[Biojava-l] IDL

[Matthew Pocock]

Martin,

David Martin wrote:

> I have a few pennys worth to throw in here..
>
> Firstly I think that IDL is definitely the way to go to get a good design
> and interoperatability between the various bio* development efforts.
>

I too am looking forward to this all working out nicely.

>
> Now for the fun..
>
> At present the IDL is focussed exclusively on linear sequences of amino
> acids or nucleotides (of whichever flavour). This may prove to be a
> mistake in future as not all biological sequences are linear.
>
> I am thinking of the bio* IDL being useable across the whole spectrum of
> biocomputing rather than basic sequence analysis.
>

The current module is under org.Biocorba.Seqcore - it is specificaly aimed at
letting us pass sequences arround. This is a realy *good* thing to be able to
do (I look forward to corba-blast) but relies on not imposing any far-reaching
design philosophies on the servers or clients. I don't know what other IDL
modules will go under Biocorba as time goes by.

>
> If we start at the basic level, everything in molecular biology (in its
> broadest sense) is a unit at some scale. It typically forms part of a
> sequence of such units where one or more units precede it and it precedes
> one or more units.
> We are looking at the nodes of a unidirected graph.
>

So, I would be happy to see these kinds of concepts represented in biojava - I
think that things like pathways will become standard stuff for biologists to
look at - and not just on a big sheet of paper. These mabe are not things that
a naive CORBA server would handle very well, but they would work as java
objects.

Some of the graphy things I have seen try to make everything part of the graph
with the concequence that objects implement very unnatural interfaces. Of
course, I am sure that we could do better :-)

> Maybe you already have very good reasons for not taking this approach, but
> it is occasionally neccessary to do something a bit different to extend
> our capabilities rather than biojava being a clone of bioperl (worthy a
> goal though that is)

Not quite a clone - we already handle single sequence DP and Thomas is working
on two sequence DP (alignment stuff). But, a lot of the BioPerl ideas were
basicaly sound, and seem to work (the io modules for example), so I took them.
Our sequences are quite a bit more flexible, as are our features & locations,
but they look prety much the same as they are doing the same sort of job.

Anyway, with any luck biojava will evolve into a good foundation for all the
bio classes we need. I look forward to pathways making an appearance.

>
>
> ..d
>
>  >
>
> ---------------------------------------------------------------------
> *  Dr. David Martin                  Biotechnology Centre of Oslo   *
> *  Node Manager                      Gaustadalleen 21               *
> *  The Norwegian EMBNet Node         P.O. box 1125 Blindern         *
> *  tel +47 22 95 87 56               N-0317 Oslo                    *
> *  fax +47 22 69 41 30               Norway                         *
> ---------------------------------------------------------------------
>
> _______________________________________________
> Biojava-l mailing list  -  Biojava-l@biojava.org
> http://biojava.org/mailman/listinfo/biojava-l