From: Alan R. <ala...@gm...> - 2008-01-26 05:25:33
|
Hi Samson, I don't think I concur on the assessments of which of the below are qualities versus dispositions. First, let me say how I determine which is which, though I welcome other's algorithms for figuring this sort of thing out. I'll start with realizable entities, of which dispositions are a subclass, as I find them easier to distinguish. A realizable entity is a dependent continuant, which means it exists only by virtue of an independent continuant that bears it. Realizable entities manifest themselves as participation of their bearers in a process, in a certain mode of participation, under appropriate circumstances. The definition of a realizable entity would explain what each of these things are. A prototypical realizable entity would be a function. We say that a water pump has a pumping function by which we that, supplied with power, in the presence of water, an active pumping process has the pump as the mover of water. To satisfy yourself that the function exists would need to observe the process over some period of time - a snapshop won't do. A quality doesn't have this same relationship to a process. That is, you can have a quality that does not require, in order for one to define it, the mentions of process. Shape would be a prototypical quality. You can take a snapshot that shows the shape, and while shape can influence how a process happens, if nothing ever happened to the bearer of the shape, we could still be convinced that it had the shape. There are certainly relationships between e.g. functions, and qualities. For example, there are a restricted set of shapes that pumps can take, and the pumping function can be explained, for the most part in terms of those qualities. However, the shape quality that explains the function is different from the function. In the light of these, let's look at the entities that you describe: On Jan 25, 2008, at 8:55 PM, Samson Tu wrote: > Let me expand on what Chris wrote about our work on developing > autism ontology, and comment on Bill's observations. > > There are three parts to our extension of birnlex ontology: > > 1. Autism-related qualities > e.g., > Savant skill > Savant negative > Savant positive I would say these are realizable entities. There is no snapshot that would let you observer that an individual has savant skill. Instead, the skill is manifested in processes where the individual demonstrates the skill. > Insistent on sameness > High insistence on sameness > Medium insistence on sameness > Low insistence on sameness Again, I would say these are realizable entities. They are realized in processes where the individual reacts in some way to a change in their environment or status. > 2. What I call "situation": the state (of one or more continuants) > at a particular time. It is often defined by a complex of > observations. > e.g., Assessment result > ADI_2003_result > ADOS1_2001_result (where ADI_2003 and ADOS1_2001 are > autism assessment instruments) > Situation satisfying DSM Part A diagnostic criteria for autism Assessments are processes, and assessment results are measurements - they yield information. From that information one may, of course, conclude that certain entities exists or that existing entities have qualities or dispositions, for instance. In OBI we have information_content_entity to capture measurements. Currently our definitions are: An information content entity is a non-realizable information entity that 'is encoded in' some digital or physical entity A non-realizable information entity is a generically dependent continuant which conveys meaning that can be communicated but whose meaning does necessarily imply that some process might follow from that meaning. The branch that these are in, DigitialEntityPlus, is still in development. Certainly it would be interesting to get your feedback on what we have so far, and suggestions for improvements and additions. > > 3. Extensions to bfo: disposition > e.g., Developmental capability > Communication capability > Ability to initiate speech > Ability to sustain speech > Social interaction > Emotional reciprocity, > ... > Autism disposition > Impairment of communication capabiltiy > Impairment of social behavior I agree that these all seem to be in the realm of realizable entities, though, as you note below, though, not seeing the definitions I'm not sure. Some certainly, though, such as emotional reciprocity. All would be most understandable with definitions that took the form "an xxxx is a realizable entity that is realized in yyyy processes where the bearer zzzzs" I hedge about dispositions (a subclass of realizable entities) only because, as Bill alludes to, I also don't have a clear way to distinguish functions from dispositions, though in language use I tend to choose one or the other in different circumstances. (more to learn :) > The assessment results are direct mappings to the NDAR codebook. > Martin O'Connor at Stanford BMIR has developed a SWRL-to-relational > database mapping so that, in effect, the SWRL rules are translated > into relational queries. > > I am not happy with our current model in the following ways. > Perhaps some of you can help me with them. > > I definitely need to add time to the assertions about individuals > having autistic qualities, because the assessments are time- > specific. I