Anomalous anatomical set members

[Alan Ruttenberg]

In reviewing, and then attempting to debug, the representation of teeth, I came across a number of cases that seem to be in error.

The definition of Anatomical set is:

material anatomical entity which has as its members the maximal number of anatomical structures of the same type.
While some of anatomical sets could be regarded as an organ system subvision, e.g set of lumbar nerves, others do not satisfy one or two defining attributes of organ system subdivision. E.g. thoracic viscera consists of organs but not belonging to one organ system, dental arcade consists of similar organs but they are not connected.

I ran a function to find cases (~350) where anatomical set members are not anatomical structures. The cases are listed at http://svn.neurocommons.org/svn/trunk/convert/fma/reports/anatomical-sets-with-not-structure-members.txt

I'd also like to test my understanding of the member versus part relationships.

member_of => part_of

part_of: transitive

member_of: not transitive

member_of o part_of => member_of

The last one is the one to check - it says that if a set has a part, and the part has members, then the set has those members. 

If that is accepted, then some of the above cases might be recast using part_of. However I suspect that they should all be checked, given the situation described below.

In the specific case of dentition, the issue is that an uncareful interpretation of the current fma is that the primary dentition include secondary teeth.

That arises because 

'Dentition'(FMA_75150) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dentition'(FMA_75150) has member 'Maxillary dentition'(FMA_269582) which is not anatomical structure

 

and 'Primary dentition' is_a 'Dentition'

and the members of 'Maxillary dentition' are all secondary teeth.

On the reading of member as has_part we get the unintended inference.

There are a number of possible fixes, but before offering them I thought I would first check in about the above. I need to sort this out for a prototype I am currently working on.

Other cases of the anatomical set issue in that area of anatomy

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

Some clerical errors that need fixing are a number of orphan tooth surfaces and parts where there is no asserted relation to the corresponding tooth:

Distolingual cusp of right upper first molar tooth 

Occlusal surface of left lower first molar tooth 

Crown proper of left upper first molar tooth 

Incisal surface of right upper canine tooth 

Lingual surface of left upper second molar tooth 

Distal surface of right lower second premolar tooth 

Mesial surface of right upper third molar tooth 

Occlusal surface of right lower third molar tooth 

Finally, I note that the relation of the surfaces of the teeth are related by part_of to the teeth, but should instead be related by  bounded_by, as is documented in the papers and as occurs elsewhere in FMA. 

Best,

Alan

[mej...@comcast.net]

Alan,

My comments in-line  below.

Onard

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>
To: "fma-owl-2009" <fma-ow...@googlegroups.com>
Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>
Sent: Thursday, December 8, 2011 11:32:56 PM
Subject: Anomalous anatomical set members

In reviewing, and then attempting to debug, the representation of teeth, I came across a number of cases that seem to be in error.

The definition of Anatomical set is:

material anatomical entity which has as its members the maximal number of anatomical structures of the same type.
While some of anatomical sets could be regarded as an organ system subvision, e.g set of lumbar nerves, others do not satisfy one or two defining attributes of organ system subdivision. E.g. thoracic viscera consists of organs but not belonging to one organ system, dental arcade consists of similar organs but they are not connected.

OM: In the FMA no anatomical set is regarded as an organ system subdivision and vice versa. The latter is an anatomical structures consisting of continuous parts as opposed to the former which consists of discontinuous members.

I ran a function to find cases (~350) where anatomical set members are not anatomical structures. The cases are listed at http://svn.neurocommons.org/svn/trunk/convert/fma/reports/anatomical-sets-with-not-structure-members.txt

I'd also like to test my understanding of the member versus part relationships.

member_of => part_of

part_of: transitive

member_of: not transitive

member_of o part_of => member_of

OM: We consider membership as a kind of parthood relation in the FMA, hence it has transitivity property. Member_of is a subproperty of part_of.

The last one is the one to check - it says that if a set has a part, and the part has members, then the set has those members. 

If that is accepted, then some of the above cases might be recast using part_of. However I suspect that they should all be checked, given the situation described below.

In the specific case of dentition, the issue is that an uncareful interpretation of the current fma is that the primary dentition include secondary teeth.

That arises because 

'Dentition'(FMA_75150) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dentition'(FMA_75150) has member 'Maxillary dentition'(FMA_269582) which is not anatomical structure

 

and 'Primary dentition' is_a 'Dentition'

and the members of 'Maxillary dentition' are all secondary teeth.

OM: Using the frame-based system and the "single inheritance" principle posed some limitations to the representation. Dentition is a good example.

Ideally we would like to declare in is_a hierarchy:

Dentition
      Primary dentition
           Primary maxillary dentition
           Primary mandibular dention
      Secondary dentition
           Secondary maxillary dentition
           Secondary mandibular dentition

As well as...

Maxillary dentition
     Primary maxillary dentition
     Secondary maxillary dentition

Mandibular dentition
     Primary mandibular dention
     Secondary mandibular dentition

And then declare the appropriate primary and secondary teeth to the corresponding maxillary and mandibular sets. We hope to achieve this in the OWL format. In the current version the maxillary and mandibular dentitions have secondary teeth members but that will change as soon as we move to OWL.

On the reading of member as has_part we get the unintended inference.

There are a number of possible fixes, but before offering them I thought I would first check in about the above. I need to sort this out for a prototype I am currently working on.

Other cases of the anatomical set issue in that area of anatomy

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

OM: What is missing in the definition is the inclusion of subsets as members (parts) of a the bigger set.

Some clerical errors that need fixing are a number of orphan tooth surfaces and parts where there is no asserted relation to the corresponding tooth:

Distolingual cusp of right upper first molar tooth 

Occlusal surface of left lower first molar tooth 

Crown proper of left upper first molar tooth 

Incisal surface of right upper canine tooth 

Lingual surface of left upper second molar tooth 

Distal surface of right lower second premolar tooth 

Mesial surface of right upper third molar tooth 

Occlusal surface of right lower third molar tooth

OM: Thank you for the list. I will add the links to the corresponding teeth.

Finally, I note that the relation of the surfaces of the teeth are related by part_of to the teeth, but should instead be related by  bounded_by, as is documented in the papers and as occurs elsewhere in FMA.

OM: That indeed is not right. I think the original intent was to represent the surface layer (3-D structure) but in the process they were classified as surfaces (2-D). I will look into this and apply the corrections.

Again, many thanks for the heads up.
Onard

Best,

Alan

[Alan Ruttenberg]

On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:

Alan,

My comments in-line  below.

Onard

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>
To: "fma-owl-2009" <fma-ow...@googlegroups.com>
Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>
Sent: Thursday, December 8, 2011 11:32:56 PM
Subject: Anomalous anatomical set members

In reviewing, and then attempting to debug, the representation of teeth, I came across a number of cases that seem to be in error.

The definition of Anatomical set is:

material anatomical entity which has as its members the maximal number of anatomical structures of the same type.
While some of anatomical sets could be regarded as an organ system subvision, e.g set of lumbar nerves, others do not satisfy one or two defining attributes of organ system subdivision. E.g. thoracic viscera consists of organs but not belonging to one organ system, dental arcade consists of similar organs but they are not connected.

OM: In the FMA no anatomical set is regarded as an organ system subdivision and vice versa. The latter is an anatomical structures consisting of continuous parts as opposed to the former which consists of discontinuous members.

Right. My point is that the definitions disallows members of anatomical sets being other anatomical sets(many of the cases listed) or things that are not anatomical structures, such as blood:

 

'Hematopoietic system'(FMA_9667) has member 'Portion of blood'(FMA_9670) which is not anatomical structure

I ran a function to find cases (~350) where anatomical set members are not anatomical structures. The cases are listed at http://svn.neurocommons.org/svn/trunk/convert/fma/reports/anatomical-sets-with-not-structure-members.txt

I'd also like to test my understanding of the member versus part relationships.

member_of => part_of

part_of: transitive

member_of: not transitive

member_of o part_of => member_of

OM: We consider membership as a kind of parthood relation in the FMA, hence it has transitivity property. Member_of is a subproperty of part_of.

It makes sense that member_of be a kind of parthood relation. However member_of is generally not considered to be transitive, and it doesn't become transitive solely by being a subproperty of another transitive property.  Suppose you have a set (call it "coin set set") whose members are sets of coins, one generally doesn't consider the coins to also be members of "coin set set". Also, since anatomical structure and anatomical set are disjoint, there shouldn't ever be an opportunity for transitivity to be exercised, since member_of isn't an appropriate relation to relate anatomical structures to other things. 

(1) Anatomical set -- has_member --> (2) Anatomical structure --- // has member (not!) // --> (3) anything else.

For has_member to be meaningfully transitive, it would need to be the case that member_of was a valid relation for (2).

 

The last one is the one to check - it says that if a set has a part, and the part has members, then the set has those members. 

If that is accepted, then some of the above cases might be recast using part_of. However I suspect that they should all be checked, given the situation described below.

In the specific case of dentition, the issue is that an uncareful interpretation of the current fma is that the primary dentition include secondary teeth.

That arises because 

'Dentition'(FMA_75150) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dentition'(FMA_75150) has member 'Maxillary dentition'(FMA_269582) which is not anatomical structure

 

and 'Primary dentition' is_a 'Dentition'

and the members of 'Maxillary dentition' are all secondary teeth.

OM: Using the frame-based system and the "single inheritance" principle posed some limitations to the representation. Dentition is a good example.

It's potentially the single inheritance, not the frame system.

 

Ideally we would like to declare in is_a hierarchy:

Dentition
      Primary dentition
           Primary maxillary dentition
           Primary mandibular dention
      Secondary dentition
           Secondary maxillary dentition
           Secondary mandibular dentition

As well as...

Maxillary dentition
     Primary maxillary dentition
     Secondary maxillary dentition

Mandibular dentition

     Primary mandibular dentition

     Secondary mandibular dentition

And then declare the appropriate primary and secondary teeth to the corresponding maxillary and mandibular sets. We hope to achieve this in the OWL format. In the current version the maxillary and mandibular dentitions have secondary teeth members but that will change as soon as we move to OWL.

Right, since that's what I thought you were trying to accomplish. But consider this, which may be more than adequate (parentheses show parent class)

Dentition (anatomical set)
      has_part Primary dentition (anatomical set)
             has_part Primary maxillary dentition (anatomical set)
             has_part Primary mandibular dentition (anatomical set)
      has_part Secondary dentition (anatomical set)
            has_part Secondary maxillary dentition (anatomical set)
            has_part Secondary mandibular dentition (anatomical set)

Maxillary dentition (anatomical set)
     has_part Primary maxillary dentition (anatomical set)
     has_part Secondary maxillary dentition (anatomical set)

Mandibular dentition (anatomical set)
     has_part Primary mandibular dentition (anatomical set)
     has_part Secondary mandibular dentition (anatomical set)

In this view, the hierarchy is completely a partonomy, and all the terms are is_a anatomical set, so no multiple inheritance is triggered.

If you wanted to pull out as a type primary versus secondary and a add a bit more is_a structure

(that would be my guess as to the sensible thing) then you would have:

Primary tooth (type)

Secondary tooth (type)

set of teeth =def anatomical set whose members are tooth

set of primary teeth =def set of teeth whose members are primary tooth

set of secondary teeth =def set of teeth whose members are secondary tooth

Dentition (set of teeth)
  has_part Maxillary dentition (set of teeth)

  has_part Mandibular dentition (set of teeth)

Maxillary dentition (set of teeth)
     has_part Primary maxillary dentition (set of primary teeth)
     has_part Secondary maxillary dentition (set of secondary teeth)

Mandibular dentition (set of teeth)
     has_part Primary mandibular dentition (set of primary teeth)
     has_part Secondary mandibular dentition (set of secondary teeth)

Primary dentition (set of primary teeth)

Secondary dentition (set of secondary teeth)

The have each tooth be 

is_a either primary or secondary tooth, 

and member either primary or secondary dentition

and member one of Primary mandibular dentition, or Primary maxillary dentition, Secondary mandibular dentition, or Secondary maxillary dentition

Assuming I have confused myself, I believe this captures all the anatomical sets you want, and avoids multiple inheritance.

  

On the reading of member as has_part we get the unintended inference.

There are a number of possible fixes, but before offering them I thought I would first check in about the above. I need to sort this out for a prototype I am currently working on.

Other cases of the anatomical set issue in that area of anatomy

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Set of periodontia'(FMA_269567) which is not anatomical structure

'Dental arcade'(FMA_59415) has member 'Dentition'(FMA_75150) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Set of mandibular periodontia'(FMA_269572) which is not anatomical structure

'Mandibular dental arcade'(FMA_55635) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

OM: What is missing in the definition is the inclusion of subsets as members (parts) of a the bigger set.

