Peter Rice (pmr at sanger.ac.uk) wrote:
: The problem is not with prettypep.cmp in this case, although I have seen
: it cause confusion over consensus sequences from time to time.
: The cause is the way in which PRETTY calculates the consensus.
: Luckily, I believe a solution is available.
: In EGCG, the PRETTYPLOT program started life as a much enhanced version
: of PRETTY. PRETTYPLOT produces a boxed alignment as graphical output,
: but for historical reasons it is also capable of generating text
: output in the same style as PRETTY.
: The consensus calculation in PRETTYPLOT is much improved. For your case,
: the /NOCOLLISIONS option allows more than one residue to match the consensus
: where all are above the plurality you specify.
: The text output of PRETTYPLOT is not an obvious choice, and as far as I
: am aware nobody has been using it so it may need testing and some
: changes. In particular, I suspect the /DIFF option may upset the way
: the consensus is stored.
: EGCG is available from ftp.sanger.ac.uk in directory pub/pmr/egcg8
: or pub/pmr/egcg8vms
I had another suggestion from Alan Heath at NISBC, which does not require installing
EGCG. Quoting Alan:
This has also fooled many of my colleagues (and myself) in the past. You can
leave the symbol comparison table unaltered, but use PRETTY/THRESH=1.5 to force
all differences to be shown up.
In the past, I also noticed a slight bug with nucleotide sequences, when the
presence of an x ambiguity code caused PRETTY to assume everything in that
column was the same. I don't know if this has been fixed in version 8.
End of quote. The /THRESH=1.5 switch does indeed do the job.
Thanks to all for your help!