[Bioperl-l] dpalign, local for one sequence, global for the other?

George Hartzell hartzell at alerce.com
Wed Oct 1 20:32:48 EDT 2008


I need to produce an alignment between a hunk of genomic sequence (in
the sense that it hasn't had introns edited out or anything) that's on
the order of a 1000 bases to a genome/chromosome/fragment of similar
genomic sequence.  In an ideal situation they'll be the same,
differences will come from variations in the sources (e.g. the hunk
might have been clipped out of genome revision X and the current
genome might be X+i, or the hunk might have come from a paper (who
knows where it came from...).  Nothing across species or across
evolutionary time or anything fun.  I'm happy to narrow the region of
the genome hunk down using some/an/... heuristic first to avoid
running dp against an entire chromosome.

I need the alignment to account for all of the bases in the hunk.  In
dynamic programming terms, if the hunk is along the vertical axis, the
path through the matrix would have to run from the top to the bottom
(or vice versa).  The projection of the path onto the
horizontal/genomic axis can start/end wherever.

I'd like to not write this [again] and was hoping to use the bio-ext
dpalign stuff.  I'm hopeful that "ENDSFREE" is just what I need, but
from the docs I'm not convinced that it is.  A more pessimistic
reading makes it sounds a lot like a local alignment.

Can anyone out there who's familiar with the dpalign code tell me
whether it can do what I need?  Out of the box?  With modifications?

g.



More information about the Bioperl-l mailing list