From: Suzanna L. <su...@be...> - 2007-04-23 19:05:44
|
I do like interstitial a lot. It is very evocative of the role played in putting ontological pieces together. What I don't quite understand is the relationship to the RO. May I assume that one uses the RO to build the interstitialOs? a bit more below. On Apr 23, 2007, at 9:56 AM, William Bug wrote: > Wonderful! > > To join Suzi in over-simplifying, I think there two additional > distinctions to consider. > > 1) Ontologies based on OLU/Foundational ontologies shared by many > OBO Foundry reference ontologies: > I realize this can be a needlessly contentious topic, but I'm > certain it will become a significant limiting factor regarding what > one can expect to be able to compute using a given artifact. It's > not that building off BFO or DOLCE necessarily establishes either > the veracity or credibility of a given artifact, but it does > greatly effect the scope of effort required to make use of the > ontology. In other words, if a new ontology artifact is built > using BFO, other researchers will have less work to do - much much > less work to do - to put that ontology to effective use if they are > also using BFO and/or Foundry reference ontologies built upon BFO. > > 2) "Interstitial" ontologies (ontologies consisting primarily of > relations amongst foundry reference ontology continuants & occurents): > It's somewhat of nebulous category, possibly so much so that it > couldn't really be established in a useful way. > > Having said that, its still worth considering how to treat as > distinct those cross-domain artifacts describing inter-relatedness > amongst foundry reference ontology continuants & occurents > characterizing observations upon whose interpretation essentially > 100% of biologists would be in agreement. These are the extended > artifacts likely to be of the most broad utility, and I don't think > I'd strictly describe them as "application" ontologies. > "relational" ontologies is a pretty lousy name, too, given how > loaded the term "relational" is and the fact that ALL ontologies > have relations of some sort - e.g., subsumption, existential links > to instances, etc.. Perhaps a common term used by biologists - > "interstitial" - would be more fitting. Sure - "interstitial" > ontologies has a nice ring to it. According to Merriam-Webster, > "interstitial" means: > "relating to or situated in the interstices", where "interstices" > means: > "1 a : a space that intervenes between things; especially : one > between closely spaced things <e.g., interstices of a wall> b : a > gap or break in something generally continuous<e.g., the > interstices of society>" > > For instance, > mitochondrion located_in eukaryotic_cell > hepatocyte part_of liver > neurotransmitter located_in neurotransmitter_vesicle > alpha_type2_adrenergic_receptor has_ligand noradrenaline > > These are exactly the sort of assertion the GO folks are building > out now, just as Suzi described. Between the 3 GO branches, ChEBI, > CL, and CARO, you should be able to cover all of these assertions > above. The only exception might be neurotransmitter which needs to > include both small molecule, classical bioamine transmitters such > as noradrenaline, 5-HT, etc., as well as the broad range of peptide > transmitters. Many of these may be in ChEBI (I haven't looked), > but as the products of post-translational processing from an > expressed gene, all the required peptide transmitters may not be > fully covered in ChEBI, yet. > > These are a simplification, of course - e.g., they don't include > the necessary existential qualifiers and such - but I think you get > the idea. In each statement, at least as applied to the "normal" - > i.e., non-pathalogical - state, few biologists I know - or know of > - would disagree. > > There will be some more subtle assertions, such comprehensively > representing all known locations of "chromaffin" cells, where > establishing the assertion is a widely accepted interpretation of > the known observations may be less straight-forward. Having said > that, even for the "chromaffin cell" example, if you limit yourself > to stating was is known without constraining the assertion to being > a "descriptive" as opposed to a "defining" assertion about > "chromaffin cell" location, there too you would probably be able to > reach essentially 100% agreement amongst the domain experts. > > As I see it, making this distinction falls in line with the idea of > identifying the "gold standard" assertions which all biologists > would want to be able to drawn on in systems engineered to use the > OBO Foundry (or foundry-related) ontologies. > > There are two significant sticking points: > 1) what will be the authority that establishes the veracity of > these "gold standard" assertions describing this consensus, core > biological reality (a topic not unrelated to the current BioTop > effort, though, as the examples make clear, I'm not think primarily > of foundational assertions); > 2) what can one do when an artifact contains a mix "gold standard" > assertions and other that cannot be considered to be accepted fact > about biological reality. > > Neither of these is insurmountable. They just need to be addressed > in some way. And the usual for consensus on a gold standard is a lot of discussion by respected authorities in that branch of biology. Like writing up a review paper, or a text book. This is much the same. > > I do believe filling out some of these details on the OBO Foundry > page (s) (and in the manuscript) would greatly enhance the utility > of - and hasten the adoption of - the OBO Foundry as a critical > bioinformatic resource. As Suzi requested, I'd be glad to get some > pages started over next weekend (this week is pretty full). I > could set them up in a Google Doc all could access that would > easily translate to HTML when they are ready to publish. I think we will have the wiki ready soon. This is just for the FAXs though, and other web site related stuff. The manuscript Barry passed to me last week. How about we do this the old-fashioned way: the ball- holder metaphor. I've got the ball (manuscript) now, and will pass it on to you this weekend. ta, S > > Cheers, > Bill > > > On Apr 23, 2007, at 7:04 AM, Smith, Barry wrote: > >> Suzi wrote: >>> Barry, I had been thinking that we would/should have a category >>> within the OBO-Foundry for artifacts that are purely composed from >>> OBO-Foundry reference ontologies. For example, when/if mammalian >>> phenotype were solely comprised of terms from mammalian anatomy >>> (OBO- >>> Foundry reference) plus PATO (another OBO-Foundry reference) then it >>> would be an OBO-Foundry application ontology (bear with me, I know >>> that this is somewhat over-simplified). I think there is a >>> fundamental difference between links out to MESH, and links between >>> entities composed from OBO-Foundry ontologies alone, if only because >>> the tools we could build. If we could depend on this, then tool- >>> building would be quite wonderful. We're adding links between the 3 >>> GO ontologies (intra-GO), and between GO and CL, and between GO and >>> ChEBI. How do you see these intra-foundry links? >> >> The nice thing about a pure OBO Foundry-based artifact, is that the >> artifact itself could be subjected to the same principles as the base >> ontologies themselves. Hence I think that we should certainly be >> working towards creating such a category. >> (We sort of had it from the start, with the distinction between >> Foundry reference ontologies and Foundry application ontologies; but >> now is the time to think it through more carefully on the basis of >> real examples). >> BS >> >> >> >> --------------------------------------------------------------------- >> ---- >> 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. >> http://sourceforge.net/powerbar/db2/ >> _______________________________________________ >> Obo-discuss mailing list >> Obo...@li... >> https://lists.sourceforge.net/lists/listinfo/obo-discuss > > > > Bill Bug > Senior Research Analyst/Ontological Engineer > > Laboratory for Bioimaging & Anatomical Informatics > www.neuroterrain.org > Department of Neurobiology & Anatomy > Drexel University College of Medicine > 2900 Queen Lane > Philadelphia, PA 19129 > 215 991 8430 (ph) > 610 457 0443 (mobile) > 215 843 9367 (fax) > > > Please Note: I now have a new email - William.Bug@DrexelMed.edu > > > > > ---------------------------------------------------------------------- > --- > 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. > http://sourceforge.net/powerbar/db2/ > _______________________________________________ > Obo-discuss mailing list > Obo...@li... > https://lists.sourceforge.net/lists/listinfo/obo-discuss |