looked at the PATO pheno-syntax w/o seeing any solution. > > I searched through Birnlex but I couldn't find any thing that > allows me to model "situation" or results of assessments. I must be > missing something. Surely OBI must have something that models the > results of investigation, like the measurements and observations > that are recorded. How do you model Bill Bug's "phenotypical > observation"? Yes, as mentioned above, information_content_entities, as well as digital_entities (and physical entities) in which they can be encoded. > "Situation" plays two roles in our ontology. First, it allows me to > model the assessment results (collections of observations) and, by > placing restrictions on the assessment results, define when a human > subject has the "autism qualities" (Savant positive, for example). > We use SWRL rules to make the connection between assessments and > the "bearing" of the qualities. (if X assessment score > 5, the > subject is_a_bearer_of "savant positive" quality). The schema in obi would be something like: protocol applications (= processes by which assessments are made, in which the assessor and patient participate, which are realizations of protocols - a kind of plan) information_content_entities (the output of such protocol applications) You would write the antecedents of the rules in terms of these entities (patient, protocol, output) and have the consequent assert dispositions that the patient bears, if appropriate. I don't think you need "situation", at least not in the BFO framework, unless it turns out to be synonymous with something described above. > Second, "situation" can be intensionally defined (e.g, Situation > satisfying DSM Part A diagnostic criteria for autism) e.g., using > OWL restrictions. They can also allow me to define when a human > subject has autism, according to DSM criteria. These criteria makes > use of concepts and relationships such as "has _impairment_of some > Communication_capability". > > The Autism-related qualities are just a collection of labels. These > labels derive their semantics through the SWRL rules. I have added > a reference to the SWRL rule as an annotation to the quality (e.g., > Savant positive) . However, the SWRL rules relate these quality > labels to assessment results. Right now there is no sense that > "savant positive", "savant negative" are manifestations ( presence, > absence) of a disposition or function, and thus relating the > qualities to the developmental capabilities that characterize autism. > > > Also, where is "developmental capability" in Birlex? I don't think > it's a function, because it is "the ability to perform a function." > Tentatively, I've added it as a new subclass of obi:disposition. That makes sense. It would be bfo:disposition, btw. > There is certainly a tension between disposition and quality. For > example, if I define "savant skill" as a developmental capability > (disposition), then (presence of savant skill) can be an autism > disposition. If "presence_of" is an OWL property, I can say that > (presence_of savant skill) is equivalent to bearer of savant skill > positive. Is that the modeling direction that we want to go? I don't see the tension between quality and disposition here. The are comparable in that both can inhere in independent continuants, and in OWL this means that they the instances can be related by a property, but the meaning of the relation is quite different. After reading my descriptions of the difference between qualities and dispositions, do you still see a problem here? Hope this helps, and sounds like a fascinating project! Regards, Alan > > > Thanks. > > Samson > > Bill Bug wrote: >> >> Thanks, Chris. >> >> I agree pretty much with everything you say below. I'd especially >> recommend "promoting" PATO:disposition to a higher level in the >> PATO asserted hierarchy - one that is more commensurate with >> BFO:disposition. I suppose it would make sense for >> PATO:disposition to be a direct child subclass of BFO:disposition, >> just as PATO:quality is a direct child subclass of BFO:disposition. >> >> As you say below, there is still considerable confusion amongst >> the community of BFO users as to exactly what differentiates a >> BFO:disposition from other BFO:dependent_continuants, though this >> is primarily in providing a clear sense of how to use formal >> languages such as OWL and/or OBO to represent the distinction >> between BFO:disposition and BFO:function. I don't believe the >> same confusion exists for BFO:disposition and BFO:quality/state, >> though as you say, whether this is truly the case requires having >> clear examples. I've been asking for a broad collection of >> BFO:dependent_continuant examples in OWL for some time. The BFO >> and SNAP/SPAN-related manuscripts give a reasonably clear (in some >> cases, a formal) description of these distinct >> BFO:dependent_continuant classes. The definitions for these >> classes in the bfo.owl are not nearly as much help. >> >> Regarding your specific example - photo-sensitivity - I believe it >> most definitely is a disposition, as opposed to a state/quality. >> It's a disposition that is very relevant to a variety of >> neurological disorders such as epilepsy, stroke, hypothyroidism >> and migraine. I also believe - at least in a BFO+OBO-RO+PATO >> framework - most diseases will end up as