Right, that was my point earlier about: member_of o part_of => member_of.

If you have subsets be part of sets, then with the above rule (which can be expressed in OWL 2) you will get the intended behavior.

 

Some clerical errors that need fixing are a number of orphan tooth surfaces and parts where there is no asserted relation to the corresponding tooth:

Distolingual cusp of right upper first molar tooth 

Occlusal surface of left lower first molar tooth 

Crown proper of left upper first molar tooth 

Incisal surface of right upper canine tooth 

Lingual surface of left upper second molar tooth 

Distal surface of right lower second premolar tooth 

Mesial surface of right upper third molar tooth 

Occlusal surface of right lower third molar tooth

OM: Thank you for the list. I will add the links to the corresponding teeth.

Finally, I note that the relation of the surfaces of the teeth are related by part_of to the teeth, but should instead be related by  bounded_by, as is documented in the papers and as occurs elsewhere in FMA.

OM: That indeed is not right. I think the original intent was to represent the surface layer (3-D structure) but in the process they were classified as surfaces (2-D). I will look into this and apply the corrections.

Again, many thanks for the heads up.

Welcome. If you plan to do this in the near future, could you let me know and I'll wait to pick up your changes, which would save me a bit of time.

 

Onard

Best,

Alan

[Alan Ruttenberg]

One more thing ;-)

The sets of teeth are unlike some other sets as they gather members that do not necessarily exist at the same time. I wonder if there is precedent elsewhere in the FMA for that, or whether it might make more sense to stay away from that and not create as sets dentition (meaning set of all primary and secondary teeth), mandibular dention (also mixes primary and secondary teeth) or maxilliary dentition (also mixes primary and secondary teeth).

-Alan

On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:

[ro...@u.washington.edu]

Alan, it is true for sets of anatomical entities in the fully developed vertebral body that their members do not exist at the same time. For example, members of sets vertebrae, sets of muscles, etc appear during ontogeny at separate times. Hence adding a time element to the definitions is always necessary. I realize that FMA definitions do not do this, because the FMA takes account of a canonical, fully developed human body. I realize that there are arguments for and against the justification of such a strategy.

Cornelius

On Fri, 9 Dec 2011, Alan Ruttenberg wrote:

> One more thing ;-)
> The sets of teeth are unlike some other sets as they gather members that do not necessarily exist at the same time. I wonder
> if there is precedent elsewhere in the FMA for that, or whether it might make more sense to stay away from that and not create
> as sets dentition (meaning set of all primary and secondary teeth), mandibular dention (also mixes primary and secondary
> teeth) or maxilliary dentition (also mixes primary and secondary teeth).
>
> -Alan
>
> On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:
> Alan,
>
> My comments in-line below.
>
> Onard
>

> ______________________________________________________________________________________________________________________________

[Alan Ruttenberg]

On Fri, Dec 9, 2011 at 12:41 PM, <ro...@u.washington.edu> wrote:

Alan, it is true for sets of anatomical entities in the fully developed vertebral body that their members do not exist at the same time. For example, members of sets vertebrae, sets of muscles, etc appear during ontogeny at separate times. Hence adding a time element to the definitions is always necessary. I realize that FMA definitions do not do this, because the FMA takes account of a canonical, fully developed human body. I realize that there are arguments for and against the justification of such a strategy.

Hi Cornelius,

Thanks for the response (happy to hear your virtual voice :) . 

Dentition is one of those in-between cases which slightly stretches the model to allow for two canonical cases - childhood and adulthood, with reasonably clear delineation between the immature and mature forms. My inclination would therefore be to only use Anatomical set for each of these separately, with the same caveats you mention earlier applicable, and not to name sets that include members of both as, given the current FMA, the members of Anatomical sets seem to generally be more approximately contemporaneous than is the case with primary and secondary teeth.

Would you concur, or do you see benefit in having Anatomical sets that have both as members?

Regards,

Alan

ps. Barry and I frequently argue about what you would say about this or that issue. It would be great fun to find a time again when we could all sit together and have it out :)

[mej...@comcast.net]

Alan,

The class "Dentition" does not have both primary and secondary teeth as members but rather it has subtypes primary and secondary. The members of class "Dentition" are maxillary dentition and mandibular dentition, and the subtype of "Dentition", "Primary dentition", has members primary maxillary and primary mandibular sets (same rationale for the secondary dentition). Currently in the FMA, you will find that "Primary dentition" has as members only the primary teeth.

Temporal property is of great interest to me and one I would like to explore for the FMA. As Cornelius mentioned, not all members exist at the same time but remember we are dealing with a continuant, one that takes into account its canonical completeness over a period of time, not at a one specific stage.

Onard

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>

To: fma-ow...@googlegroups.com
Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>, "Cornelius Rosse" <ro...@u.washington.edu>, "Onard Mejino" <mej...@u.washington.edu>
Sent: Friday, December 9, 2011 10:19:57 AM
Subject: Re: Anomalous anatomical set members

[Alan Ruttenberg]

On Fri, Dec 9, 2011 at 1:32 PM, mej...@u.washington.edu <mej...@u.washington.edu> wrote:

Alan,

The class "Dentition" does not have both primary and secondary teeth as members but rather it has subtypes primary and secondary. The members of class "Dentition" are maxillary dentition and mandibular dentition, and the subtype of "Dentition", "Primary dentition", has members primary maxillary and primary mandibular sets (same rationale for the secondary dentition).

Yes. But I think that's wrong. If Dentition has members and subclasses, then instances of the subclasses also have those members= by inheritance. As maxillary dentition would include both primary and secondary teeth, the inheritance would lead to an incorrect membership for Primary dentition (and secondary dentition).

In general, having a set type that has both members and subtypes seems to me to be a recipe for confusion and error, and while it might work in certain cases, it seems best avoided.

  

Currently in the FMA, you will find that "Primary dentition" has as members only the primary teeth.

Plus the ones inherited from dentition, which include secondary teeth.

 

Temporal property is of great interest to me and one I would like to explore for the FMA. As Cornelius mentioned, not all members exist at the same time but remember we are dealing with a continuant, one that takes into account its canonical completeness over a period of time, not at a one specific stage.

Agreed. What did you think of the analysis in my response with regard to primary and secondary teeth, where the temporal separation is more clear?

Best,

Alan

Onard

[Alan Ruttenberg]

Onard wrote:

Temporal property is of great interest to me and one I would like to explore for the FMA.

For starters I think it would help to simply clarify the status of Anatomical set as it is used in the FMA and augment the definition with the resulting clarification.

Is an anatomical set instantiated at any time any of its members develops, perhaps having changing membership over time, until none are present? Or is it the intention that the term denote the full set of entities only during the time that they are all present. These are different from an ontological point of view, though in both cases continuant.

-Alan

[mej...@comcast.net]

Can I ask you a favor? Can you include my UW address, mej...@uw.edu, to the email group? I'd rather respond using the UW mail than this comcast service.

Thanks.
Onard

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>

To: mej...@u.washington.edu
Cc: fma-ow...@googlegroups.com, "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>, "Cornelius Rosse" <ro...@u.washington.edu>
Sent: Friday, December 9, 2011 10:40:37 AM

Subject: Re: Anomalous anatomical set members

[mej...@comcast.net]

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>

To: mej...@u.washington.edu
Cc: fma-ow...@googlegroups.com, "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>, "Cornelius Rosse" <ro...@u.washington.edu>
Sent: Friday, December 9, 2011 10:40:37 AM
Subject: Re: Anomalous anatomical set members

On Fri, Dec 9, 2011 at 1:32 PM, mej...@u.washington.edu <mej...@u.washington.edu> wrote:

Alan,

The class "Dentition" does not have both primary and secondary teeth as members but rather it has subtypes primary and secondary. The members of class "Dentition" are maxillary dentition and mandibular dentition, and the subtype of "Dentition", "Primary dentition", has members primary maxillary and primary mandibular sets (same rationale for the secondary dentition).

Yes. But I think that's wrong. If Dentition has members and subclasses, then instances of the subclasses also have those members= by inheritance. As maxillary dentition would include both primary and secondary teeth, the inheritance would lead to an incorrect membership for Primary dentition (and secondary dentition).

OM: I'm not quite sure I follow you but consider these:

Dentition is  a class whose members consist only of teeth, without specifying which particular set of teeth.
Entire dentition is a class whose members include all kinds of teeth, primary and secondary.
Primary dentition is a subclass of 'dentition' which specifies the membership only to include primary teeth.
Maxillary dentition is a subclass of 'dentition' which specifies the membership only to include all kinds of teeth attached to the maxillae, whether they are primary or secondary.
Entire maxillary dentition is a subclass of 'maxillary dentition' which specifies the membership only to include all kinds of teeth attached to the maxillae, whether they are primary or secondary.
Primary maxillary dentition is a subclass of 'maxillary dentition' which specifies the membership only to include primary teeth attached to the maxillae.

And the instances of each class are those which satisfy the properties of that particular class.

Onard

[mej...@comcast.net]

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>

To: fma-ow...@googlegroups.com

Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>

Sent: Friday, December 9, 2011 7:37:31 AM

Subject: Re: Anomalous anatomical set members

On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:

Alan,

My comments in-line  below.

Onard

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>
To: "fma-owl-2009" <fma-ow...@googlegroups.com>
Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>
Sent: Thursday, December 8, 2011 11:32:56 PM
Subject: Anomalous anatomical set members

In reviewing, and then attempting to debug, the representation of teeth, I came across a number of cases that seem to be in error.

The definition of Anatomical set is:

material anatomical entity which has as its members the maximal number of anatomical structures of the same type.
While some of anatomical sets could be regarded as an organ system subvision, e.g set of lumbar nerves, others do not satisfy one or two defining attributes of organ system subdivision. E.g. thoracic viscera consists of organs but not belonging to one organ system, dental arcade consists of similar organs but they are not connected.

OM: In the FMA no anatomical set is regarded as an organ system subdivision and vice versa. The latter is an anatomical structures consisting of continuous parts as opposed to the former which consists of discontinuous members.

Right. My point is that the definitions disallows members of anatomical sets being other anatomical sets(many of the cases listed) or things that are not anatomical structures, such as blood:

 

'Hematopoietic system'(FMA_9667) has member 'Portion of blood'(FMA_9670) which is not anatomical structure

OM: The definition of anatomical set does not disallow body substances, that why we have the class "Hematopoietic system".
We also have sets of immaterial anatomical entities, such as "Set of subarachnoid sulci" and "Set of external nares".

I ran a function to find cases (~350) where anatomical set members are not anatomical structures. The cases are listed at http://svn.neurocommons.org/svn/trunk/convert/fma/reports/anatomical-sets-with-not-structure-members.txt

I'd also like to test my understanding of the member versus part relationships.

member_of => part_of

part_of: transitive

member_of: not transitive

member_of o part_of => member_of

OM: We consider membership as a kind of parthood relation in the FMA, hence it has transitivity property. Member_of is a subproperty of part_of.

It makes sense that member_of be a kind of parthood relation. However member_of is generally not considered to be transitive, and it doesn't become transitive solely by being a subproperty of another transitive property.  Suppose you have a set (call it "coin set set") whose members are sets of coins, one generally doesn't consider the coins to also be members of "coin set set". Also, since anatomical structure and anatomical set are disjoint, there shouldn't ever be an opportunity for transitivity to be exercised, since member_of isn't an appropriate relation to relate anatomical structures to other things.

OM:
Set of C1, C2 and C3 vertebrae is a set of Cervical vertebrae
Set of Cervical vertebrae is a set of vertebrae
Set of C1, C2 and C3 vertebrae is a set of vertebrae
Set of vertebrae is a set of axial bones
Set of axial bones is a set of bones
Set of C1, C2 and C3 vertebrae is a set of bones

And yes, there is a relationship between an anatomical structure and a set of anatomical structures:

Hand has constitutional_part
      Skin of hand
      Superficial fascia of hand
      Musculature of hand (Anatomical set)
      Skeleton of hand (Anatomical set)
      Neural network of hand
      Vasculature of hand
  

(1) Anatomical set -- has_member --> (2) Anatomical structure --- // has member (not!) // --> (3) anything else.

For has_member to be meaningfully transitive, it would need to be the case that member_of was a valid relation for (2).

 

The last one is the one to check - it says that if a set has a part, and the part has members, then the set has those members. 

If that is accepted, then some of the above cases might be recast using part_of. However I suspect that they should all be checked, given the situation described below.

In the specific case of dentition, the issue is that an uncareful interpretation of the current fma is that the primary dentition include secondary teeth.

