Infernal 1.1 seems to have two primary improvements that might make it worth using: 1) better handling of end-gaps and 2) significant alignment speedup. Note that multithreading is also available without using MPI (not sure if that was the case in the earlier version).
Unfortunately, the interface has changed a bit:
cmalign --merge is deprecated; use esl-alimerge --dna -o $merged $ref_sto $sto instead
The default .sto output includes rows containing posterior probabilities; suppress with --noprob
There's now an option to write alignment scores to a file, so no need to capture stdout.
It isn't clear whether the default alignment params are still appropriate - I have been using cmalign -o $sto --sfile $scores --noprob --dnaout $profile $fasta but more investigation may be worthwhile.
Note that alignment profiles now have a new format and need to be updated with cmconvert
Infernal 1.1 seems to have two primary improvements that might make it worth using: 1) better handling of end-gaps and 2) significant alignment speedup. Note that multithreading is also available without using MPI (not sure if that was the case in the earlier version).
Unfortunately, the interface has changed a bit:
esl-alimerge --dna -o $merged $ref_sto $sto
instead--noprob
cmalign -o $sto --sfile $scores --noprob --dnaout $profile $fasta
but more investigation may be worthwhile.Note that alignment profiles now have a new format and need to be updated with
cmconvert
More illustrations of integrating infernal into the pplacer pipeline can be found in https://github.com/fhcrc/yapp