[MISC] [MOBY-l] A simpler Ontology question

[Mark Wilkinson]

I'll tackle your question first (sorry Ben), since it is an easier one
:-)

There are several directions of possible answer to your question,
perhaps none of which will please you, but they are the answers:

1)  In the current ontology you *cannot* get a child of VirtualSequence
that does not have a Length.

2)  The ontology is open - if you wish to define a sequence object that
does not carry length, you are free to do so.  Length was at the root of
the sequence hierarchy to avoid situations where you unexpectedly find
yourself downloading the entirety of Chromosome 1, and all of its
subparts, from genbank because you simply chose to get the sequences of
a lists of gi numbers.  Retrieving the VirtualSequence object first
allows you to do a quick check of whether the sequences you are about to
retrieve are of a reasonable size (big or small), and then only submit
your query for the "real" data when you have done this pre-screening. 
(it was also based in the desire to be somewhat like BioPerl, where
there are sequence objects that have only length - it just seemed like a
good idea at the time)

3)  It isn't difficult to determine the length of a string... I don't
know if I agree with your assessment that it is "expensive" to compute
the size of a sequence.  The length field in the object does add a bit
of message-size overhead, but it is minor compared to the sequence
itself.

4)  We also have a FASTA object, if that fits your needs better, which
is just straight text.


Your choice among these options is entirely up to you.  I am always
loathe to design new objects that are effectively identical to existing
objects, because that is what got us into the state of
non-interoperability that we are already in!  

I would argue that interoperability demands sacrifice... and if that
means adding a line to a script to calculate the length of a string,
then that is a small sacrifice. 

But again, the ontologies are open, and you are free to register
whatever objects you wish :-)

Cheers!

M


On Thu, 2004-07-29 at 12:51, Hudson, Thomas C. wrote:
> Here's a simpler question about the ontology I don't understand:
>   As far as I can see, the root of the sequence hierarchy is the
> VirtualSequence, which is a MOBY object with an integer Length. So
> there's no way to have a sequence object without the length attached?
> 
> Tom, who has a database that doesn't store length, meaning length is
> expensive, and wants to just use accession numbers out of the first
> database to do lookups in another database, meaning length is
> unnecessary.
> 
> _______________________________________________
> moby-l mailing list
> moby-l at biomoby.org
> http://biomoby.org/mailman/listinfo/moby-l
-- 
Mark Wilkinson (mwilkinson at mrl.ubc.ca)
University of British Columbia iCAPTURE Centre