'Dialign 2' seems to be useful for this sort of problem
in article <7b7fc1$8jq at net.bio.net>, Alan Wilton <dna at unsw.edu.au> wrote:
> Dear sequencers,
>> Can anyone recommend a program that will extract a region of high
> similarity from a bunch of sequences of different length and then do
> multiple alignment on them?
> Even an 'in house' protocol on how to go from BLAST matches,
> then extracting regions of similarity to put into a
> multiple alignment program would do?