That arises because 

'Dentition'(FMA_75150) has member 'Mandibular dentition'(FMA_269584) which is not anatomical structure

'Dentition'(FMA_75150) has member 'Maxillary dentition'(FMA_269582) which is not anatomical structure

 

and 'Primary dentition' is_a 'Dentition'

and the members of 'Maxillary dentition' are all secondary teeth.

OM: Using the frame-based system and the "single inheritance" principle posed some limitations to the representation. Dentition is a good example.

It's potentially the single inheritance, not the frame system.

You're right, I can have multiple inheritance in frames.

OM: I addressed this in the previous message.

[mej...@comcast.net]

Corrections, I meant:

Maxillary dentition is a subclass of 'dentition' which specifies the membership only to include teeth attached to the maxillae, without specifying whether they are primary or secondary.
Entire maxillary dentition is a subclass of 'maxillary dentition' which specifies the membership to include all kinds of teeth attached to the maxillae.

---

From: mej...@comcast.net
To: fma-ow...@googlegroups.com
Cc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai" <tor...@ohsu.edu>, "Cornelius Rosse" <ro...@u.washington.edu>, mej...@u.washington.edu
Sent: Friday, December 9, 2011 11:22:16 AM

[ro...@u.washington.edu]

Alan,

Yes; an eminently sensible proposition.

Cornelius

On Fri, 9 Dec 2011, Alan Ruttenberg wrote:

>
>
> On Fri, Dec 9, 2011 at 12:41 PM, <ro...@u.washington.edu> wrote:
> Alan, it is true for sets of anatomical entities in the fully developed vertebral body that their members do not
> exist at the same time. For example, members of sets vertebrae, sets of muscles, etc appear during ontogeny at
> separate times. Hence adding a time element to the definitions is always necessary. I realize that FMA definitions
> do not do this, because the FMA takes account of a canonical, fully developed human body. I realize that there are
> arguments for and against the justification of such a strategy.
>
>
> Hi Cornelius,
>

> Thanks for the response (happy to hear your virtual voice :) .ￜ

>
> Dentition is one of those in-between cases which slightly stretches the model to allow for two canonical cases - childhood and
> adulthood, with reasonably clear delineation between the immature and mature forms. My inclination would therefore be to only
> use Anatomical set for each of these separately, with the same caveats you mention earlier applicable, and not to name sets
> that include members of both as, given the current FMA, the members of Anatomical sets seem to generally be more approximately
> contemporaneous than is the case with primary and secondary teeth.
>
> Would you concur, or do you see benefit in having Anatomical sets that have both as members?
>
> Regards,
> Alan
>
> ps. Barry and I frequently argue about what you would say about this or that issue. It would be great fun to find a time again
> when we could all sit together and have it out :)

> ￜ

>
>
> Cornelius
>
> On Fri, 9 Dec 2011, Alan Ruttenberg wrote:
>
> One more thing ;-)
> The sets of teeth are unlike some other sets as they gather members that do not necessarily exist at the
> same time. I wonder
> if there is precedent elsewhere in the FMA for that, or whether it might make more sense to stay away from
> that and not create
> as sets dentition (meaning set of all primary and secondary teeth), mandibular dention (also mixes primary
> and secondary
> teeth) or maxilliary dentition (also mixes primary and secondary teeth).
>
> -Alan
>
> On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:

> ￜ ￜ ￜAlan,
>
> ￜ ￜ ￜMy comments in-line ￜbelow.
>
> ￜ ￜ ￜOnard
>
> _____________________________________________________________________________________________________________________________
> _
> ￜ ￜ ￜFrom: "Alan Ruttenberg" <alanrut...@gmail.com>
> ￜ ￜ ￜTo: "fma-owl-2009" <fma-ow...@googlegroups.com>
> ￜ ￜ ￜCc: "Melissa Haendel" <hae...@ohsu.edu>, "Bill Duncan" <wddu...@gmail.com>, "Carlo Torniai"
> <tor...@ohsu.edu>
> ￜ ￜ ￜSent: Thursday, December 8, 2011 11:32:56 PM
> ￜ ￜ ￜSubject: Anomalous anatomical set members
>
> ￜ ￜ ￜIn reviewing, and then attempting to debug, the representation of teeth, I came across a number of cases that
> seem
> ￜ ￜ ￜto be in error.
>
> ￜ ￜ ￜThe definition of Anatomical set is:
>
> ￜ ￜ ￜ ￜ ￜ ￜmaterial anatomical entity which has as its members the maximal number of anatomical structures of the
> ￜ ￜ ￜ ￜ ￜ ￜsame type.
> ￜ ￜ ￜ ￜ ￜ ￜWhile some of anatomical sets could be regarded as an organ system subvision, e.g set of lumbar
> ￜ ￜ ￜ ￜ ￜ ￜnerves, others do not satisfy one or two defining attributes of organ system subdivision. E.g.
> ￜ ￜ ￜ ￜ ￜ ￜthoracic viscera consists of organs but not belonging to one organ system, dental arcade consists of
> ￜ ￜ ￜ ￜ ￜ ￜsimilar organs but they are not connected.

> ï¿œand 'Primary dentition' is_a 'Dentition'

>
> and the members of 'Maxillary dentition' are all secondary teeth.
>
> OM: Using the frame-based system and the "single inheritance" principle posed some limitations to the
> representation.
> Dentition is a good example.
>
> Ideally we would like to declare in is_a hierarchy:
>
> Dentition

> ￜ ￜ ￜPrimary dentition
> ￜ ￜ ￜ ￜ ￜ Primary maxillary dentition
> ￜ ￜ ￜ ￜ ￜ Primary mandibular dention
> ￜ ￜ ￜSecondary dentition
> ￜ ￜ ￜ ￜ ￜ Secondary maxillary dentition
> ￜ ￜ ￜ ￜ ￜ Secondary mandibular dentition

>
> As well as...
>
> Maxillary dentition

> ￜ ￜ Primary maxillary dentition
> ￜ ￜ Secondary maxillary dentition
>
> Mandibular dentition
> ￜ ￜ Primary mandibular dention
> ￜ ￜ Secondary mandibular dentition

> be related by ï¿œbounded_by, as is documented in the papers and as occurs elsewhere in FMA.

[Chris Mungall]

On Dec 9, 2011, at 7:37 AM, Alan Ruttenberg wrote:

OM: Using the frame-based system and the "single inheritance" principle posed some limitations to the representation. Dentition is a good example.

It's potentially the single inheritance, not the frame system.

Yes, and there are unfortunately many problem cases like this where SI has been adhered to in too rigid a fashion.

I think there's been a lot of confusion about SI vs MI. What we all generally agree on is that asserted SI plus inferred MI is a good thing.

I would go further - the _worst_ of all worlds is the combination of:

* pure SI

* large numbers of compositional classes

* no mechanism for inferring the full polyhierarchy

Unfortunately the FMA, in it's currently released forms, falls into this category. The consequence is that there are many is_a links missing in FMA between compositional classes. This confuses biologists and results in false negatives when FMA is used for querying.

The situation is easily fixed by having equivalence axioms for the compositional classes. However, as I understand this isn't possible in the legacy frame system that FMA uses. In my opinion moving away from this system (e.g. to something OWL-based) is high priority for the FMA.

An alternative solution would be to have the equivalence axions maintained in a separate OWL ontology that imports an automated OWL translation of the core FMA. Then the official releases of the FMA in OWL could be pre-reasoned and include the complete polyhierarchy.

However you choose to implement it, I would say it's high priority.

In the absence of any such solution, a way to mitigate the problem would be to release an FMA-core that has fewer pre-coordinated compositional classes. This will mitigate, though not solve the problem. I think a lot of your users might like to see this in any case.

[Alan Ruttenberg]

On Fri, Dec 9, 2011 at 4:25 AM, <mej...@comcast.net> wrote:

> OM: That indeed is not right. I think the original intent was to represent
> the surface layer (3-D structure) but in the process they were classified as
> surfaces (2-D). I will look into this and apply the corrections.

I think that while the 2d surfaces may perhaps be useful in some
contexts, I anticipate the 3d surfaces (and in general the
constitutional parts) of each tooth will be helpful to have explicitly
represented, as dental information systems do refer to damage and
restorations to such parts of specific teeth.

It has been raised by a colleague at the dental school that in
addition it may be of benefit to represent those anatomical sites in
which the teeth place, as they are also referred to (as when there is
such a site without a tooth occupying it, and when work is done to
place a tooth in such a site).

Thanks,
Alan

[Alan Ruttenberg]

On Fri, Dec 9, 2011 at 2:41 PM, <mej...@comcast.net> wrote:
> Right. My point is that the definitions disallows members of anatomical sets
> being other anatomical sets(many of the cases listed) or things that are not
> anatomical structures, such as blood:
>
> 'Hematopoietic system'(FMA_9667) has member 'Portion of blood'(FMA_9670)
> which is not anatomical structure
>
>
> OM: The definition of anatomical set does not disallow body substances, that
> why we have the class "Hematopoietic system".
> We also have sets of immaterial anatomical entities, such as "Set of
> subarachnoid sulci" and "Set of external nares".

The definition of Anatomical set as provided by the FMA is:

"material anatomical entity which has as its members the maximal
number of anatomical structures of the same type."

As I read the current FMA, portion of body substance, being a sibling
of Anatomical set, is disjoint from it. So the definition contradicts
the assertion. I am advocating that one of them be changed so that
there is no longer a contradiction.

I would also suggest that the definition be changed to not say
anything about the "maximal number", as this raises issues regarding
time. Instead, push any kind of such maximality down to the particular
terms. For example, the term "dentition" could be defined as being the
set of all the teeth that are present at any time, if desired.

-Alan

[Alan Ruttenberg]

> On Fri, Dec 9, 2011 at 1:32 PM, mej...@u.washington.edu
> <mej...@u.washington.edu> wrote:
>>
>> Alan,
>>
>> The class "Dentition" does not have both primary and secondary teeth as
>> members but rather it has subtypes primary and secondary. The members of
>> class "Dentition" are maxillary dentition and mandibular dentition, and the
>> subtype of "Dentition", "Primary dentition", has members primary maxillary
>> and primary mandibular sets (same rationale for the secondary dentition).
>
>
> Yes. But I think that's wrong. If Dentition has members and subclasses, then
> instances of the subclasses also have those members= by inheritance.
> As maxillary dentition would include both primary and secondary teeth, the
> inheritance would lead to an incorrect membership for Primary dentition (and
> secondary dentition).
>
> OM: I'm not quite sure I follow you but consider these:
>

Note we are going to stumble on words (and perhaps already have) by
using "members" to both name instances of the class, for members of a
set (a particular). In the following I will rename to "instance" uses
of "member" that I think you meant to mean instance of class. Please
check and see I got your sense correctly, or correct?

> Dentition is  a class whose members consist only of teeth, without specifying which particular set of teeth.

Dentition is an Anatomical set whose members are only teeth.

> Entire dentition is a class whose members include all kinds of teeth,

(see below re: discussion with Cornelius)

> primary and secondary.
> Primary dentition is a subclass of 'dentition' which specifies the
> membership only to include primary teeth.

Primary dentition is a dentition, all of whose members are all (all
extant?) primary teeth
Secondary dentition is a dentition, all of whose members are (all
extant?) secondary teeth

> Maxillary dentition is a subclass of 'dentition' which specifies the membership only to include teeth attached to the maxillae, without specifying whether they are primary or secondary.

Maxillary dentition is a dentition, all of whose members are attached
to the maxillae.

> Entire maxillary dentition is a subclass of 'maxillary dentition' which specifies the membership to include all kinds of teeth attached to the maxillae.

Given the discussion with Cornelius I don't think we should have this
set. It would make sense, however to have a class

You might want
Maxillary tooth - a tooth that is attached to the maxillae.
(in the site formulation it would be a tooth that occupies a site
hosted by the maxillae)

> Primary maxillary dentition is a subclass of 'maxillary dentition' which
> specifies the membership only to include primary teeth attached to the
> maxillae.

Primary maxillary dentition is a maxillary dentition all of whose
members are primary teeth.

I think we also want (looking at
http://en.wikipedia.org/wiki/Tooth_development for a reasonable
discriminator)
(but note this would raise multiple inheritance issues with maxillary
tooth, above)

Primary tooth
A tooth whose initial calcification is completed in-utero.

Secondary tooth
A tooth whose initial calcification is completed after birth.

