Michael A. Lonetto
lonetto at CGL.UCSF.EDU
Mon Aug 14 11:03:22 EST 1995
>I was wondering whether one has the possibility to specify kind of anchors (i.e
>segments which are known to belong to each other and should not be interrupted
>by the alignement process) in the input file of the multiple alignement
>programs first instead of shifting and moving around the output later on.
>Do clustal, treealign, pileup,... provide this possibility? If not is it
>possible easiliy to add this feature? Is this a feature?
It's always possible to add this feature, provided you can edit the
comparison matrix and sequences: Just pick a symbolt that is not used
in your sequence alignment (eg: X) and give it a really high score
vs itself and a very low or negative score vs. others. Now insert an
"X" in each sequence where you want them to be anchored. If you
want to anchor a segment you may need to add X's at each end.
I've used this method successfully with Pileup to force the location of
a large (300 aa) insertion relative to other sequences. Note that you
are really forcing the alignment, so you should think hard about how
you can justify the location of the anchor.
I haven't used any programs that allow expicit definition of anchors.
If you find any please let me know.
Michael Lonetto // Phone: 415-476-1493 // Email: lonetto at cgl.ucsf.edu
UCSF Depts. of Stomatology and Microbiology, San Francisco, CA 94143-0512
"The name of the game in the entertainment business short-term profit.
This is exactly what what Congress is all about -- how can we get
re-elected in the short term and every other value be damned."
Norman Lear 5/31/95
More information about the Bio-soft