dispositions. What is >> less clear is how classifying phenotypic observations of a given >> disease-associated symptom will link to the variety of >> dispositions that collectively indicate when a subject can be >> considered to be in a particular disease stage. As you imply, >> there is a subtle balance between what should be specified as a >> disposition (roughly 8% of diagnosed sufferers of epilepsy >> demonstrate a disposition toward being sensitive to light - which >> they don't "realize" when in a dark room), and what will be >> specified as a phenotype observation. In this particular example, >> analysis of the body of clinical knowledge would lead one to >> represent a "disposition toward photosensitivity" in a mode of >> diagnosed epileptics. Analysis of a particular subject's clinical >> data, would likely be used to diagnose that patient as being in a >> particular disease population based on phenotype assertions >> (diagnostic assays, instruments, assessments). It's not yet clear >> to me whether one of these representational approaches is >> preferred (is more useful), OR if both are required, how they can >> effectively be combined to help add a measure of automation either >> to the diagnostic process or to the process of defining nuanced >> representations of complex disease states that extend all the way >> down to the level of expressed genes and environmental factors. >> >> In the end, examples are what us BFO+OBO-RO+PATO users require >> more than anything else - examples that can empower query or >> inference based meta-analysis. >> >> Cheers, >> Bill >> >> >> On Jan 24, 2008, at 6:20 PM, Chris Mungall wrote: >> >>> >>> Just a bit of background for those not familiar with the BFO >>> upper ontology we are using to help organize ontologies. >>> >>> BFO makes a distinction between dispositions and qualities - see >>> the definitions Bill provides below. You can think of a >>> disposition as something that can be realized, whereas something >>> does or does not have a quality at any given time. A glass may >>> have a disposition to shatter (when dropped on the ground, say), >>> whereas at this particular time it may have the quality of being >>> solid, in a certain shape etc. >>> >>> Does this distinction matter? It might do, for integrating data >>> from different sources, assays etc. It's always good to be >>> absolutely clear about what a term means. A disposition to shed >>> foliage is different from not having foliage: >>> >>> PATO has a very minimal disposition branch at the moment: >>> >>> +/ PATO:0000001 quality >>> ++is_a PATO:0001241 quality of continuant >>> +++is_a PATO:0001238 relational quality of continuant >>> ++++is_a PATO:0001727 disposition >>> +++++is_a PATO:0001728 multi-cellular organismal disposition >>> ++++++is_a PATO:0001729 disposition to shed >>> +++++++is_a PATO:0001730 deciduous (generic) >>> +++++++is_a PATO:0001731 deciduous (plant) >>> +++++++is_a PATO:0001732 non-deciduous (any body part) >>> +++++++is_a PATO:0001733 evergreen (plant) >>> +++++++is_a PATO:0001734 semi-deciduous(plant) >>> >>> Now this is causing a problem for folks like Bill who are pulling >>> together ontologies from multiple sources. One ontology says >>> quality and disposition are disjoint, but PATO violates that. >>> >>> Bill's problem would be solved by shifting disposition up to be a >>> root node alongside disposition (they share the same parent in >>> the bridging ontology; we could add that to PATO to give a shared >>> root if we wanted, it's not so important). Note that we would >>> want to keep this in PATO as to move it out would be too >>> confusing. For most use cases, quality terms and disposition >>> terms would be used in the same contexts. >>> >>> Now this does create a potential confusion for users - wherever >>> we have previously been talking of qualities we are not talking >>> of qualities-or-dispositions (or dependents). >>> >>> This isn't coming out of nowhere: it's been languishing in the >>> tracker the last 9 months: >>> http://sourceforge.net/tracker/index.php? >>> func=detail&aid=1675411&group_id=76834&atid=595654 >>> >>> Thanks for bringing it up Bill. >>> >>> My feeling is that we need a stronger sense of what the >>> requirements are for dispositions from users. I know a lot of >>> people are currently interested in behaviour phenotypes, and many >>> of these would seem intimately connected to dispositions. We also >>> want to avoid creating new sources of confusion (example: is a >>> disposition to blink in bright light the same as sensitivity to >>> light?). It may be the case that users would prefer to see >>> dispositions in a new ontology, distinct from PATO. >>> >>> For now I suggest Bill carries on treating the root node of PATO >>> as a BFO:quality, as the BFO bridge states. We can have a brief >>> moratorium on placing new child terms of PATO:0001727 until we >>> have collected a mass of terms, and then decide what to do (which >>> may involve obsoleting PATO:0001727 and creating a repaced_by >>> node, in a new ontology or new root). >>> >>> On Jan 21, 2008, at 1:48 AM, Bill Bug wrote: >>> >>>> Dear PATO folk, >>>> >>>> I just noticed PATO includes a class "pato:disposition" as a sub- >>>> type of "pato:relational quality of continuant". >>>> >>>> Can someone provide info on how this relates to >>>> "bfo:disposition" in BFO, which