
UCSF Computer Graphics Lab
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Chimera-users