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

Mark Johnson johnsonm at gmail.com
Thu Apr 12 15:11:18 EDT 2007


> So to reiterate, I'm suggesting that the module also parse out the frame and
> score information from Glimmer output.  I take back my suggestion of
> overriding the source / primary tags through the module as this can easily
> be done post-parser.  Other annotations can also be edited post-parser
> easily enough.

The reason the source tags are what they are for my addition(s) is
that the output from glimmer2/glimmer3 does not include a version
string.  You can figure out the major version from the output
formatting, but that's about it.  Also, being my first significant
contribution, I wasn't out to break new ground.  I did what some of
the other gene predictors seem to do, and what the existing code
already did.  Maybe there should be a method to pass in the exact
version if you know it.  Further than that, I think the Glimmer module
should stay consistent with what the other gene predictors do.  No
reason, though, that they couldn't *all* be enhanced similarly, if you
want to be able to further control the source tag.  8)

Part of the reason I didn't parse out the frame / score info for
either glimmer2 or glimmer3 was that I didn't need it.  The other part
being that my regexp kung-fu is nothing special.  This sounds like a
no-brainer to me.  Extend the regexps to capture it and tag it (and
the tests).

As far as the ORFs go, I guess you could just use
Bio::SeqFeature::Generic to represent them.  I haven't been keeping
track of the relevant feature/annotation interfaces, but maybe there
should be some kind of relation between the ORFs and predictions?

The glimmer3 detail file is a little trickier.  The least disruptive
thing to do, interface wise, might be to specify that as a seperate
input via an argument to the constructor.  Then you've got *two* input
files, and are going to have to override the automagic stuff that
expects one input file and takes care of it all.

As far as process, I just got on the list and started pestering
people, and they haven't thrown me off yet.  8)  I'm afraid that
you're going to find that while people are happy to discuss
implementation details, when it comes time to fire up the editor,
you're usually on your own, if it's an enhancement.

I'd love to work on Bioperl more, but so far, it's only been for what
I need for my job.


More information about the Bioperl-l mailing list