Re: [Fuge-devel] FuGE v1 - Issue summary

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 454-5900

Hi Junmin,

Yes, agreed it could be a protocol, but for a gene expression
investigation this information would be included in, the details of the
protocol are irrelevant, these rats post-sacrifice are the initial input
material so using ontology annotations is the best way to record it.

cheers,
Michael

> -----Original Message-----
> From: fug...@li...=20
> [mailto:fug...@li...] On Behalf=20
> Of Junmin Liu
> Sent: Friday, October 13, 2006 8:39 AM
> To: fug...@li...
> Cc: MGED Ontology Working Group
> Subject: Re: [Fuge-devel] FuGE v1 - Issue summary
>=20
>=20
> Hi, Michael,
>=20
> Thank for providing use cases.
>=20
> > But imagine I have ten rats that are sacrificed on=20
> successive days and
> > this is recorded as a class Sacrificed_date with a property=20
> has_day with
> > a class Day.  Obviously I must have ten individual sacrifice dates
> > distinct from each other or I will not be able to=20
> distinguish which rat
> > was sacrificed on which day.  They are distinct from each=20
> other by the
> > 10 object slots related to the ten distinct day individuals.
> >
>=20
> I think in this use case, most likely you should end up annotating it=20
> using protocol param.
>=20
> biosouce: rat 1 --> protocol app 1 (using step 1. sacrifice=20
> protocol, step=20
> 2, extract sample protocol) --> biosample 1
>=20
> The protocolApp for protocol step 1 should have parameter value --=20
> execution date, which is just regular MO date which will have=20
> measument.
>=20
> I will try to give feedback on the other cases later.
> ---junmin
>=20
>=20
>=20
>=20
>=20
>=20
>=20
> > Note that these individual sacrificed_date are the rat's
> > sacrificed_date, so really they shouldn't be part of the ontology,
> > simply individuals of the class Sacrificed_date that=20
> relates these rats
> > to that class.  But they can be considered synonymous with=20
> the class.
> >
> > Now one could optimize this so that if two rats were=20
> sacrificed on the
> > same date that they shared the same individual sacrificed_date.  Or
> > there might be an ontology somewhere that enumerates every single
> > possible day as individuals but I doubt that.
> >
> > Now imagine I want to capture clinical data about my rats.  From my
> > ontology I have class Obseveration_set with a property
> > has_observation_day to class Observation_day which has properties
> > has_observation to class Observation and has_date to Date. =20
> Further the
> > classes Weight and Feed have an is_a relationship to Observation.
> >
> > So for a particular Rat I might have an observation_set=20
> individual which
> > has seven slots to observation_day individuals which are=20
> distinguished
> > by having a different has_date slot to seven successive=20
> days.  They are
> > further distinguished by having for each date two=20
> observation slots, one
> > to a Feed individual that takes on the individual wheat,=20
> corn or rice
> > and one weight individual for the weight of the rat on a particular
> > date.
> >
> > Again I may or may not have an ontology that defines the individuals
> > above, but hopefully I have an ontology for wheat, corn and=20
> rice which
> > may or may not be distinct (in a similar way a unit=20
> ontology is stand
> > alone) from the ontology that Feed comes from so that, as=20
> you say Andy,
> > we can do comparisons.
> >
> > So each rat has a distinct tree of individuals and slots=20
> that represent
> > its individual observation_set.  Again, if by chance two=20
> rats shared the
> > same individual tree, they could share that individual.
> >
> > As to the usefulness of Ontologies in this, the values of value
> > properties can be compared, when the individuals are named=20
> the same as
> > their class, say from FuGE (the only thing one can do),=20
> then comparisons
> > are probably safe, when the individuals come from a=20
> different ontology
> > but it can be known which ontology it came from, then the=20
> comparisons
> > can also take place.
> >
> >> If this is the case, then the ontology is incomplete or
> >> wrongly designed. If
> >> an arbitrary user-entered value is allowed in the ontology,
> >> then fine, this
> >> is a DataProperty. If only valid individuals are allowed,=20
> then 'death'
> >> should be added as an individual... If the data model says it
> >> is okay to put
> >> whatever you like in here the ontology never gets improved.
> >
> > It is hubris to think that any human designed ontology can=20
> be complete.
