[Gusdev-gusdev] BioPerl and GUS

[Aaron J. M. <am...@pc...>]

A quick google search reveals that efforts to make GUS more 
BioPerl-aware have (seemingly) been attempted since (at least) 2002.  I 
wonder then why everyone on the mailing lists sounds so startled to 
hear about BioPerl's capabilities, and where the previous efforts have 
lead ...

Steve, you mention that you're afraid BioPerl won't be capable of 
capturing all the information contained in a GUS deployment; though I 
have only a passing acquaintance with GUS, I believe the only aspect of 
GUS that BioPerl would not be able to represent (in some meaningful 
form) is the RAD schema.  BioPerl has both strongly-typed and 
weakly-typed feature/annotation systems, and includes facilities for 
such data related to (in no particular order) Sequences, Assemblies, 
Alignments, Trees, Graphs, Ontologies, Pedigrees, Analyses, Coordinate 
Systems, Databases, Events, Matrices, Phenotypes, Species, Symbols, 
SNPs, etc.  For instance, I would argue that using even just a 
weakly-typed sequence feature system, one can successfully represent 
all sequence-related data (although the prior post about semantic 
difficulties remains relevant; BioPerl may not share the same semantics 
[or be missing it entirely] as GUS, and this will have to be 
programmed).

Also, I don't see GenBank locations as a big issue ("fuzzy locations"), 
since you're only talking about exposing GUS data to BioPerl (which 
doesn't have fuzzy locations) or importing GUS-capable data (which 
presumably will also not have fuzzy locations).  That said, BioPerl has 
an entire suite of location/coordinate handling modules (including 
support for split and/or fuzzy locations, representing the entire 
GenBank location moddel), and mechanisms/policies for "focussing" fuzzy 
locations into exact locations.  You're not the first group to ever 
face such problems.

Now, you will certainly find as you get more into BioPerl that there 
are particularly dusty corners (or ill-documented, but in active use); 
I would argue that these are areas where your developers might spend 
some other their valuable time that was saved by using BioPerl in the 
first place to make improvements to BioPerl, rather than backing away 
and saying "Oh look, the last 5% of the functionality we need isn't 
there, I guess we'll have to do it all ourselves" (cf: the Ensembl 
project).

BioPerl is becoming the lingua-franca of wet-bench biologists who are 
picking up Perl programming for their own lightweight bioinformatics 
tasks.  In 2003, various members of the BioPerl team gave many short 
lectures and workshops are their various host institutions, and at 
national meetings; in 2004, week-long BioPerl workshops were held for 
biologist/programmers at the University of Montreal, CSHL and the NIH.  
Requests for BioPerl training in 2005 are numerous and growing.  
Leaders of the NCI's caBIO and EBI's EnsEMBL projects recognize this, 
and are currently designing their own "bridges" between their code 
bases and BioPerl.  As such, there is much chatter on the BioPerl 
lissts about ways to improve the interoperability of BioPerl objects 
while maintaining current "ease of use" (arguable, of course, but there 
are even plans to make that better as well).  It would be great to have 
more GUS developers as part of the BioPerl community (and, frankly, it 
boggles my mind that there are professional bioinformatics Perl 
programmers who have little or no exposure to, or awareness of, 
BioPerl; remind me to buy more GoogleAds for bioperl.org).

Finally, I won't speak for Lincoln Stein, but I know that the GMOD 
project does want to include GUS in its effort to assimilate existing 
tools, but the current GMOD developers find GUS unapproachable.  I 
would guess that Lincoln and friends would be willing to help out an 
intrepid GUS aficionado get GUS at least talking to the GMOD toolkit 
(presumably leveraging BioPerl to talk ChadoXML; which isn't probably 
as straightforward as it sounds; cf. semantic issues raised earlier).

I'll keep listening ...

Best wishes,

-Aaron

--
Aaron J. Mackey, Ph.D.
Dept. of Biology, Goddard 212
University of Pennsylvania       email:  am...@pc...
415 S. University Avenue         office: 215-898-1205
Philadelphia, PA  19104-6017     fax:    215-746-6697