[Bioperl-l] Bio::Tools

[Jason Stajich]

First off Rob - this is not to dissuade you from trying to mop up the
dusty corners of Bioperl but I am wary of just ripping things apart and
putting them into new dirs unless it is really going to make things
better.

I was definitely struck by the potpourri approach of the Tools dir - so we
tried to corall things to Run which has to deal with actually running.
Aaron and I have started putting phylogenetically related things in
Tools::Phylo and so forth.  I am of the mind to leave things as they may
and to just start putting new things in the proper place.

more below

On Mon, 23 Jun 2003, Rob Edwards wrote:

> Hi all,
>
> As others have pointed out Bio::Tools is a real mess of modules and it is
> hard to know what is there.
>
> I propose the following changes based in part on what Hilmar and Heikki (and
> others) suggested.
>
> 1. Move all the parsers out of Bio::Tools. It seems to make (more?) sense to
> put these in a separate directory, e.g. Bio::Tools::Parsers, and perhaps
> distribute them with Bio::Tools::Run instead of core. There are at least 35
> parsers. It seems that if you don't have the Run modules you are unlikely to
> use the parsers (I know this isn't always true if you have run the app
> without using bioperl, but it is an approximation).

Well - we quite clearly want to keep these separate for folks who want to
run analysis on a farm without using the bioperl wrappers.  I still fully
feel that parser should live in bioperl-live as it is hard enough to sync
the schedules and dependancies of these for a release.  The parsers are
our strength and I don't want to impeede their release.

>
> 2. Use
> Bio::Tools::Analysis
> Bio::Tools::Analysis::Nucleotide
> Bio::Tools::Analysis::Protein
>
> The last one is currently in the cvs but has a wrapper for NetPhos that
> should be in Bio::Tools::Run
>
> In that case, several modules could be moved out of Tools such as:
>
> ECnumber.pm -> Analysis/Protein
> Gel.pm -> Analysis/Nucleotide
> OddCodes.pm -> Analysis/Protein
> SeqStats.pm -> Analysis
> SeqWords.pm -> Analysis
> SiRNA.pm -> Analysis/Nucleotide
> Sigcleave.pm -> Analysis/Protein
> pICalculator.pm -> Analysis/Protein (but I think this should be part of
> SeqStats)
>
> 3. Aren't the blast parsers being folded into SearchIO, in which case they
> could be moved to the attic at somepoint?
>

Bio::Tools::Blast is a deprecated module  - it will go in the attic for
1.4.   BPlite is still going to be with us for a while because it is
lighter and faster than SearchIO and is still used by a lot of people.  I
don't know what the schedule

The HMMer parser is also on the chopping block since I think SearchIO is
doing its job now.  I think things the BLAT parser could be folded in
SearchIO with a little work as well.

However, I don't think it is as simple as all alignment parsers go in
SearchIO - we are going to eventually have a genewise alignment parser in
SearchIO which will produce Bio::Search objects and we already have a
Bio::Tools::Genewise which produces gene objects.  I don't envision this
all be served by the same object even if the implementation framework is
shared.

> This is quite drastic and would probably break a few things, but I think it
> would make Bio::Tools a lot cleaner and clearer in the long run.
>

Well, we've resisted doing this type of cleanup for exactly the reason
that a lot of things will break, including documentation and tutorials
that exist and are being put togther.  So I'm not convinced about the
potential tradeoffs here just yet.  What do users, who've gotten this far
in reading this mail want to see?

Not to say I don't like a tidy house - just that maybe better effort in
documenting what is in Tools is more productive than just putting them
into directories.  I could be acting like a stick-in-the-mud here so let
me know.

> Rob
>
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at portal.open-bio.org
> http://portal.open-bio.org/mailman/listinfo/bioperl-l
>

--
Jason Stajich
Duke University
jason at cgt.mc.duke.edu