> > 'death' was only one that came to mind, as new terms are=20
> NEEDED because
> > a new use case has been discovered, they need to be recorded for the
> > investigation independent of being incorporated into an=20
> ontology.  Think
> > about the complete set of characteristics for Material--it=20
> is logically
> > complete only if it includes everything there is that can=20
> be known about
> > the universe.  So ontologies have to limit what they try to=20
> capture to
> > the part of the universe which is the domain of the=20
> ontology.  Even then
> > it is hard to envision all that is needed.
> >
> > cheers,
> > Michael
> >
> >> -----Original Message-----
> >> From: fug...@li...
> >> [mailto:fug...@li...] On Behalf
> >> Of Andy Jones
> >> Sent: Thursday, October 12, 2006 9:35 AM
> >> To: fug...@li...
> >> Subject: Re: [Fuge-devel] FuGE v1 - Issue summary
> >>
> >>
> >> Hi Michael,
> >>
> >>> I'm not sure when we went from having the ontology as an
> >> attribute of
> >>> OntologyTerm to a URI association, (good idea by the by),
> >> but then it
> >>> brings up an ambiguity of whether the OntologySource is for
> >> the term,
> >>> that represents an ontology individual, or the termAccession, that
> >>> represents a ontology class.
> >>
> >> OntologySource is meant to be defined only once for each
> >> ontology in the
> >> entire document i.e. once for OBI, once for NCBI Taxonomy etc. The
> >> termAccession is the one unique identifier for each term.
> >> There is no longer
> >> a URI at the term level.
> >>
> >>> For instance, my favorite example is the MO class
> >> InitialTimePoint that
> >>> has a set of Individuals associated with it.  But 'death',
> >> which seems
> >>> like a perfectly valid individual for that class, as would
> >> many others,
> >>> is not in that list.
> >>
> >> If this is the case, then the ontology is incomplete or
> >> wrongly designed. If
> >> an arbitrary user-entered value is allowed in the ontology,
> >> then fine, this
> >> is a DataProperty. If only valid individuals are allowed,=20
> then 'death'
> >> should be added as an individual... If the data model says it
> >> is okay to put
> >> whatever you like in here the ontology never gets improved.
> >>
> >>> If OntologyTerm has lost this distinction between
> >> individual and class,
> >>> object property and object slot, and value property and
> >> value slot, we
> >>> need to seriously revisit to put that back in.  Yes, they
> >> often are the
> >>> same but even for the 'Age' class and the set of 'age'
> >> individuals, the
> >>> distinction is that each 'age' will be associated via slots
> >> to distinct
> >>> different age values and units.
> >>
> >> I'm lost on this point. Ontology individuals are only classes
> >> that can't
> >> have child nodes, hence in the most ontologies now, including OBI,
> >> individuals are not used, everything is a class or a
> >> property. In the model,
> >> classes and individuals are both represented by OntologyIndividual.
> >>
> >> We still have a DataProperty (e.g term =3D "has_value" value =3D
> >> "12); value
> >> property and value slot.
> >>
> >> What is an object property and object slot? Is this not just
> >> a property
> >> (hasIntialTimePoint) and an associated class/individual
> >> (InitialTimePoint).
> >>
> >> I think the point you're getting at is that, classes must=20
> come from an
> >> ontology but they can be completed with any individuals the
> >> user wants? If
> >> so, I disagree with this distinction, this removes the whole
> >> point of using
> >> an ontology or CV.
> >>
> >> I agree with Junmin's point, terms in an ontology are context
> >> dependent. If
> >> you remove them from their context they are no longer valid.
> >> A person might
> >> be able to guess what the data producer intended but a
> >> machine could not do
> >> anything with it.
> >>
> >> Cheers,
> >> Andy
> >>
> >>
> >>
> >>
> >>
> >>> -----Original Message-----
> >>> From: fug...@li... [mailto:fuge-devel-
> >>> bo...@li...] On Behalf Of Miller, Michael
> >> D (Rosetta)
> >>> Sent: 12 October 2006 16:45
> >>> To: fug...@li...
> >>> Subject: Re: [Fuge-devel] FuGE v1 - Issue summary
> >>>
> >>> Hi Junmin,
> >>>
> >>>> But besides that, we should
> >>>> discourage people to use different ontology sources
> >>>
> >>> Every OntologyTerm has an association to ontologyURI that
> >> must be used
> >>> in this case to clearly identify where classes and
> >> properties each are
> >>> being drawn from.