There is a minor twist, also mentioned by my Dental School colleague.
It is occurs not so unoften that a developmentally primary tooth is
classified as a secondary tooth later in life if a corresponding
secondary tooth does not develop to push the primary tooth out. A
definition of primary and secondary based on site would be able to
avoid this problem. Instead we would define primary tooth sites and
secondary tooth sites, then primary and secondary teeth as those that
eventually occupy those sites. This would also require that primary
and secondary tooth not be disjoint (since some primary teeth could
transform (in the BFO relations sense) into secondary teeth.

-Alan

[Todd Detwiler]

Alan, where are you getting this definition of Anatomical set? I have looked in the live FMA, the last release version of the FMA (v3.2) and the version before that (from early 2010, v3.1). All have the following definition for Anatomical set:

"Material anatomical entity which consists of the maximum number of discontinuous members of the same class. Examples: set of cranial nerves, ventral branches of aorta, set of mammary arteries, thoracic viscera, dental arcade."

It does not say that the members of the set must be from the class Anatomical structure.

Todd

Landon Todd Detwiler
Structural Informatics Group (SIG)
University of Washington

[mej...@comcast.net]

---

From: "Alan Ruttenberg" <alanrut...@gmail.com>

To: fma-ow...@googlegroups.com
Sent: Tuesday, December 13, 2011 10:08:29 AM

Subject: Re: Anomalous anatomical set members

On Fri, Dec 9, 2011 at 4:25 AM,  <mej...@comcast.net> wrote:
> OM: That indeed is not right. I think the original intent was to represent
> the surface layer (3-D structure) but in the process they were classified as
> surfaces (2-D). I will look into this and apply the corrections.

I think that while the 2d surfaces may perhaps be useful in some
contexts, I anticipate the 3d surfaces (and in general the
constitutional parts) of each tooth will be helpful to have explicitly
represented, as dental information systems do refer to damage and
restorations to such parts of specific teeth.

OM: I agree. I will look into this.

It has been raised by a colleague at the dental school that in
addition it may be of benefit to represent those anatomical sites in
which the teeth place, as they are also referred to (as when there is
such a site without a tooth occupying it, and when work is done to
place a tooth in such a site).

OM: We do have "Alveolar compartment"  and representation is carried out to the specific tooth involved, such as "Alveolar compartment in tooth socket 9" (syn: Alveolar compartment in socket for left upper central secondary incisor tooth). The cavity or space that contains the compartment is called "Tooth socket" (e.g. Tooth socket 9 or Socket for left upper central secondary incisor tooth). The alveolar compartment has parts, root of tooth, periodontal ligament and associated neural and vascular networks. So if the tooth is missing, one can create a class called "Adental alveolar compartment 9" (I'm making up the "adental" term, the dental experts may have a better term) which consists of the same parts minus the tooth (root of tooth).




Thanks,
Alan

[Chris Mungall]

Are these definitions being coordinated with CARO?

I don't remember the exact reasons we don't have anatomical set in CARO.

In the original CARO, which was based heavily on the FMA, we had:

  is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity that has mass."]

   is_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical entity that has inherent 3D shape and is generated by coordinated expression of the organism's own genome."]

    is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure consisting of at least two non-overlapping organs, multi-tissue aggregates or portion of tissues or cells of different types that does not constitute an organism, organ, multi-tissue aggregate, or portion of tissue."]

     is_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is has as its parts distinct anatomical structures interconnected by anatomical structures at a lower level of granularity."]

     is_a CARO:0000041 ! anatomical cluster  [DEF: "Anatomical group that has its parts adjacent to one another."]

Dentition would seem to fit 'anatomical cluster' - unless we consider a child who has a mixed dentition, consisting of a mixture of primary dentition, secondary dentition, and sets of unerupted teeth (maybe this isn't "canonical")

[Chris Mungall]

The naming in the FMA is a little confusing:

 / FMA:62955 ! Anatomical entity

  is_a FMA:61775 ! Physical anatomical entity

   is_a FMA:67112 ! Immaterial anatomical entity

    is_a FMA:71917 ! Set of immaterial anatomical entities ***

     is_a FMA:71918 ! Set of anatomical spaces

      is_a FMA:76576 ! Set of cavities

   is_a FMA:67165 ! Material anatomical entity

    is_a FMA:55652 ! Anatomical set *** 

 

"anatomical set" should be renamed "set of material anatomical entities".

You could introduce "anatomical set" as a different class that is a set of material or immaterial entities, but this would go against the rigid SI principle in the FMA.

[Alan Ruttenberg]

On Tue, Dec 13, 2011 at 1:38 PM, Todd Detwiler <d...@u.washington.edu> wrote:
> Alan, where are you getting this definition of Anatomical set? I have looked
> in the live FMA, the last release version of the FMA (v3.2) and the version
> before that (from early 2010, v3.1). All have the following definition for
> Anatomical set:
>
> "Material anatomical entity which consists of the maximum number of
> discontinuous members of the same class. Examples: set of cranial nerves,
> ventral branches of aorta, set of mammary arteries, thoracic viscera, dental
> arcade."

I'll check my version. I see that the FMAE has the version you cite,
which I'm happy to substitute.
It too is flawed, because the constraint "being members of the same
class" is effectively always true for anything, since everything is a
n instance of Anatomical entity. So you might as well leave that out.
Indeed I see that "Sets of parts of human body" is a subclass, which
makes the point.

-Alan

[Alan Ruttenberg]

On Tue, Dec 13, 2011 at 3:07 PM, Chris Mungall <cjmu...@lbl.gov> wrote:
>
> Are these definitions being coordinated with CARO?
>
> I don't remember the exact reasons we don't have anatomical set in CARO.
> In the original CARO, which was based heavily on the FMA, we had:
>
>   is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity
> that has mass."]
>    is_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical
> entity that has inherent 3D shape and is generated by coordinated expression
> of the organism's own genome."]
>     is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure
> consisting of at least two non-overlapping organs, multi-tissue aggregates
> or portion of tissues or cells of different types that does not constitute
> an organism, organ, multi-tissue aggregate, or portion of tissue."]

I'm not sure what case is covered by cells of different types as
opposed to tissues. Is there a collection of cells of different types
that is not an organ, multi-tissue aggregate or portion of tissue, and
yet still has "inherent 3D shape and is generated by coordinated
expression of the organism's own genome"?

>      is_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is
> has as its parts distinct anatomical structures interconnected by anatomical
> structures at a lower level of granularity."]
>      is_a CARO:0000041 ! anatomical cluster  [DEF: "Anatomical group that
> has its parts adjacent to one another."]
>
> Dentition would seem to fit 'anatomical cluster' - unless we consider a
> child who has a mixed dentition, consisting of a mixture of primary
> dentition, secondary dentition, and sets of unerupted teeth (maybe this
> isn't "canonical")

Teeth aren't necessarily adjacent (at least in the common sense),
though I see the RO sense is simply nearby "in spatial proximity".
Since the mouth can open quite wide it's not clear they would be
considered adjacent in even that sense.

FMA has both Anatomical Set as well as Anatomical Structure, BTW.

Ii haven't been coordinating with CARO but will make an effort to do
so in the subset of terms I extract to represent teeth.

-Alan

[Alan Ruttenberg]

OK, I tracked this down.

The definition is as you say. The "documentation" is:

"OM def (1.2.2007): material anatomical entity which has as

its members the maximal number of anatomical structures of the same

type.nn10.04.01: While some of anatomical sets could be regarded as an

organ system subvision, e.g set of lumbar nerves, others do not
satisfy one or two defining attributes of organ system subdivision.
E.g. thoracic viscera consists of organs but not belonging to one
organ system, dental arcade consists of similar organs but they are

not connected.")

So what is the status of the documentation slot value?

-Alan

On Tue, Dec 13, 2011 at 1:38 PM, Todd Detwiler <d...@u.washington.edu> wrote:

[ro...@u.washington.edu]

I feel that if CARO adopts FMA definitions, those definitions in CARO should indicate that they are derived from the FMA, as is the case with SNOMED for example.

Cornelius

On Tue, 13 Dec 2011, Chris Mungall wrote:

>
> Are these definitions being coordinated with CARO?
>
> I don't remember the exact reasons we don't have anatomical set in CARO.
>
> In the original CARO, which was based heavily on the FMA, we had:
>

> ￜ is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity that has mass."]
> ￜ ￜis_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical entity that has inherent 3D shape and is generated by

> coordinated expression of the organism's own genome."]

> ￜ ￜ is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure consisting of at least two non-overlapping organs,

> multi-tissue aggregates or portion of tissues or cells of different types that does not constitute an organism, organ,
> multi-tissue aggregate, or portion of tissue."]

> ￜ ￜ ￜis_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is has as its parts distinct anatomical structures

> interconnected by anatomical structures at a lower level of granularity."]

> ￜ ￜ ￜis_a CARO:0000041 ! anatomical cluster ￜ[DEF: "Anatomical group that has its parts adjacent to one another."]

[Alan Ruttenberg]

Onard responds saying it is more about provenance and history and as
such I would map it to IAO's editor note.

On Tue, Dec 13, 2011 at 6:33 PM, Alan Ruttenberg
<alanrut...@gmail.com> wrote:
> OK, I tracked this down.
>
> The definition is as you say. The "documentation" is:
>
>           "OM def (1.2.2007): material anatomical entity which has as
> its members the maximal number of anatomical structures of the same
> type.nn10.04.01: While some of anatomical sets could be regarded as an
> organ system subvision, e.g set of lumbar nerves, others do not
> satisfy one or two defining attributes of organ system subdivision.
> E.g. thoracic viscera consists of organs but not belonging to one
> organ system, dental arcade consists of similar organs but they are
> not connected.")
>
> So what is the status of the documentation slot value?

Onard

[Alan Ruttenberg]

On Wed, Dec 14, 2011 at 4:50 AM, David Osumi-Sutherland
<dj...@gen.cam.ac.uk> wrote:
> Fabian and I both objected to 'anatomical set' in CARO1, as its main use seemed to be to duplicate classification as partonomy. Taken to the extreme, you could make such set terms for every class (type) in an ontology.

Yup. I think I vaguely get why in some cases these terms are
desirable, but it would be better to hear from Onard/Cornelius about
what their intent and goal was in introducing these terms. They also
seem to prone to these inheritance issues (I'll produce a list soon) -
general case

class A
class B
B subclass of A
member assertions on A that are false for B.

Anyways, I want to be faithful to FMA as much as I can where it makes
sense, so my preference is that we reach some consensus on how to
handle these terms and then all of us (FMA included) adopt it.

-Alan
-Alan

>
> I'm not sure why you would want to have:
> incisor is_a tooth
> AND
> incisor part_of some 'set of teeth'
>
> Bur if you do have both, you need some way to infer the second from the first or the two will go out a sync.
>
> 'anatomical group' is needed for cases where an anatomical entity is made up of disconnected parts of various classes.  The immune system is a good example.
>
> In some cases the parts are in close proximity. In those cases 'anatomical cluster' seems appropriate to me.
>
> CARO1 didn't do a good job of making these distinctions.
>
> Note, I have never been entirely comfortable with 'anatomical cluster' as a classification for joint as, AFAIK, the various parts of joints are connected to each other.

[mej...@u.washington.edu]

On Wed, 14 Dec 2011, Alan Ruttenberg wrote:

> On Wed, Dec 14, 2011 at 4:50 AM, David Osumi-Sutherland
> <dj...@gen.cam.ac.uk> wrote:
>> Fabian and I both objected to 'anatomical set' in CARO1, as its main use seemed to be to duplicate classification as partonomy. Taken to the extreme, you could make such set terms for every class (type) in an ontology.

I vaguely remember what Fabian objected to. I had a number of discussions with him but we never had a final agreement.

>
> Yup. I think I vaguely get why in some cases these terms are
> desirable, but it would be better to hear from Onard/Cornelius about
> what their intent and goal was in introducing these terms. They also
> seem to prone to these inheritance issues (I'll produce a list soon) -
> general case
>

There are references to groups of anatomical entities in anatomical discourse. Some other connected (anatomical cluster), e.g. root of lung and some are not (anatomical set), e.g. dentition (set of teeth and nothing else, ergo separate, independent structures taken as a set). The two groups are separated by the differentia/property "connectivity", hence are represented as separate classes/types. Cornelius may have other opinion.

> class A
> class B
> B subclass of A
> member assertions on A that are false for B.

What's the example?