there is a sub-type of >>>> "bfo:realizable_entity" and thus a "nephew (or is it niece)" >>>> class to "bfo:quality". I realize the debate amongst BFO users >>>> & developers as how to interpret the distinctions between >>>> realizable entities - specifically "bfo:function" and >>>> "bfo:disposition" is still somewhat up in the air, but I hadn't >>>> thought this included clouding the distinction between >>>> "bfo:quality" and "bfo:realizable entity". >>>> >>>> Even if this distinction between "bfo:realizable entity" and >>>> "bfo:quality" is somewhat in doubt, given the bfo-to-pato bridge >>>> defines "bfo:quality" the parent of "pato:quality", it's >>>> difficult to understand how to relate the potential sibling (or >>>> equivalence) relation between "bfo:quality" and >>>> "bfo:disposition" to the position of "pato:disposition" as being >>>> classes deeper in the hierarchy from "bfo:quality" - i.e.: >>>> "pato:disposition" -> "pato:relational quality of continuant" -- >>>> > "pato:quality of continuant" --> "pato:quality" --> "bfo:quality" >>>> >>>> I'd like to think I'm just being confused by a pair of disparate >>>> classes that happen to have the same name, but I don't think >>>> that is the case, as the PATO definition of "disposition" - >>>> despite the enormous difficulty in defining a useful definition >>>> for bfo:disposition (i.e., one that is consistent, cogent, and >>>> complete) - appears to be trying to specify the same sort of >>>> entities as bfo:disposition is trying to representation - i.e.: >>>> >>>> pato:disposition: >>>> A prevailing tendency, mood, or inclination. (from Merriam- >>>> Webster's dictionary apparently) >>>> >>>> bfo:disposition: >>>> Definition: A realizable entity that essentially causes a >>>> specific process or transformation in the object in which it >>>> inheres, under specific circumstances and in conjunction with >>>> the laws of nature. A general formula for dispositions is: X >>>> (object) has the disposition D to (transform, initiate a >>>> process) R under conditions C. >>>> >>>> I would also note that "pato:disposition" inherits useful links >>>> to PATO relations - i.e.: >>>> inheres_in: >>>> domain: "pato:quality of continuant" >>>> range: "snap:continuant" >>>> towards: >>>> domain: "pato:relational quality of continuant" >>>> range: "bfo:entity" >>>> >>>> These properties are in fact required to make effective use of >>>> "bfo:disposition" as well (but are not currently being asserted >>>> for "bfo:dispisition" - nor can they be, since bfo.owl doesn't >>>> make reference to OBO-RO or PATO), so it would be useful to >>>> somehow reconcile what appears right now to be a pair of >>>> parallel classes that ultimately seek to provide the same >>>> representational capability. >>>> >>>> Many thanks for any specifics you can offer regarding this >>>> relation between "pato:disposition" and "bfo:disposition". >>>> >>>> Cheers, >>>> Bill >>>> >>>> P.S.: I'm including Holger Stenzhorn (curator of the BFO OWL >>>> file) on this post to make sure he is in the loop on this >>>> discussion. I thought that might be more expeditious to actually >>>> resolving the issue than to simply double post this to both the >>>> PATO and BFO lists. >>>> >>>> >>>> William Bug, M.S., >>>> M.Phil. email: >>>> wb...@nc... >>>> Ontological Engineer (Programmer Analyst III) work: (610) 457-0443 >>>> Biomedical Informatics Research Network (BIRN) >>>> and >>>> National Center for Microscopy & Imaging Research (NCMIR) >>>> Dept. of Neuroscience, School of Medicine >>>> University of California, San Diego >>>> 9500 Gilman Drive >>>> La Jolla, CA 92093 >>>> >>>> Please note my email has recently changed >>>> >>>> >>>> ------------------------------------------------------------------- >>>> ------ >>>> This SF.net email is sponsored by: Microsoft >>>> Defy all challenges. Microsoft(R) Visual Studio 2008. >>>> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ >>>> _______________________________________________ >>>> Obo-phenotype mailing list >>>> Obo...@li... >>>> https://lists.sourceforge.net/lists/listinfo/obo-phenotype >>> >> >> William Bug, M.S., >> M.Phil. email: >> wb...@nc... >> Ontological Engineer (Programmer Analyst III) work: (610) 457-0443 >> Biomedical Informatics Research Network (BIRN) >> and >> National Center for Microscopy & Imaging Research (NCMIR) >> Dept. of Neuroscience, School of Medicine >> University of California, San Diego >> 9500 Gilman Drive >> La Jolla, CA 92093 >> >> Please note my email has recently changed >> >> >> --------------------------------------------------------------------- >> ---- >> This SF.net email is sponsored by: Microsoft >> Defy all challenges. Microsoft(R) Visual Studio 2008. >> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ >> _______________________________________________ >> Obo-phenotype mailing list >> Obo...@li... >> https://lists.sourceforge.net/lists/listinfo/obo-phenotype >> > > -- > Samson Tu email: sw...@st... > Senior Research Scientist web: www.stanford.edu/ > ~swt/ > Center for Biomedical Informatics Research phone: 1-650-725-3391 > Stanford University fax: 1-650-725-7944 > ---------------------------------------------------------------------- > --- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Obo-phenotype mailing list > Obo...@li... > https://lists.sourceforge.net/lists/listinfo/obo-phenotype |