> >>>
> >>> I'm not sure when we went from having the ontology as an
> >> attribute of
> >>> OntologyTerm to a URI association, (good idea by the by),
> >> but then it
> >>> brings up an ambiguity of whether the OntologySource is for
> >> the term,
> >>> that represents an ontology individual, or the termAccession, that
> >>> represents a ontology class.
> >>>
> >>> For instance, my favorite example is the MO class
> >> InitialTimePoint that
> >>> has a set of Individuals associated with it.  But 'death',
> >> which seems
> >>> like a perfectly valid individual for that class, as would
> >> many others,
> >>> is not in that list.
> >>>
> >>> I suppose it would be an okay rule that the OntologySource
> >> is always for
> >>> the termAccession, which is the class in the ontology.
> >>>
> >>> If OntologyTerm has lost this distinction between
> >> individual and class,
> >>> object property and object slot, and value property and
> >> value slot, we
> >>> need to seriously revisit to put that back in.  Yes, they
> >> often are the
> >>> same but even for the 'Age' class and the set of 'age'
> >> individuals, the
> >>> distinction is that each 'age' will be associated via slots
> >> to distinct
> >>> different age values and units.
> >>>
> >>> cheers,
> >>> Michael
> >>>
> >>>> -----Original Message-----
> >>>> From: fug...@li...
> >>>> [mailto:fug...@li...] On Behalf
> >>>> Of Junmin Liu
> >>>> Sent: Thursday, October 12, 2006 8:26 AM
> >>>> To: fug...@li...
> >>>> Subject: Re: [Fuge-devel] FuGE v1 - Issue summary
> >>>>
> >>>>
> >>>> Hi, Michael,
> >>>>
> >>>> my 2c.
> >>>>
> >>>> Unit is special, maybe only the special case, because
> >> they are more
> >>>> internationalized and generalized, every biologists around world
> >>>> understand what "minute, hour" means. But besides that, we should
> >>>> discourage people to use different ontology sources, for
> >>>> "splitting" in
> >>>> MO may not have the same meaning as in some meteorology ontology.
> >>>>
> >>>> ---junmin
> >>>>
> >>>>
> >>>> On Wed, 11 Oct 2006, Miller, Michael D (Rosetta) wrote:
> >>>>
> >>>>> Hi All,
> >>>>>
> >>>>>> I am not quite sure about this: allow ontologyTerm tree
> >>>>>> having terms from different source, maybe ONLY IF those
> >>>>>> different ontology
> >>>>>> sources are interoperable.
> >>>>>
> >>>>> We will never be able to keep people from doing bad things
> >>>> when it comes
> >>>>> to how they assign values to attributes.  Anything between "" is
> >>>>> invisible to the model.
> >>>>>
> >>>>> I do agree we should guide people to the correct behavior
> >>>> but on this
> >>>>> point, since there are good use cases where one might use
> >>>> classes and
> >>>>> properties from different ontologies, it is hard to
> >> offer a blanket
> >>>>> decree.
> >>>>>
> >>>>> Being able to define one's own kinds of measurements
> >> but then use a
> >>>>> standard measurement ontology for its units seems
> >> highly desirable,
> >>>>> also, one part of the semantic web is the creation of bridging
> >>>>> ontologies whose sole purpose is to connect terms from different
> >>>>> ontologies.
> >>>>>
> >>>>> cheers,
> >>>>> Michael
> >>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: fug...@li...
> >>>>>> [mailto:fug...@li...] On Behalf
> >>>>>> Of Junmin Liu
> >>>>>> Sent: Wednesday, October 11, 2006 2:45 PM
> >>>>>> To: fug...@li...
> >>>>>> Subject: Re: [Fuge-devel] FuGE v1 - Issue summary
> >>>>>>
> >>>>>>
> >>>>>> Hi,
> >>>>>>
> >>>>>>> "user should not develop pseudo ontologies involving
> >>>>>> ObjectProperties"
> >>>>>>>
> >>>>>>> I don't see the problem, I think having an
> >> OntologyTerm tree that
> >>>>>>> incorporates terms from a variety of ontologies is a
> >>>> perfectly sane
> >>>>>>> scenario.
> >>>>>>
> >>>>>> I am not quite sure about this: allow ontologyTerm tree
> >>>>>> having terms from different source, maybe ONLY IF those
> >>>>>> different ontology
> >>>>>> sources are interoperable.