>
> Anyways, I want to be faithful to FMA as much as I can where it makes
> sense, so my preference is that we reach some consensus on how to
> handle these terms and then all of us (FMA included) adopt it.
>
> -Alan
> -Alan
>
>
>>
>> I'm not sure why you would want to have:
>> incisor is_a tooth
>> AND
>> incisor part_of some 'set of teeth'

We don't have the latter relationship in the FMA unless Alan's program indirectly infers that. In the FMA the actual membership declared includes only the individual teeth (by transitivity). View the part hierarchy in the FME:

Dentition (part relation)
Maxillary dentition
Left upper third secondary molar tooth
Right upper third secondary molar tooth
etc.,
Mandibular dentition
Left lower third secondary molar tooth
Right lower third secondary molar tooth
etc.

>>
>> Bur if you do have both, you need some way to infer the second from the first or the two will go out a sync.
>>

>> 'anatomical group' is needed for cases where an anatomical entity is made up of disconnected parts of various classes. ï¿œThe immune system is a good example.

Exactly, the immune system is an anatomical set, with discontinuous parts.

>>
>> In some cases the parts are in close proximity. In those cases 'anatomical cluster' seems appropriate to me.

Proximity but not connected is also an anatomical set.

Anatomical cluster has parts connected, such as the joint.

>>
>> CARO1 didn't do a good job of making these distinctions.

We elaborated on the distinction in the FMA but I'm still not clear what the objection is. I would certainly entertain a better representation that captures the distinctions.

>>
>> Note, I have never been entirely comfortable with 'anatomical cluster' as a classification for joint as, AFAIK, the various parts of joints are connected to each other.

>>
>>
>> On 13 Dec 2011, at 20:53, Alan Ruttenberg <alanrut...@gmail.com> wrote:
>>
>>> On Tue, Dec 13, 2011 at 3:07 PM, Chris Mungall <cjmu...@lbl.gov> wrote:
>>>>
>>>> Are these definitions being coordinated with CARO?
>>>>
>>>> I don't remember the exact reasons we don't have anatomical set in CARO.
>>>> In the original CARO, which was based heavily on the FMA, we had:
>>>>

>>>> ￜ is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity
>>>> that has mass."]
>>>> ￜ ￜis_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical

>>>> entity that has inherent 3D shape and is generated by coordinated expression
>>>> of the organism's own genome."]

>>>> ￜ ￜ is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure

>>>> consisting of at least two non-overlapping organs, multi-tissue aggregates
>>>> or portion of tissues or cells of different types that does not constitute
>>>> an organism, organ, multi-tissue aggregate, or portion of tissue."]
>>>
>>> I'm not sure what case is covered by cells of different types as
>>> opposed to tissues. Is there a collection of cells of different types
>>> that is not an organ, multi-tissue aggregate or portion of tissue, and
>>> yet still has "inherent 3D shape and is generated by coordinated
>>> expression of the organism's own genome"?
>>>

>>>> ￜ ￜ ￜis_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is

>>>> has as its parts distinct anatomical structures interconnected by anatomical
>>>> structures at a lower level of granularity."]

>>>> ￜ ￜ ￜis_a CARO:0000041 ! anatomical cluster ￜ[DEF: "Anatomical group that

[mej...@u.washington.edu]

Hi David,

My response below.

Onard

On Wed, 14 Dec 2011, David Osumi-Sutherland wrote:

> Hi Onard,
> Few comments in line.

> ￜￜMaxillary dentition
> ￜￜￜￜￜLeft upper third secondary molar tooth
> ￜￜￜￜￜRight upper third secondary molar tooth
> ￜￜￜￜￜetc.,
> ￜￜMandibular dentition
> ￜￜￜￜￜLeft lower third secondary molar tooth
> ￜￜￜￜￜRight lower third secondary molar tooth
> ￜￜￜￜￜetc.

>
>
>
> Bur if you do have both, you need some way to infer the second from the first or the two will go out a sync.
>
>
> 'anatomical group' is needed for cases where an anatomical entity is made up of disconnected parts of various classes. ï¿œThe immune system is a good example.
>
>
> Exactly, the immune system is an anatomical set, with discontinuous parts.
>
>

> FMA has 'Anatomical Set':

>
> "Material anatomical entity which consists of the maximum number of discontinuous members of the same class."
>

> ï¿œIn what sense are the disconnect parts of the immune system members of the same class ?
>
> The point of 'anatomical group' in CARO was, I believe, to encompass entities that had disconnected parts of various classes.

Very good point. I knew this was coming. A while ago I have considered two kinds of sets, one with members of the same class (homogenous anatomical set) and the other, with various classes (heterogeneous anatomical set). Cornelius and I have not had the chance to revisit this but probably now is a good time. Of course suggestions are always welcome.

But do you agree that there is a difference between groups that are connected and those that are discontinuous?

>
>
>
>
> In some cases the parts are in close proximity. In those cases 'anatomical cluster' seems appropriate to me.
>
>
> Proximity but not connected is also an anatomical set.
>
> Anatomical cluster has parts connected, such as the joint.
>
>

> OK. ï¿œThe FMA def seems reasonably clear on this. ï¿œStill, it is useful to have a term for anatomical groups whose members are in close proximity. ï¿œFor example, many of the sense organs of the Drosophila larva
> are arranged in clusters - and these clusters are referred to in anatomical discourse. ï¿œCan you suggest any alternative to 'cluster' that we could use for this?
>
> Cheers,
>
> David

> David Osumi-Sutherland, PhD
> Ontologist
> FlyBase / Virtual Fly Brain
> Department of Genetics,
> University of Cambridge,
> Downing Street,
> Cambridge, CB2 3EH, UK
> Tel: +44 (0)1223 333 963
> Fax: +44 (0)1223 766 732
> http://www.virtualflybrain.org
>
>
>

[Chris Mungall]

I just committed a new version of CARO - the definitions that were derived in part or in whole from FMA should have the definition source field now filled in.

On Dec 13, 2011, at 5:17 PM, ro...@u.washington.edu wrote:

> I feel that if CARO adopts FMA definitions, those definitions in CARO should indicate that they are derived from the FMA, as is the case with SNOMED for example.
>
> Cornelius
>
> On Tue, 13 Dec 2011, Chris Mungall wrote:
>
>> Are these definitions being coordinated with CARO?
>> I don't remember the exact reasons we don't have anatomical set in CARO.
>> In the original CARO, which was based heavily on the FMA, we had:

>> is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity that has mass."]

>> is_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical entity that has inherent 3D shape and is generated by
>> coordinated expression of the organism's own genome."]

>> is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure consisting of at least two non-overlapping organs,
>> multi-tissue aggregates or portion of tissues or cells of different types that does not constitute an organism, organ,
>> multi-tissue aggregate, or portion of tissue."]

>> is_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is has as its parts distinct anatomical structures
>> interconnected by anatomical structures at a lower level of granularity."]

>> is_a CARO:0000041 ! anatomical cluster [DEF: "Anatomical group that has its parts adjacent to one another."]

[mej...@u.washington.edu]

On Wed, 14 Dec 2011, David Osumi-Sutherland wrote:

>
> ￜￜￜￜￜExactly, the immune system is an anatomical set, with discontinuous parts.
>
> FMA has 'Anatomical Set':
>

> "Material anatomical entity which consists of the maximum number of discontinuous members of the same class."
>

> ï¿œIn what sense are the disconnect parts of the immune system members of the same class ?
>
> The point of 'anatomical group' in CARO was, I believe, to encompass entities that had disconnected parts of various classes.
>
>
>
> Very good point. I knew this was coming. A while ago I have considered two kinds of sets, one with members of the same class (homogenous anatomical set) and the other, with various classes
> (heterogeneous anatomical set). Cornelius and I have not had the chance to revisit this but probably now is a good time. Of course suggestions are always welcome.
>
> But do you agree that there is a difference between groups that are connected and those that are discontinuous?
>
>

> By connected group, I assume you mean an anatomical structure that is made up of very heterogeneous parts? ï¿œA joint seems like a good example of this. ï¿œMy worry is how heterogenous do the parts have to be to
> count. ï¿œThis seems to leave a lot of room for grey areas. ï¿œWith the term 'anatomical cluster' the FMA does a pretty good job of specifying what types of heterogeneity count:
>
> "Anatomical structure, which has as its parts a heterogeneous collection of organs, organ parts, cells, cell parts or body part subdivisions that are adjacent to, or continuous with one another; does not
> constitute a cell part, cell, tissue, organ, organ system or organ system subdivision, cardinal body part, body part subdivision or anatomical junction."

First of all, the definition has been updated: "Anatomical structure, which has as its parts a heterogeneous collection of organs, organ parts, cells, cell parts or body part subdivisions that are connected to one another; does not constitute a cell part, cell, tissue, organ, organ system or organ system subdivision, cardinal body part, body part subdivision."
There are two changes implemented in the definition:
1. the parts are connected (by attachment or by continuity). Adjacency does not necessarily mean the adjacent parts are connected.
2. anatomical junction is a type of anatomical cluster.

>
> The challenge for CARO is that this references many FMA classes and so is not really suitable unless we add these classes, or some more general equivalent, to CARO. ï¿œJudging by the arguments we had over CARO1,
> many of these are hard to generalise for multiple species and so hard to add to CARO.ￜ

If the anatomical structure consists of parts, however heterogeneous, that are connected to one another but does not fit or satisfy the definitions you have for the major organizational classes such as cell, organ part, organ, organ system, cardinal body part, etc., then it would probably fall in the "cluster" category. I think other people have used the term "aggregate".

>
>
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜIn some cases the parts are in close proximity. In those cases 'anatomical cluster' seems appropriate to me.
>
> ￜￜￜￜￜProximity but not connected is also an anatomical set.
>
> ￜￜￜￜￜAnatomical cluster has parts connected, such as the joint.

>
> OK. ï¿œThe FMA def seems reasonably clear on this. ï¿œStill, it is useful to have a term for anatomical groups whose members are in close proximity. ï¿œFor example, many of the sense organs of
> the Drosophila larva
>
> are arranged in clusters - and these clusters are referred to in anatomical discourse. ï¿œCan you suggest any alternative to 'cluster' that we could use for this?
>
>

> Still interested in any suggestions for naming ï¿œanatomical entities that are disconnected groups defined by their close proximity (assuming anatomical cluster is ruled out).
>
> Cheers again,
>
> David
> Cheers,
>
> David
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜCARO1 didn't do a good job of making these distinctions.
>
> ￜￜￜￜￜWe elaborated on the distinction in the FMA but I'm still not clear what the objection is. I would certainly entertain a better representation that captures the distinctions.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜNote, I have never been entirely comfortable with 'anatomical cluster' as a classification for joint as, AFAIK, the various parts of joints are connected to each other.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOn 13 Dec 2011, at 20:53, Alan Ruttenberg <alanrut...@gmail.com> wrote:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOn Tue, Dec 13, 2011 at 3:07 PM, Chris Mungall <cjmu...@lbl.gov> wrote:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜAre these definitions being coordinated with CARO?
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜI don't remember the exact reasons we don't have anatomical set in CARO.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜIn the original CARO, which was based heavily on the FMA, we had:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜthat has mass."]
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ ￜis_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜentity that has inherent 3D shape and is generated by coordinated expression
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜof the organism's own genome."]
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ ￜ is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜconsisting of at least two non-overlapping organs, multi-tissue aggregates
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜor portion of tissues or cells of different types that does not constitute
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜan organism, organ, multi-tissue aggregate, or portion of tissue."]
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜI'm not sure what case is covered by cells of different types as
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜopposed to tissues. Is there a collection of cells of different types
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜthat is not an organ, multi-tissue aggregate or portion of tissue, and
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜyet still has "inherent 3D shape and is generated by coordinated
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜexpression of the organism's own genome"?
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ ￜ ￜis_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜhas as its parts distinct anatomical structures interconnected by anatomical
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜstructures at a lower level of granularity."]
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ ￜ ￜis_a CARO:0000041 ! anatomical cluster ￜ[DEF: "Anatomical group that
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜhas its parts adjacent to one another."]
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜDentition would seem to fit 'anatomical cluster' - unless we consider a
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜchild who has a mixed dentition, consisting of a mixture of primary
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜdentition, secondary dentition, and sets of unerupted teeth (maybe this
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜisn't "canonical")
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜTeeth aren't necessarily adjacent (at least in the common sense),
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜthough I see the RO sense is simply nearby "in spatial proximity".
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜSince the mouth can open quite wide it's not clear they would be
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜconsidered adjacent in even that sense.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜFMA has both Anatomical Set as well as Anatomical Structure, BTW.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜIi haven't been coordinating with CARO but will make an effort to do
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜso in the subset of terms I extract to represent teeth.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ-Alan
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOn Dec 13, 2011, at 10:38 AM, Todd Detwiler wrote:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜAlan, where are you getting this definition of Anatomical set? I have looked
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜin the live FMA, the last release version of the FMA (v3.2) and the version
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜbefore that (from early 2010, v3.1). All have the following definition for
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜAnatomical set:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ"Material anatomical entity which consists of the maximum number of
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜdiscontinuous members of the same class. Examples: set of cranial nerves,
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜventral branches of aorta, set of mammary arteries, thoracic viscera, dental
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜarcade."
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜIt does not say that the members of the set must be from the class
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜAnatomical structure.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜTodd
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜLandon Todd Detwiler
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜStructural Informatics Group (SIG)
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜUniversity of Washington
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOn 12/13/11 10:15 AM, Alan Ruttenberg wrote:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOn Fri, Dec 9, 2011 at 2:41 PM, ￜ<mej...@comcast.net> wrote:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜRight. My point is that the definitions disallows members of anatomical sets
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜbeing other anatomical sets(many of the cases listed) or things that are not
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜanatomical structures, such as blood:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ'Hematopoietic system'(FMA_9667) has member 'Portion of blood'(FMA_9670)
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜwhich is not anatomical structure
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜOM: The definition of anatomical set does not disallow body substances, that
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜwhy we have the class "Hematopoietic system".
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜWe also have sets of immaterial anatomical entities, such as "Set of
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜsubarachnoid sulci" and "Set of external nares".
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜThe definition of Anatomical set as provided by the FMA is:
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ"material anatomical entity which has as its members the maximal
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜnumber of anatomical structures of the same type."
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜAs I read the current FMA, portion of body substance, being a sibling
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜof Anatomical set, is disjoint from it. So the definition contradicts
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜthe assertion. I am advocating that one of them be changed so that
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜthere is no longer a contradiction.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜI would also suggest that the definition be changed to not say
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜanything about the "maximal number", as this raises issues regarding
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜtime. Instead, push any kind of such maximality down to the particular
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜterms. For example, the term "dentition" could be defined as being the
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜset of all the teeth that are present at any time, if desired.
>
> ￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜￜ-Alan

