On 08/25/2010 06:20 PM, Richard H. Scheuermann, Ph.D. wrote:
> I think you are on the right track here. I do think that we need both the dispositions (disease) as well as the occurrents (pathological processes). In theory, we might also want to describe the abnormal independent continuant that bears the disposition (the abnormal brain).
> Your questions raise and imply a more general question, should the Human Phenotype Ontology be restricted to phenotypic qualities as implied by its name, or should it abandon this notion (and perhaps this name) and define itself as an "application" ontology that can select terms from different BFO branches and reference ontologies as needed for your applications. I would support the latter.
> In terms of the inflation of parallel terms in the independent continuant, dependent continuant and occurrent branches, unfortunately I don't think there is any way to avoid that. In fact, for many applications we do want to make very clear distinction between them, i.e. in medical records and clinical research databases. But I would also argue that we should not automatically add terms unless we find that we definitely need them. For example, I know that we would want to talk about the epilepsy disposition and the seizure occurrent, but I am less convinced that I need to describe the abnormal brain that bears the epilepsy disposition for my applications (although I do want to talk about the patient with the disease epilepsy).
> I will leave others to comment on the need for new relations since I am not active in the RO discussions.
> I would be fine with starting with a more generic epilepsy term for the disposition and add more specific children terms if needed at some future time. And I agree that we will need to specialized occurrent terms sooner rather than later.
I am thinking that it might be interesting to work out design patterns
to describe abnormal processes such as "seizure". I think it would be
difficult to do this using a subsumption hierarchy such as we currently
have in the HPO. I like the idea of using a number of other OBO
ontologies for the definitions (I would mention that we are already
using a logical definition approach based upon cross product definitions
in an effort that was proposed and developed mainly by Michael
Ashburner, Suzi Lewis, Paul Schofield, Chris Mungall, and George
Gkoutos, in no particular order, and that we are refering to FMA, GO,
Cell O, CHEBI, PRO).
However, I think that defining processes such as seizure in a useful way
will require a more complex structure. I am thinking that some of the
design patterns typically seen in OWL ontologies might be required to do
this well, but I am aware of the difficulties this would entail.
Nonetheless, I would like to try and make a small prototype ontology for
an area such as seizure to explore some of the possibilities. If this is
of interest to you or any others on the list, we should continue
conversations about it.
best wishes Peter
> On Jul 19, 2010, at 12:04 PM, Peter Robinson wrote:
>> Hi Phenotypers& Barry and Richard,
>> I have been thinking about where HPO terms fit into BFO (<gasp>), and I wonder if it may be wise to think about revising some of the structure of the HPO in the medium-term.
>> HPO terms can be imagined to fall into a number of categories. At the moment, the HPO is conceived as an ontology of abnormalities rather than an ontology of abnormal entities. But if we do examine the abnormal entities, we have entities such as say microphthalmia (small eye) that are continuants, and other entities such as tachycardia (heart beating too fast) that are occurents.
>> If this was all, then we would have no big problem conceiving of the abnormalities as (more or less abstract) independent continuants.
>> However, there are a number of abnormal entities such as epilepsy, which are dispositions (dependent continuants) to have seizures (each of which is an occurent).
>> At the moment, the distinction between the disposition (epilepsy/seizures) and the occurents (a particular seizure) is rather vague in the HPO.
>> The MD concludes that a patient has a disposition (epilepsy) based on the finding that a patient has already had at least several seizures. If we confine the HPO term to the disposition, then we have no way of saying that some patient has had one single seizure.
>> Therefore, for terms like this, we need to have one term that refers to the disposition, and another term that refers to the occurrent. We need a relationship such as
>> epilepsy ---(is predisposed to suffer episodes of)-->seizure
>> One danger of this approach is that there can be an inflation of terms. Do we need a different term for each kind of disposition? (Remembering that there are epileptics who can have different mixes of seizure types also).
>> Another point is that I think the only way of making a clean distinction between the dependent continuants and the occurents would be to make the HPO an ontology of abnormal entities rather than of abnormalities. This would have its own disadvantages.
>> Any comments? I would not plan on making any hurried changes at the this point, but would like to make an action plan this Fall...My questions:
>> 1) Does this sound like a reasonable idea?
>> 2) Do we need a new relation?
>> 3) Is there a good way of having one general term such as epilepsy for the disposition, while retaining the numerous more specialized terms for the occurrents?
>> thanks Peter
>> PD Dr. med. Peter N. Robinson, MSc.
>> Institut für Medizinische Genetik
>> Charité - Universitätsmedizin Berlin
>> Augustenburger Platz 1
>> 13353 Berlin
>> voice: 49-30-450566042
>> fax: 49-30-450569915
>> email: peter.robinson@...
> Richard H. Scheuermann, Ph.D.
> Chief, Division of Biomedical Informatics
> John H. Childers Professorship in Pathology
> Department of Pathology
> U.T. Southwestern Medical Center
> 5323 Harry Hines Blvd.
> Dallas, TX 75390-9072
> phone: 214-648-4115
> FAX: 214-648-4070
> email: richard.scheuermann@...
PD Dr. med. Peter N. Robinson, MSc.
Institut für Medizinische Genetik
Charité - Universitätsmedizin Berlin
Augustenburger Platz 1