[Bioperl-l] Re: Integrating caBIOperl with BIOperl

[Ewan Birney]

On Mon, 11 Oct 2004, Jiang, Shan (NIH/NCI) wrote:

>
> Hi Ewan,
>
> I would like to introduce myself. I am a colleage of Gene Levinson at the
> National Cancer Institue in the US. I am the original developer of
> caBIOperl, which Gene presented at BOSC '04. I believe Gene talked to you
> quite extensively during the meeting as well. (Gene asked me to say hi!)
>
> Currently, I am undertaking the task of integrating caBIOperl with
> BIOperl.Gene indicated that you would be a great source to talk to. I am in
> the process of learning BIOperl before deciding how to proceed. So I would
> much appreciate your help in learning BIOperl as well as looking into
> possible ways of integrating caBIOperl with BIOperl.
>

Great - I'm cc'ing this message to the main bioperl list to check I give
you the best advice!

> Let me start asking some questions to start the ball rolling.
>
> 1. Has similar kinds of integration work been done before? If so is there a
> general recommended approach?

The recommendation is definitely to have an caBIOperl "bridge" to Bioperl
objects. The main ones you want to have are Bio::SeqI,
Bio::DB::RandomAccessI and Bio::AnnotationCollectionI and Bio::SeqFeatureI

The "I" means interface (a bit like Java)

In each case you would have wrapper classes that has-a caBIOPerl object
and is-a Bioperl object, for example, imagining the caBIOPerl sequence
object has methods "human_readable_name" and "sequence_as_string" (of
course, they might have something completely different...)

  package Bio::caBIOBridge::Seq;

  @ISA = qw( Bio::SeqI );

  ...
  ...

  # Bio::SeqI isa Bio::PrimarySeqI, and needs to implement
  # display_id. this should give back the human readable name
  sub display_id {
    my $obj = shift;

    # the caBIOPerl method is "human_readable_name"
    return $obj->{'_cabioperl_object'}->human_readable_name()
  }

  # Bio::SeqI needs to implement seq
  sub seq {
    return $obj->{'_cabioperl_object'}->sequence_as_string()
  }


etc etc


This is, BTW, something I am planning to do with Ensembl as - make an
Ensembl-Bioperl bridge.

> 2. Do you have a repository where people can just "donate" their code into?

I would suggest that the caBIO-Bioperl bridge was its own cvs module and
donated into CPAN. You could run the cvs module at Bioperl.org or do it in
your own shop - entirely up to you.

> 3. caBIOperl has its own object model, if the end vision is to integrate
> this model with BIOperl, how should I proceed?

see above

> 4. Can I get access to the CVS repository?
>

You shouldn't need access to the bioperl cvs repository to come up with
some working code - if you want to have the caBio-Bioperl bridge
repository hosted at bioperl.org that's feasible, but probably building
some proof-of-concept classes first off would be great.


A great first step would be if someone could write a script like:


use Bio::caBIOBridge::DBAccess;
use Bio::SeqIO;

# default to well known caBio server
$db = Bio::caBIOBridge::DBAccess->new();

$ca_wrapped_seq = $db->get_Seq_by_id('some_id');

# $ca_wrapped_seq is Bio::SeqI object but is actually a thin wrapper over
# caBIO objects

# Bio::SeqIO is a Bioperl object writer that works with Bio::SeqI
# compliant objects
$seqout = Bio::SeqIO->new( -format => 'EMBL');

# Here we see the bridge in action!
$seqout->write_seq($ca_wrapped_seq);

> I am not sure how familiar you are with caBIOperl. So if you have any
> question, please do not hesistate to ask me.
>
> Regards,
> Shan Jiang
> (Contractor)
>
>