>
> David Osumi-Sutherland, PhD
>
> Ontologist
>
> FlyBase / Virtual Fly Brain
>
> Department of Genetics,
>
> University of Cambridge,
>
> Downing Street,
>
> Cambridge, CB2 3EH, UK
>
> Tel: +44 (0)1223 333 963
>
> Fax: +44 (0)1223 766 732
>
> http://www.virtualflybrain.org
>
>
>

> David Osumi-Sutherland, PhD
> Ontologist
> Virtual Fly Brain / FlyBase
> Department of Genetics
> University of Cambridge
> Downing Street
> Cambridge, CB2 3EH,ï¿œUK
> +44 (0)1223 333 963
> http://www.virtualflybrain.org
>
>
>
>
>
>

[mej...@u.washington.edu]

Chris,

That's great, thanks!

Onard

[ro...@u.washington.edu]

Chris, I am also very pleased that you ahve made this addition. Many thanks.

Cornelius

[mej...@u.washington.edu]

Chris,

We most certainly agree with you that it's high priority to move the FMA to OWL (2). Our group is looking into this right now but there are no immediate solutions...yet.

Onard

[mej...@u.washington.edu]

On Fri, 9 Dec 2011, Alan Ruttenberg wrote:

> Onard wrote:
> Temporal property is of great interest to me and one I would like to explore for the FMA.
>
>

> For starters I think it would help to simply clarify the status of Anatomical set as it is used in the FMA and augment the definition with the resulting clarification.
>
> Is an anatomical set instantiated at any time any of its members develops, perhaps having changing membership over time, until none are present? Or is it the intention that the term denote the full set of
> entities only during the time that they are all present. These are different from an ontological point of view, though in both cases continuant.

You raised a very interesting point. By canonical dentition, I meant a class that subsumes 1) a set of primary teeth which is complete over a specified period of time (primary dentition), 2) a set of secondary teeth which is complete over a specified period of time (secondary dentition) and 3) a set of primary and secondary teeth over the entire life (Entire dentition). And probably a fourth kind where, in various periods of time, varying mixture of primary and secondary teeth exist (mixed dentition). Maybe analogous to this is the class "Portion of blood". It subsumes both "portion of blood in right ventricle" and "entire portion of blood". The constituents of all types of "portion of blood" also change over time (faster, of course) and so when we say "portion of blood" do we refer only to a portion present during childhood or during the individual's entire life?

>
> -Alan
>
>

[Alan Ruttenberg]

On Wed, Dec 14, 2011 at 2:40 PM, mej...@u.washington.edu
<mej...@u.washington.edu> wrote:
>
>
>
> On Fri, 9 Dec 2011, Alan Ruttenberg wrote:
>
>> Onard wrote:
>>      Temporal property is of great interest to me and one I would like to
>> explore for the FMA.
>>
>>
>> For starters I think it would help to simply clarify the status of
>> Anatomical set as it is used in the FMA and augment the definition with the
>> resulting clarification.
>>
>> Is an anatomical set instantiated at any time any of its members develops,
>> perhaps having changing membership over time, until none are present? Or is
>> it the intention that the term denote the full set of
>> entities only during the time that they are all present. These are
>> different from an ontological point of view, though in both cases
>> continuant.
>
>
> You raised a very interesting point. By canonical dentition, I meant a class
> that subsumes 1) a set of primary teeth which is complete over a specified
> period of time (primary dentition),

This I like.

2) a set of secondary teeth which is
> complete over a specified period of time (secondary dentition) and

This I like.

3) a set
> of primary and secondary teeth over the entire life (Entire dentition). And
> probably a fourth kind where, in various periods of time, varying mixture of
> primary and secondary teeth exist (mixed dentition).

This I don't like. I think that statements would be better made in
terms of the type tooth than this set. My understanding was that
Cornelius also thought it a reasonable principle that we don't make
*sets* of things like this.
However if you have any statements that you think appropriate to make
about Entire dentition that are not made well using only the type
tooth, I'd be interested in having a look.

Otherwise I propose we just decide to drop "Entire dentition" if it
will mean that set.
If it would have the meaning "at any given time having members all the
teeth in a mouth" that might be a useful class.

> Maybe analogous to this
> is the class "Portion of blood". It subsumes both "portion of blood in right
> ventricle" and "entire portion of blood". The constituents of  all types of
> "portion of blood" also change over time (faster, of course) and so when we
> say "portion of blood" do we refer only to a portion present during
> childhood or during the individual's entire life?

Here we are not talking about a set, but rather a continuant and its
parts. If we were to talk about something set-like, it would be a kind
of aggregate that acts similar to our bodies losing and gaining parts
in time. I don't think it is analogous to teeth.

-Alan

[mej...@u.washington.edu]

Jose Leonardo V. Mejino Jr., M.D.
Senior Scientist and Project Director,
Foundational Model of Anatomy Ontology (FMA)
Structural Informatics Group
Department of Biological Structure
University of Washington School of Medicine
Seattle, WA 98195
(206)543-7118 FAX:(206)543-1524

On Wed, 14 Dec 2011, Alan Ruttenberg wrote:

> On Wed, Dec 14, 2011 at 2:40 PM, mej...@u.washington.edu
> <mej...@u.washington.edu> wrote:
>>
>>
>>
>> On Fri, 9 Dec 2011, Alan Ruttenberg wrote:
>>
>>> Onard wrote:

>>> ￜ ￜ ￜTemporal property is of great interest to me and one I would like to

So if I want to record patient's dental history, I need the patient's entire dentition record and I build that by going to the FMA, get the class "Entire dentition" and with it the list of all individual teeth, both primary and secondary. If I go by class "Tooth", I have to filter classes I don't need, e.g. Incisor tooth, primary incisor tooth, upper primary incisor tooth,
and upper central primary incisor tooth before I get to "Right upper central primary incisor tooth" which I need. Then repeat the process for lower primary incisor teeth and then again for the secondary versions.

>
> Otherwise I propose we just decide to drop "Entire dentition" if it
> will mean that set.
> If it would have the meaning "at any given time having members all the
> teeth in a mouth" that might be a useful class.
>
>> Maybe analogous to this
>> is the class "Portion of blood". It subsumes both "portion of blood in right

>> ventricle" and "entire portion of blood". The constituents of ï¿œall types of

>> "portion of blood" also change over time (faster, of course) and so when we
>> say "portion of blood" do we refer only to a portion present during
>> childhood or during the individual's entire life?
>
> Here we are not talking about a set, but rather a continuant and its
> parts. If we were to talk about something set-like, it would be a kind
> of aggregate that acts similar to our bodies losing and gaining parts
> in time. I don't think it is analogous to teeth.

Sorry I was not clear on this one. I didn't mean that portion of blood is a set. What I wanted to say is that a class subsumes both portions (parts) and the entire whole. Portion of substance may be a smaller portion of a whole or the entire portion. A set may mean a subset or a complete set.

Then there's the second issue, inclusion of the temporal property. This is probably a separate issue best left for later, for Barry and Cornelius.

Onard

>
> -Alan
>

[mej...@u.washington.edu]

On Tue, 13 Dec 2011, Chris Mungall wrote:

>
> The naming in the FMA is a little confusing:
>

> ￜ/ FMA:62955 ! Anatomical entity
> ￜ is_a FMA:61775 ! Physical anatomical entity
> ￜ ￜis_a FMA:67112 ! Immaterial anatomical entity
> ￜ ￜ is_a FMA:71917 ! Set of immaterial anatomical entities ***
> ￜ ￜ ￜis_a FMA:71918 ! Set of anatomical spaces
> ￜ ￜ ￜ is_a FMA:76576 ! Set of cavities
> ￜ ￜis_a FMA:67165 ! Material anatomical entity
> ￜ ￜ is_a FMA:55652 ! Anatomical set ***ￜ
> ￜ

> "anatomical set" should be renamed "set of material anatomical entities".

Yes this makes sense. For now I will make "set of material anatomical entities" a synonym. Once we have the capability for inferencing, then anatomical set will be a separate class, encompassing material and immaterial entities.

>
> You could introduce "anatomical set" as a different class that is a set of material or immaterial entities, but this would go against the rigid SI principle in the FMA.
>
> On Dec 13, 2011, at 12:07 PM, Chris Mungall wrote:
>
>
> Are these definitions being coordinated with CARO?
>
> I don't remember the exact reasons we don't have anatomical set in CARO.
>
> In the original CARO, which was based heavily on the FMA, we had:
>

> ￜ is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity that has mass."]
> ￜ ￜis_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical entity that has inherent 3D shape and is generated by coordinated expression of the organism's own genome."]
> ￜ ￜ is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure consisting of at least two non-overlapping organs, multi-tissue aggregates or portion of tissues or cells of different types that does

> not constitute an organism, organ, multi-tissue aggregate, or portion of tissue."]

> ￜ ￜ ￜis_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is has as its parts distinct anatomical structures interconnected by anatomical structures at a lower level of granularity."]
> ￜ ￜ ￜis_a CARO:0000041 ! anatomical cluster ￜ[DEF: "Anatomical group that has its parts adjacent to one another."]

[mej...@u.washington.edu]

On Tue, 13 Dec 2011, Alan Ruttenberg wrote:

> On Tue, Dec 13, 2011 at 3:07 PM, Chris Mungall <cjmu...@lbl.gov> wrote:
>>
>> Are these definitions being coordinated with CARO?
>>
>> I don't remember the exact reasons we don't have anatomical set in CARO.
>> In the original CARO, which was based heavily on the FMA, we had:
>>

>> ￜ is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity
>> that has mass."]
>> ￜ ￜis_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical

>> entity that has inherent 3D shape and is generated by coordinated expression
>> of the organism's own genome."]

>> ￜ ￜ is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure

>> consisting of at least two non-overlapping organs, multi-tissue aggregates
>> or portion of tissues or cells of different types that does not constitute
>> an organism, organ, multi-tissue aggregate, or portion of tissue."]
>
> I'm not sure what case is covered by cells of different types as
> opposed to tissues. Is there a collection of cells of different types
> that is not an organ, multi-tissue aggregate or portion of tissue, and
> yet still has "inherent 3D shape and is generated by coordinated
> expression of the organism's own genome"?

Yes, a motor unit which consists of a motor neuron and a skeletal muscle cell (Cell cluster). Another is the "juxtaglomerular complex" (juxtaglomerular apparatus).

>
>> ￜ ￜ ￜis_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is

>> has as its parts distinct anatomical structures interconnected by anatomical
>> structures at a lower level of granularity."]