> >>>>>>
> >>>>>> This is approach we should be cautious about.
> >>>>>> ---junmin
> >>>>>>
> >>>>>>> 	-----Original Message-----
> >>>>>>> 	From: fug...@li...
> >>>>>>> [mailto:fug...@li...] On
> >>>> Behalf Of Andy
> >>>>>>> Jones
> >>>>>>> 	Sent: Tuesday, October 10, 2006 3:25 AM
> >>>>>>> 	To: fug...@li...
> >>>>>>> 	Subject: [Fuge-devel] FuGE v1 - Issue summary
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	Hi all,
> >>>>>>>
> >>>>>>> 	A few issues have been floating around that I'd
> >> like to tie
> >>> up.
> >>>>>>> If anyone has a major problem with what is proposed below,
> >>>>>> then we'll
> >>>>>>> re-open the discussion otherwise I'll assume these are
> >>>>>> agreed ;-) I want
> >>>>>>> to get the specs into the document process shortly, during
> >>>>>> which we hope
> >>>>>>> everyone will work through the specs really carefully to
> >>>>>> make sure we
> >>>>>>> haven't missed anything vital!
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	Describable - No change except to
> >> documentation: "The URI,
> >>>>>>> Annotations, PropertySets and Descriptions associations
> >>>>>> inherited from
> >>>>>>> Describable should only be used to capture additional
> >>>>>> information where
> >>>>>>> the model contains no other structures that could used to
> >>>>>> capture the
> >>>>>>> information".
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	Material - I would strongly argue that it is
> >> bad practice to
> >>>>>>> allow users to make arbitrary distinctions between conditions,
> >>>>>>> characteristics, QC etc., these decisions are better taken
> >>>>>> by ontology
> >>>>>>> developers. I vote we remove QualityControlStatistics and
> >>>>>> stick to two
> >>>>>>> ontology associations, MaterialType and either
> >> Characteristics or
> >>>>>>> MaterialDescriptions? My vote is for the latter as it is
> >>>>>> clear that this
> >>>>>>> should be used for all things someone wants to say about
> >>>>>> the material.
> >>>>>>> We would add to the documentation that for Material the
> >>>> annotations
> >>>>>>> association should not be used.
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	OntologyTerm - We add to the documentation that
> >> user defined
> >>>>>>> terms are allowable in OntologyTerm if the required term
> >>>>>> does not exist
> >>>>>>> in an ontology and for practical reasons could not be
> >>>>>> added. For these
> >>>>>>> cases the termAccession should be completed with "CUSTOM".
> >>>>>> If it is a
> >>>>>>> term with a value, it is allowable to have a DataProperty:
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	    <OntologyIndividual term=3D"gene_knockout"
> >>>>>>> termAccession=3D"CUSTOM" identifier=3D"ex01:OntInd17">
> >>>>>>>
> >>>>>>> 	        <DataProperty term=3D"has_value"
> >> termAccession=3D"CUSTOM"
> >>>>>>> identifier=3D"ex01:DataProperty" value=3D"p53"></DataProperty>
> >>>>>>>
> >>>>>>> 	    </OntologyIndividual>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	I think the best we can do is put in a
> >> recommendation that
> >>> user
> >>>>>>> should not develop pseudo ontologies involving
> >>>>>> ObjectProperties i.e. we
> >>>>>>> want to avoid the following (I think):
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	    <OntologyIndividual term=3D"rotofor_temperature"
> >>>>>>> termAccession=3D"CUSTOM" identifier=3D"ex01:OntInd17">
> >>>>>>>
> >>>>>>> 	        <DataProperty term=3D"has_value"
> >> termAccession=3D"CUSTOM"
> >>>>>>> identifier=3D"ex01:DP4" value=3D"p53"/>
> >>>>>>>
> >>>>>>> 	        <ObjectProperty term=3D"has_unit"
> >> termAccession=3D"CUSTOM"
> >>>>>>> identifier=3D"ex01:OP4">
> >>>>>>>
> >>>>>>> 	            <OntologyIndividual term=3D"celcius"
> >>>>>>> termAccession=3D"OBI:9999" identifier=3D"ex01:OntInd18"
> >>>>>>> OntologySource_ref=3D"ex01:OS_:OBI"/>
> >>>>>>>
> >>>>>>> 	        </ObjectProperty>
> >>>>>>>
> >>>>>>> 	    </OntologyIndividual>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	Factor and FactorValue - we received a couple
> >> of comments
> >>> about
> >>>>>>> this in the NBT paper but my take is that is an issue
> >> of lack of
> >>>>>>> understanding rather than problems with model. I think we
> >>>>>> can sort this
> >>>>>>> by clarifying the paper.
