OK then. We will continue to build out this list as needed and be
explicit about what exactly we are localizing.
On Jun 9, 2007, at 7:18 AM, Chris Mungall wrote:
> I actually quite like your idea of explicitly representing types of
> protein in your ontology - this makes a lot of sense for you since
> you (BIRN) may be focused mainly on a more limited set - SNCA for
> example. You probably want a lot of control over the representation
> of these important proteins and depending on the situation you will
> want to refer to:
> The SNCA gene in mouse
> The SNCA gene in mammals
> The generic product of SNCA
> A specific protein as produced by SNCA (say, uniquely identified
> by the AA seq)
> SNCA spliceforms
> two spliceforms that are different in sequence but similar enough
> to be grouped into a supertype
> A post-translationally modified SNCA protein
> The synuclein family of genes
> A variant form of the SNCA gene
> SNCA driven by some other promoter in a transgenic animal
> An aggregate of SNCA proteins
> Now this could be done by using a mish-mash of existing databases,
> plus a set of conventions on how we are to interpret an ID to such
> and such a database. But part of the point of using ontologies is
> to remove the need for these conventions and have a shared
> unambiguous representation.
> Another approach is to represent everything explicitly using a
> database schema like Chado, which allows you to encode explicitly
> the relationships between the various entities above (and given
> coordinates, sequence, etc). But there is still a certain amount of
> convention involved in how to unambiguously represent these things.
> For BIRN there is some value in shifting as much as possible to a
> pure ontological representation, with terms/classes representing
> all the entity types denoted above (plus xrefs to the various
> databases of course). Not everything would have to be pre-
> coordinated of course. The kind of questions you want to ask are
> very spatial in nature, so it's important to distinguish between
> the gene, the protein, it's modified form, all of which have
> different locations, especially with the particular anatomy of the
> brain (sorry if I'm mangling your description of this). We'd have
> to formalize the relations that can hold between these kinds of
> entities, SO provides a start here.
> But this approach will be hard to scale with current tools beyond a
> small set of genes. For the rest of us I think we may have to make
> do with the database IDs + conventions approach in the short term,
> in which case it's better to formalize those conventions sooner
> rather than later. One way of doing this is to provide a means to
> get an ontological representation of a "genomics entity" in OBO or
> OWL - this could be done in bulk or via some kind of webservice.
> More on the expt vs conclusion distinction later....
> On Jun 8, 2007, at 10:36 PM, Maryann Martone wrote:
>> Hi Nicole:
>> Chris and I discussed this a bit in Innsbruck. It was recommended to
>> me that I use the Uniprot ID for proteins, although we have been
>> ennumerating some proteins in SAO. I'm not happy in a strict
>> philosophical sense just using the protein ID as if it were a single
>> protein, but for practical purposes, I'm not sure how much more
>> nuanced we need to be for the moment. He and also discussed the need
>> to keep separate the experimental observation, i.e., staining for
>> protein X is increased, vs the conclusion: expression of protein is
>> increased. In fact, we can't always assume that increased staining
>> is increased expression. In our community, we tend to use expression
>> to mean "presence" rather than a statement about translation or some
>> other process. So perhaps we ought to move to some more neutral term
>> like "increased protein amount". WIth respect to the experimental
>> results vs the conclusion, I think I like the idea of using
>> experimental terms when describing image data and stating the
>> conclusions when we annotate the study.
>> On Jun 8, 2007, at 12:50 PM, Nicole Washington wrote:
>>> So, out of our montly Phenote Working Group meeting there arose a
>>> to start list-discussion of how to represent expression (DNA,
>>> RNA, and
>>> Protein) in EQ formalism, as well as how to go about it in
>>> Phenote. This
>>> topic is becoming increasingly important for the Zfin, Xenbase, and
>>> communities, and probably many others I am unaware of.
>>> There was a suggestion to take this to another email list, such as
>>> phenote-expression or somesuch. Shall I go ahead and create this
>>> list, or
>>> do we want to leave it on the obo-phenote mailing list?
>>> Anyway, Are there any groups on these lists that already have a plan
>>> in-hand for how to tackle this problem? I'll throw out a discussion
>>> *When trying to capture that "protein X expression is increased",
>>> do you
>>> represent the (a)uniprot or local database ID as your entity and if
>>> how do you represent the "pool" of proteins vs the individual
>>> itself? And really, if looking at microscopy or westerns, do you
>>> know that the protein *expression* is increased (vs. turnover
>>> or do you record simply that the protein concentration is increased?
>>> Discuss away...
>>> This SF.net email is sponsored by DB2 Express
>>> Download DB2 Express C - the FREE version of DB2 express and take
>>> control of your XML. No limits. Just data. Click to get it now.
>>> Obo-phenote mailing list
>> This SF.net email is sponsored by DB2 Express
>> Download DB2 Express C - the FREE version of DB2 express and take
>> control of your XML. No limits. Just data. Click to get it now.
>> Obo-phenote mailing list