>> ￜ ￜ ￜis_a CARO:0000041 ! anatomical cluster ￜ[DEF: "Anatomical group that

[Alan Ruttenberg]

On Wed, Dec 14, 2011 at 5:54 PM, mej...@u.washington.edu <mej...@u.washington.edu> wrote:

I'm not sure what case is covered by cells of different types as
opposed to tissues. Is there a collection of cells of different types
that is not an organ, multi-tissue aggregate or portion of tissue, and
yet still has "inherent 3D shape and is generated by coordinated
expression of the organism's own genome"?

Yes, a motor unit which consists of a motor neuron and a skeletal muscle cell (Cell cluster). Another is the "juxtaglomerular complex" (juxtaglomerular apparatus).

Excellent! Thanks for the examples :)

I think it would be great if they could added to the editor notes or examples of usage.

-Alan

 

[Alan Ruttenberg]

You just ask for the leaf subclasses of tooth. Below I'll do that for the mouth module I shared, and then inside the native FMA. Each gets a different answer, exposing two bugs that should be fixed.

Here's the version on the OWL module:

(setq mouth (load-ontology "http://svn.neurocommons.org/svn/trunk/convert/fma/owlfiles/mouth-hierarchy.owl"))

(loop for tooth-type in (descendants !obo:FMA_12516 mouth) ;; here is the DL query for all subclasses

   when

     (null (children tooth-type mouth))

   do

     (format t "~a (~a)~%" (rdfs-label tooth-type mouth) tooth-type)

            and  sum 1)

(Left upper central primary incisor tooth) (!obo:FMA_84221)

(Right upper central primary incisor tooth) (!obo:FMA_84220)

(Left upper lateral primary incisor tooth) (!obo:FMA_84223)

(Right upper lateral primary incisor tooth) (!obo:FMA_84222)

(Left lower central primary incisor tooth) (!obo:FMA_84225)

(Right lower central primary incisor tooth) (!obo:FMA_84224)

(Left lower lateral primary incisor tooth) (!obo:FMA_84227)

(Right lower lateral primary incisor tooth) (!obo:FMA_84226)

(Left upper central secondary incisor tooth) (!obo:FMA_55682)

(Left upper lateral secondary incisor tooth) (!obo:FMA_55683)

(Right upper lateral secondary incisor tooth) (!obo:FMA_55680)

(Right upper central secondary incisor tooth) (!obo:FMA_55681)

(Left lower secondary canine tooth) (!obo:FMA_55687)

(Right lower secondary canine tooth) (!obo:FMA_55686)

(Right upper first secondary premolar tooth) (!obo:FMA_55689)

(Right upper second secondary premolar tooth) (!obo:FMA_55688)

(Right lower second secondary premolar tooth) (!obo:FMA_55695)

(Right upper third secondary molar tooth) (!obo:FMA_55696)

(Left lower first secondary premolar tooth) (!obo:FMA_55693)

(Right lower first secondary premolar tooth) (!obo:FMA_55694)

(Left upper second secondary premolar tooth) (!obo:FMA_55691)

(Left lower second secondary premolar tooth) (!obo:FMA_55692)

(Left upper first secondary premolar tooth) (!obo:FMA_55690)

(Left upper first secondary molar tooth) (!obo:FMA_55699)

(Right upper first secondary molar tooth) (!obo:FMA_55698)

(Right upper second secondary molar tooth) (!obo:FMA_55697)

(Right lower lateral secondary incisor tooth) (!obo:FMA_57140)

(Right lower central secondary incisor tooth) (!obo:FMA_57142)

(Left lower lateral secondary incisor tooth) (!obo:FMA_57141)

(Left lower central secondary incisor tooth) (!obo:FMA_57143)

(Right lower first primary molar tooth) (!obo:FMA_84242)

(Left lower first primary molar tooth) (!obo:FMA_84243)

(Right upper secondary canine tooth) (!obo:FMA_55798)

(Left upper secondary canine tooth) (!obo:FMA_55799)

(Left upper first primary molar tooth) (!obo:FMA_84239)

(Right upper first primary molar tooth) (!obo:FMA_84238)

(Left lower primary canine tooth) (!obo:FMA_84233)

(Left upper primary canine tooth) (!obo:FMA_84231)

(Right lower primary canine tooth) (!obo:FMA_84232)

(Right upper primary canine tooth) (!obo:FMA_84230)

(Right lower first secondary molar tooth) (!obo:FMA_55705)

(Right lower second secondary molar tooth) (!obo:FMA_55706)

(Right lower third secondary molar tooth) (!obo:FMA_55707)

(Left upper third secondary molar tooth) (!obo:FMA_55701)

(Left lower third secondary molar tooth) (!obo:FMA_55702)

(Left lower second secondary molar tooth) (!obo:FMA_55703)

(Left lower first secondary molar tooth) (!obo:FMA_55704)

(Left upper second secondary molar tooth) (!obo:FMA_55700)

48

Note we are missing 4 teeth, the second primary molars, as they are missing as members of primary dentition and my module heuristic is to follow the partonomy downward from a selected term, and because the teeth are only asserted to be part of the mouth by virtue of dentition being part of the mouth. fix: Add second primary molars to primary dentition.

Now the same thing on the FMA native

(let ((count 0))

  (map-fma-children fma::|Tooth|

      (lambda(e) (when (null (fma-children e))

   (incf count)

   (format t "~a(~a)~%" (string e) (fma-uri e)))))

  count)

Left upper second primary molar tooth(!obo:FMA_84241)

Right upper second primary molar tooth(!obo:FMA_84240)

Left upper first primary molar tooth(!obo:FMA_84239)

Right upper first primary molar tooth(!obo:FMA_84238)

Right lower second primary molar tooth(!obo:FMA_84244)

Left lower second primary molar tooth(!obo:FMA_84245)

Right lower first primary molar tooth(!obo:FMA_84242)

Left lower first primary molar tooth(!obo:FMA_84243)

Right upper second secondary molar tooth(!obo:FMA_55697)

Left upper second secondary molar tooth(!obo:FMA_55700)

Left upper first secondary molar tooth(!obo:FMA_55699)

Right upper first secondary molar tooth(!obo:FMA_55698)

Left upper third secondary molar tooth(!obo:FMA_55701)

Right upper third secondary molar tooth(!obo:FMA_55696)

Right lower third secondary molar tooth(!obo:FMA_55707)

Left lower third secondary molar tooth(!obo:FMA_55702)

Left lower first secondary molar tooth(!obo:FMA_55704)

Right lower first secondary molar tooth(!obo:FMA_55705)

Left lower second secondary molar tooth(!obo:FMA_55703)

Right lower second secondary molar tooth(!obo:FMA_55706)

Left upper first secondary premolar tooth(!obo:FMA_55690)

Right upper first secondary premolar tooth(!obo:FMA_55689)

Left upper second secondary premolar tooth(!obo:FMA_55691)

Right upper second secondary premolar tooth(!obo:FMA_55688)

Right lower second secondary premolar tooth(!obo:FMA_55695)

Left lower second secondary premolar tooth(!obo:FMA_55692)

Left lower first secondary premolar tooth(!obo:FMA_55693)

Right lower first secondary premolar tooth(!obo:FMA_55694)

Right upper secondary canine tooth(!obo:FMA_55798)

Left upper secondary canine tooth(!obo:FMA_55799)

Left lower secondary canine tooth(!obo:FMA_55687)

Right lower secondary canine tooth(!obo:FMA_55686)

Right lower primary canine tooth(!obo:FMA_84232)

Left lower primary canine tooth(!obo:FMA_84233)

Left upper primary canine tooth(!obo:FMA_84231)

Right upper primary canine tooth(!obo:FMA_84230)

Left lower lateral secondary incisor tooth(!obo:FMA_57141)

Right lower lateral secondary incisor tooth(!obo:FMA_57140)

Left lower central secondary incisor tooth(!obo:FMA_57143)

Right lower central secondary incisor tooth(!obo:FMA_57142)

Left upper lateral secondary incisor tooth(!obo:FMA_55683)

Right upper lateral secondary incisor tooth(!obo:FMA_55680)

Left upper central secondary incisor tooth(!obo:FMA_55682)

Right upper central secondary incisor tooth(!obo:FMA_55681)

Right upper central primary incisor tooth(!obo:FMA_84220)

Left upper central primary incisor tooth(!obo:FMA_84221)

Left upper lateral primary incisor tooth(!obo:FMA_84223)

Right upper lateral primary incisor tooth(!obo:FMA_84222)

Left lower lateral primary incisor tooth(!obo:FMA_84227)

Right lower lateral primary incisor tooth(!obo:FMA_84226)

Left lower central primary incisor tooth(!obo:FMA_84225)

Right lower central primary incisor tooth(!obo:FMA_84224)

Primary tooth(!obo:FMA_55655)

Secondary tooth(!obo:FMA_55654)

54

Here we have a couple of extra due to the classes Primary tooth and  Secondary tooth not being superclass of the appropriate tooth classes. Since they have no children they are also caught by this query. Fix: Populate those.

 

I have noted the problems with the current dentition terms having members which are sets a well as teeth. Suffice to say, OWL queries will not return what you want for the teeth, at the moment.

Bottom line: You do not need the set classes in order to easily retrieve the specific tooth classes you want.

BTW, in another email you asked for another example where the member/subclass confusion occurred. Here is a query I used to fish for them

(map-fma 

 (lambda (e) 

   (when (and (some 

(lambda(m) 

 (find fma::|Anatomical set| (fma-superclasses m)))

(slot-values e fma::|member|))

      (fma-children e)) (print e))))

Which means: Find me sets that have sets as members and also subclasses.

Musculature of trunk 

Set of joints of foot 

Set of basal ganglia 

Set of all viscera 

Dentition 

Musculature of perineum 

Set of arteries of foot 

Musculature of trunk proper 

Skeleton of free lower limb 

Set of viscera of abdomen 

Set of phalanges of foot 

Skeleton of foot 

Skeleton of upper limb

If you look at Skeleton of foot  you see, for example, that it has member Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply that the right foot has parts all of the tarsal bones of both feet.

-Alan

<I'll take the below up in another message, but essentially: there's no time like the present (heh, dual use phrase :) ). Hopefully Barry and Cornelius will chime in, but in their absence I think there's enough grownups around for us to make progress>

Otherwise I propose we just decide to drop "Entire dentition" if it
will mean that set.
If it would have the meaning "at any given time having members all the
teeth in a mouth" that might be a useful class.

Maybe analogous to this
is the class "Portion of blood". It subsumes both "portion of blood in right

ventricle" and "entire portion of blood". The constituents of  all types of

"portion of blood" also change over time (faster, of course) and so when we
say "portion of blood" do we refer only to a portion present during
childhood or during the individual's entire life?

Here we are not talking about a set, but rather a continuant and its
parts. If we were to talk about something set-like, it would be a kind
of aggregate that acts similar to our bodies losing and gaining parts
in time. I don't think it is analogous to teeth.

[Alan Ruttenberg]

On Tue, Dec 13, 2011 at 6:33 PM, Alan Ruttenberg
<alanrut...@gmail.com> wrote:

> OK, I tracked this down.
>
> The definition is as you say. The "documentation" is:
>
>           "OM def (1.2.2007): material anatomical entity which has as
> its members the maximal number of anatomical structures of the same
> type.nn10.04.01: While some of anatomical sets could be regarded as an
> organ system subvision, e.g set of lumbar nerves, others do not
> satisfy one or two defining attributes of organ system subdivision.
> E.g. thoracic viscera consists of organs but not belonging to one
> organ system, dental arcade consists of similar organs but they are
> not connected.")
>
> So what is the status of the documentation slot value?

Onard wrote back privately: "The comment (documnetation) field is for
internal use only and no longer distributed or included in release
versions. These are notes to the authors, they give us some historical
information or in some cases, provenance."

So I would put these in the editor notes. They are too interesting to
be curator notes.

-Alan

[Melissa Haendel]

hi,

few comments below,

On Dec 14, 2011, at 12:39 PM, mej...@u.washington.edu wrote:

On Wed, 14 Dec 2011, David Osumi-Sutherland wrote:

                Exactly, the immune system is an anatomical set, with discontinuous parts.

           FMA has 'Anatomical Set':

           "Material anatomical entity which consists of the maximum number of discontinuous members of the same class."

            In what sense are the disconnect parts of the immune system members of the same class ?

           The point of 'anatomical group' in CARO was, I believe, to encompass entities that had disconnected parts of various classes.

yes exactly

     Very good point. I knew this was coming. A while ago I have considered two kinds of sets, one with members of the same class (homogenous anatomical set) and the other, with various classes

     (heterogeneous anatomical set). Cornelius and I have not had the chance to revisit this but probably now is a good time. Of course suggestions are always welcome.

     But do you agree that there is a difference between groups that are connected and those that are discontinuous?

yes.