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>> 	If anyone has any other model issues, raise them now
> >> otherwise
> >>>>>>> I'll go ahead with updating the specs and getting them into
> >>>>>> the document
> >>>>>>> process,
> >>>>>>>
> >>>>>>> 	Cheers
> >>>>>>>
> >>>>>>> 	Andy
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>> --------------------------------------------------------------
> >>>>>> -----------
> >>>>>> Using Tomcat but need to do more? Need to support web
> >>>>>> services, security?
> >>>>>> Get stuff done quickly with pre-integrated technology to make
> >>>>>> your job easier
> >>>>>> Download IBM WebSphere Application Server v.1.0.1 based on
> >>>>>> Apache Geronimo
> >>>>>> =
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&
> >>>>> dat=3D121642
> >>>>> _______________________________________________
> >>>>> Fuge-devel mailing list
> >>>>> Fug...@li...
> >>>>> https://lists.sourceforge.net/lists/listinfo/fuge-devel
> >>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>> --------------------------------------------------------------
> >>>> -----------
> >>>>> Using Tomcat but need to do more? Need to support web
> >>>> services, security?
> >>>>> Get stuff done quickly with pre-integrated technology to
> >>>> make your job easier
> >>>>> Download IBM WebSphere Application Server v.1.0.1 based on
> >>>> Apache Geronimo
> >>>>>
> >>>> =
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&
> >>> dat=3D121642
> >>>> _______________________________________________
> >>>> Fuge-devel mailing list
> >>>> Fug...@li...
> >>>> https://lists.sourceforge.net/lists/listinfo/fuge-devel
> >>>>
> >>>
> >>>
> >> --------------------------------------------------------------
> >> ----------
> >>> -
> >>> Using Tomcat but need to do more? Need to support web services,
> >>> security?
> >>> Get stuff done quickly with pre-integrated technology to
> >> make your job
> >>> easier
> >>> Download IBM WebSphere Application Server v.1.0.1 based on Apache
> >>> Geronimo
> >>>
> >> =
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&
> > dat=3D121642
> >> _______________________________________________
> >> Fuge-devel mailing list
> >> Fug...@li...
> >> https://lists.sourceforge.net/lists/listinfo/fuge-devel
> >>
> >>
> >>
> >>
> >=20
> --------------------------------------------------------------
> ----------
> > -
> >> Using Tomcat but need to do more? Need to support web services,
> > security?
> >> Get stuff done quickly with pre-integrated technology to=20
> make your job
> >> easier
> >> Download IBM WebSphere Application Server v.1.0.1 based on Apache
> > Geronimo
> >>
> >=20
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&
> dat=3D121642
> >> _______________________________________________
> >> Fuge-devel mailing list
> >> Fug...@li...
> >> https://lists.sourceforge.net/lists/listinfo/fuge-devel
> >
> >
> >=20
> --------------------------------------------------------------
> ----------
> > -
> > Using Tomcat but need to do more? Need to support web services,
> > security?
> > Get stuff done quickly with pre-integrated technology to=20
> make your job
> > easier
> > Download IBM WebSphere Application Server v.1.0.1 based on Apache
> > Geronimo
> >=20
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&
dat=3D121642
> _______________________________________________
> Fuge-devel mailing list
> Fug...@li...
> https://lists.sourceforge.net/lists/listinfo/fuge-devel
>
>
>
>
------------------------------------------------------------------------
-
> Using Tomcat but need to do more? Need to support web services,
security?
> Get stuff done quickly with pre-integrated technology to make your job
easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache
Geronimo
>
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D=
121642
> _______________________________________________
> Fuge-devel mailing list
> Fug...@li...
> https://lists.sourceforge.net/lists/listinfo/fuge-devel
>

------------------------------------------------------------------------
-
Using Tomcat but need to do more? Need to support web services,
security?
Get stuff done quickly with pre-integrated technology to make your job
easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache
Geronimo
http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D=
121642
_______________________________________________
Fuge-devel mailing list
Fug...@li...
https://lists.sourceforge.net/lists/listinfo/fuge-devel