CASA is a hacked up version of AMAP for producing consensus alignments through sequence annealing. As it currently stands, CASA takes a set of sequences in fasta format and a set of arbitrarily-defined pairwise alignment probability matrices to produce multiple sequence alignments. It also includes a script, consent.pl, which takes a stockholm format file containing multiple alignments over the same set of sequences and produces count matrices normalized by the number of alignments present, suitable for feeding to CASA. Besides the perl script, there isn't a significant amount of new code, the new name is largely to avoid confusion.
CASA is available via anonymous CVS as follows:
export CVS_RSH=ssh cvs -z3 -d:ext:firstname.lastname@example.org:/usr/local/cvs co casa
CASA also requires DART Perl.
Sample sequences and matrices can be found in the subdirectory postprob_sample. To run it try:
./casa -prefix postprob_sample/postalign postprob_sample/nanos.fasta
To produce a consensus alignment from a stockholm format file run:
./consent.pl yourstockholmfile.stock ./casa -prefix yourstockholmfile.stock yourstockholmfile.stock.fasta
consent.pl will leave the normalzed count matrices in the current directory named things like yourstockholmfile.stock-seqa-seqb
consent.pl also has a -v verbose mode which will output a good amount of log data to stderr.
CASA is in active development, and the implementation may change significantly from what is described here. If you are interested in using and/or (especially!) contributing to/extending CASA or consent.pl, please contact Lars.
-- Lars Barquist - 05 Mar 2008