[DAS2] modification to /type request & response
suzi at fruitfly.org
Mon Apr 11 19:52:13 UTC 2005
On Apr 11, 2005, at 12:35 PM, Chervitz, Steve wrote:
> Allen Day writes:
>> I'd like to remove the constraint that all types inherit from a type.
>> makes sense to keep a constraint like this in place for genome
>> features, but not any other record type.
> One aim of DAS/2 is to disallow the "anything goes" policy for feature
> typing, and to constrain them to be part of some ontology or related
> to a
> term in an existing ontology so that DAS/2 clients have a chance of
> sense of them.
> The DAS/2 grant contemplates how to extend an existing ontology, but
> discuss plans for incorporating types that don't fit into any
> ontology. Here's the relevant snippet from the grant:
> "Using a standard sequence ontology does not preclude extending the
> ontology. One of the advantages of using RDF and OWL to specify
> is that there is a formal mechanism for anyone to extend an existing
> ontology, which gives rise to distributed ontologies. This ability to
> extend ontologies is built on the same URI linking concepts previously
> discussed regarding feature ids. Thus for example someone with
> knowledge of some the biology of repetitive sequences, for instance,
> may be
> annotating repeat data and want a much finer classification of repeat
> than is specified in the reference ontology. They can define an
> extension of
> the existing ontology via RDF that refers to the existing ontology and
> specifies the extension (such as LINE-variantX isa type of LINE), then
> up DAS sequence features that refer to this extended ontology rather
> the reference ontology. But because the referenced ontology formally
> extends the standard ontology, a DAS client could still be aware that
> LINE-variantX is a type of repeat, if for example it wants to display
> repeats using the same color. Many software tools are now available
> to make
> such ontology extensions easy."
> Regarding protein feature types, I think the SO team plans to put them
> SO/SOFA since they summarize it as, "a set of terms used to describe
> features on a nucleotide or protein sequence." But there is no support
> proteins in SO/SOFA yet. Suzi?
Yes, that is pretty much where it stands. There is the intention of
doing this, but we have not yet gotten to that point. Including the
basic structural descriptions of a protein (alpha helices, beta
seems an obvious new addition and easy enough to do.
I strongly agree with Steve here that we absolutely must stay
away from ad hoc types. They are the sirens on the rocks:
alluring, but lethal.
> It does seem like there should be an avenue for people to say, "Here's
> a new
> type that doesn't belong to any established ontology (or if it does,
> I'm not
> sure where it goes)." Could such an avenue be opened without having it
> Interesting google result: Searching for "protein sequence feature
> gives about 35,000 hits, the second of which is the DAS/2 spec
> DAS2 mailing list
> DAS2 at portal.open-bio.org
More information about the DAS2