[Bioperl-l] Bio::Tools::Glimmer

[Torsten Seemann]

Mark,

>     I've got one main issue to solve, the rest is just stuff like updating
> the POD.  Torsten Seemann very helpfully added example output for all 4
> formats to t/data.  Looking at GlimmerHMM.out, the first line is
> 'GlimmerHMM'.  However, I think there is a bug in the existing
> _parse_predictions:
> Shouldn't this:
> } elsif( /^(Glimmer\S*)$/ ) { # GlimmerHMM has no version
> be this instead:
> } elsif( /^(GlimmerHMM\S*)$/ ) { # GlimmerHMM has no version

I think it should be what it says, or perhaps now /^(Glimmer(M|HMM))/.
Here's why:

I came onto the scene at Glimmer.pm rev 1.4. At that stage it only
parse GlimmerM. I noted that GlimmerHMM was the same output format as
GlimmerM, except for the first line. So in rev 1.5 I modified the
regexp to match both ie. \S* . This would also hopefully match any
other Glimmer-clone formats that arose. I also fixed the pdocs to say
this, and added tests to t/Genpred.t.
% cvs diff -r 1.4 -r 1.5 Bio/Tools/Glimmer.pm
% cvs diff -r 1.15 -r 1.16 t/Genpred.t

I then planned to extend support to Glimmer2 and Glimmer3. I added the
4 test files (t/Glimmer*.out) but never wrote the code. This is where
you have come in Mark :-)

> I lifted that bit of code to do format detection...we don't have GlimmerHMM
> installed locally, so I'm assuming Torsten's output is correct and the above
> is a bug.  Guess I'll go check bugzilla...

I'm pretty sure my 4 test files are correct - I spent a lot of time
ensuring they were consistent etc, as I was getting very confused with
the different "glimmer" versions!

Hope this all helps,

--Torsten