Re: Immunity of SW statements to changes in location - data integration use case

M. Scott Marshall wrote:
>  From the recent threads, I get the impression that we
are trying to do 
> combine two functions into the URI:
> 
> 1) the unambiguous *identification* of a given concept
in our own RDF
> 2) retrieve associated data records from the same URI
> 
> Although 2) seems like a nice convenience feature, we
have the problem 
> of ambiguity that Alan pointed out: It should be
possible to distinguish 
> statements about a particular data record associated
with a concept from 
> statements about the concept.

What kinds of statements would you want to make about the
"data record" 
(not tied to any specific representation) that do not apply
to the 
"concept" (not tied to any specific
representation, either)?


> See, for example, http://beta.uniprot.org/uniprot/?query=p53&sort=sco
re 
> . "Which data record for P53 did you mean?"

This is one abstraction level above UniProt, we consider P53
a "protein 
family": http://beta.uniprot.org/uniprot/?query=famil
y:"P53+family", 
unfortunately we don't have unique identifiers for those, at
the moment.

But you'll find various domain and family databases that
talk about P53, 
e.g. InterProt [htt
p://www.ebi.ac.uk/interpro/IEntry?ac=IPR002117].

> What kinds of statements would you want to make about
the "data
> record" (not tied to any specific representation)
that do not apply
> to the "concept" (not tied to any specific
representation, either)?

Properties of data record: title, date of submission to
database, submitted by, curated by, edited by, newer version
of...
Properties of "concept" (or rater physical
entity): molecular weight, molecular structure, sequence,
translated from mRNA, part of protein complex...

Can a database record have a molecular weight or be part of
a protein complex? Can a protein be curated and added to a
database?

-- Matthias