By connected group, I assume you mean an anatomical structure that is made up of very heterogeneous parts?  A joint seems like a good example of this.  My worry is how heterogenous do the parts have to be to

count.  This seems to leave a lot of room for grey areas.  With the term 'anatomical cluster' the FMA does a pretty good job of specifying what types of heterogeneity count:

"Anatomical structure, which has as its parts a heterogeneous collection of organs, organ parts, cells, cell parts or body part subdivisions that are adjacent to, or continuous with one another; does not

constitute a cell part, cell, tissue, organ, organ system or organ system subdivision, cardinal body part, body part subdivision or anatomical junction."

First of all, the definition has been updated: "Anatomical structure, which has as its parts a heterogeneous collection of  organs, organ parts, cells, cell parts or body part subdivisions that are connected to one another; does not constitute a cell part, cell,  tissue,  organ, organ system or organ system subdivision, cardinal body part, body part subdivision."
There are two changes implemented in the definition:
1. the parts are connected (by attachment or by continuity). Adjacency does not necessarily mean the adjacent parts are connected.
2. anatomical junction is a type of anatomical cluster.

The challenge for CARO is that this references many FMA classes and so is not really suitable unless we add these classes, or some more general equivalent, to CARO.  Judging by the arguments we had over CARO1,

many of these are hard to generalise for multiple species and so hard to add to CARO.

I think we *may* be able to do this in the new caro without too much trouble, because really its a grab bag of everything else. In any case, I am determined not to let perfect be the evil of good here and go through what we went through with caro1 again. I think you all will agree. ;-) 

If the anatomical structure consists of parts, however heterogeneous, that are connected to one another but does not fit or satisfy the definitions you have for the major organizational classes such as cell, organ part, organ, organ system, cardinal body part, etc., then it would probably fall in the "cluster" category. I think other people have used the term "aggregate".

                            In some cases the parts are in close proximity. In those cases 'anatomical cluster' seems appropriate to me.

                Proximity but not connected is also an anatomical set.

                Anatomical cluster has parts connected, such as the joint.

           OK.  The FMA def seems reasonably clear on this.  Still, it is useful to have a term for anatomical groups whose members are in close proximity.  For example, many of the sense organs of

           the Drosophila larva

           are arranged in clusters - and these clusters are referred to in anatomical discourse.  Can you suggest any alternative to 'cluster' that we could use for this?

Still interested in any suggestions for naming  anatomical entities that are disconnected groups defined by their close proximity (assuming anatomical cluster is ruled out).

Cheers again,

David

           Cheers,

           David

                            CARO1 didn't do a good job of making these distinctions.

                We elaborated on the distinction in the FMA but I'm still not clear what the objection is. I would certainly entertain a better representation that captures the distinctions.

                            Note, I have never been entirely comfortable with 'anatomical cluster' as a classification for joint as, AFAIK, the various parts of joints are connected to each other.

                            On 13 Dec 2011, at 20:53, Alan Ruttenberg <alanrut...@gmail.com> wrote:

                                  On Tue, Dec 13, 2011 at 3:07 PM, Chris Mungall <cjmu...@lbl.gov> wrote:

                                        Are these definitions being coordinated with CARO?

                                        I don't remember the exact reasons we don't have anatomical set in CARO.

                                        In the original CARO, which was based heavily on the FMA, we had:

                                          is_a CARO:0000006 ! material anatomical entity [DEF: "Anatomical entity

                                        that has mass."]

                                           is_a CARO:0000003 ! anatomical structure [DEF: "Material anatomical

                                        entity that has inherent 3D shape and is generated by coordinated expression

                                        of the organism's own genome."]

                                            is_a CARO:0000054 ! anatomical group [DEF: "Anatomical structure

                                        consisting of at least two non-overlapping organs, multi-tissue aggregates

                                        or portion of tissues or cells of different types that does not constitute

                                        an organism, organ, multi-tissue aggregate, or portion of tissue."]

                                  I'm not sure what case is covered by cells of different types as

                                  opposed to tissues. Is there a collection of cells of different types

                                  that is not an organ, multi-tissue aggregate or portion of tissue, and

                                  yet still has "inherent 3D shape and is generated by coordinated

                                  expression of the organism's own genome"?

                                             is_a CARO:0000011 ! anatomical system [DEF: "Anatomical group that is

                                        has as its parts distinct anatomical structures interconnected by anatomical

                                        structures at a lower level of granularity."]

                                             is_a CARO:0000041 ! anatomical cluster  [DEF: "Anatomical group that

                                        has its parts adjacent to one another."]

                                        Dentition would seem to fit 'anatomical cluster' - unless we consider a

                                        child who has a mixed dentition, consisting of a mixture of primary

                                        dentition, secondary dentition, and sets of unerupted teeth (maybe this

                                        isn't "canonical")

                                  Teeth aren't necessarily adjacent (at least in the common sense),

                                  though I see the RO sense is simply nearby "in spatial proximity".

                                  Since the mouth can open quite wide it's not clear they would be

                                  considered adjacent in even that sense.

                                  FMA has both Anatomical Set as well as Anatomical Structure, BTW.

                                  Ii haven't been coordinating with CARO but will make an effort to do

                                  so in the subset of terms I extract to represent teeth.

                                  -Alan

                                        On Dec 13, 2011, at 10:38 AM, Todd Detwiler wrote:

                                        Alan, where are you getting this definition of Anatomical set? I have looked

                                        in the live FMA, the last release version of the FMA (v3.2) and the version

                                        before that (from early 2010, v3.1). All have the following definition for

                                        Anatomical set:

                                        "Material anatomical entity which consists of the maximum number of

                                        discontinuous members of the same class. Examples: set of cranial nerves,

                                        ventral branches of aorta, set of mammary arteries, thoracic viscera, dental

                                        arcade."

                                        It does not say that the members of the set must be from the class

                                        Anatomical structure.

                                        Todd

                                        Landon Todd Detwiler

                                        Structural Informatics Group (SIG)

                                        University of Washington

                                        On 12/13/11 10:15 AM, Alan Ruttenberg wrote:

                                        On Fri, Dec 9, 2011 at 2:41 PM,  <mej...@comcast.net> wrote:

                                        Right. My point is that the definitions disallows members of anatomical sets

                                        being other anatomical sets(many of the cases listed) or things that are not

                                        anatomical structures, such as blood:

                                        'Hematopoietic system'(FMA_9667) has member 'Portion of blood'(FMA_9670)

                                        which is not anatomical structure

                                        OM: The definition of anatomical set does not disallow body substances, that

                                        why we have the class "Hematopoietic system".

                                        We also have sets of immaterial anatomical entities, such as "Set of

                                        subarachnoid sulci" and "Set of external nares".

                                        The definition of Anatomical set as provided by the FMA is:

                                        "material anatomical entity which has as its members the maximal

                                        number of anatomical structures of the same type."

                                        As I read the current FMA, portion of body substance, being a sibling

                                        of Anatomical set, is disjoint from it. So the definition contradicts

                                        the assertion. I am advocating that one of them be changed so that

                                        there is no longer a contradiction.

                                        I would also suggest that the definition be changed to not say

                                        anything about the "maximal number", as this raises issues regarding

                                        time. Instead, push any kind of such maximality down to the particular

                                        terms. For example, the term "dentition" could be defined as being the

                                        set of all the teeth that are present at any time, if desired.

                                        -Alan

           David Osumi-Sutherland, PhD

           Ontologist

           FlyBase / Virtual Fly Brain

           Department of Genetics,

           University of Cambridge,

           Downing Street,

           Cambridge, CB2 3EH, UK

           Tel: +44 (0)1223 333 963

           Fax: +44 (0)1223 766 732

           http://www.virtualflybrain.org

David Osumi-Sutherland, PhD

Ontologist

Virtual Fly Brain / FlyBase

Department of Genetics

University of Cambridge

Downing Street

Cambridge, CB2 3EH, UK

+44 (0)1223 333 963

http://www.virtualflybrain.org

Dr. Melissa Haendel

eagle-i Networking Research Resources

OHSU Library http://www.ohsu.edu/library/

Department of Medical Informatics and Epidemiology
Oregon Health & Science University
hae...@ohsu.edu
skype: melissa.haendel

503-407-5970

[Melissa Haendel]

Hi David,

Yes I agree, we want to avoid the duplication. I think the question is though, do we need the set class if it can be accommodated as a group or cluster? Do we really gain anything by having the set class?

cheers

melissa

On Dec 15, 2011, at 9:34 AM, David Osumi-Sutherland wrote:

On 15 Dec 2011, at 15/Dec/2011 14:08:01, Melissa Haendel wrote:

> Sorry I've been out of the loop this week and have had super spotty internet (government agencies for you).
>
> Anyway, while we did not include sets in caro for the reason that David mentions, we do in fact have things that really are sets in our ontologies - we just don't have the "set" supertype- call it whatever you like. For example, we have the the chondrocranium, which is the set of cranial cartilage elements or cranial bone-that-has-replaced cartilage elements. The chondroranium is currently classified as an anatomical cluster in uberon: "Anatomical cluster that is part of the cranium and composed entirely of either cartilage, cartilage replacement bones or a mixture of both[ZFA]. that part of the neurocranium formed by endochondral ossification and comprising the bones of the base of the skull[TFD]." comment: note the definition here specifically mentions cartilage or cartilage replacement bones. we make chondrocranium part_of neurocranium. The cartilage elements and endochondral bones all have a part_of relation to the chondrocranium.
>
> The point here is that the dentition is much the same issue- it changes its composition as a "set" over time.

These seem like cases where you both need an anatomical group term (not all the parts are connected to each other) and where the parts might sensibly be considered to all belong to the same class.  So, I can see the justification for an anatomical set term here. But the FMA seems to use anatomical set much more broadly - to the degree that I have a general worry about the duplication of classification as partonomy.  This is what which and I were keen to avoid.  For example, see the FMA explorer screenshot below:

<PastedGraphic-1.png>

- - David

>
> As for joint, it is defined (in VAO) as "anatomical cluster that consists of two or more adjacent bones or cartilages, which may be interconnected by various types of tissues". I recall a lot of discussion about this, I think perhaps because they are not always connected it was defined this way? This would go against the current definition in the FMA so should probably check on this (in that parts must be connected).
>
> -Melissa
>
>
>

[Alan Ruttenberg]

On Thu, Dec 15, 2011 at 9:34 AM, David Osumi-Sutherland
<dj...@gen.cam.ac.uk> wrote:
> On 15 Dec 2011, at 15/Dec/2011 14:08:01, Melissa Haendel wrote:
>
>> Sorry I've been out of the loop this week and have had super spotty internet (government agencies for you).
>>
>> Anyway, while we did not include sets in caro for the reason that David mentions, we do in fact have things that really are sets in our ontologies - we just don't have the "set" supertype- call it whatever you like. For example, we have the the chondrocranium, which is the set of cranial cartilage elements or cranial bone-that-has-replaced cartilage elements. The chondroranium is currently classified as an anatomical cluster in uberon: "Anatomical cluster that is part of the cranium and composed entirely of either cartilage, cartilage replacement bones or a mixture of both[ZFA]. that part of the neurocranium formed by endochondral ossification and comprising the bones of the base of the skull[TFD]." comment: note the definition here specifically mentions cartilage or cartilage replacement bones. we make chondrocranium part_of neurocranium. The cartilage elements and endochondral bones all have a part_of relation to the chondrocranium.
>>
>> The point here is that the dentition is much the same issue- it changes its composition as a "set" over time.

I understand that. There are many cases like that. And for primary
dentition I would advocate that point of view. But it seemed to go to
far with the complete set of dentition. I proposed:

"My inclination would therefore be to only use Anatomical set for each
of these separately, with the same caveats you mention earlier
applicable, and not to name sets that include members of both as,
given the current FMA, the members of Anatomical sets seem to
generally be more approximately
contemporaneous than is the case with primary and secondary teeth."

For some analogies, I don't believe BFO would name an entity that was
the set of presidents, envisaged as a bag the extends over time to
scoop up all of them. In FMA I don't think we would have "Entire set
of eyelash hairs" to include all the ones you now have as well as the
ones that fell out, or "Entire set of portions of mucous" to mean all
the portions of mucous you have ever generated, including those
relegated to tissue over the years.

> These seem like cases where you both need an anatomical group term (not all the parts are connected to each other) and where the parts might sensibly be considered to all belong to the same class.  So, I can see the justification for an anatomical set term here. But the FMA seems to use anatomical set much more broadly - to the degree that I have a general worry about the duplication of classification as partonomy.  This is what which and I were keen to avoid.  For example, see the FMA explorer screenshot below:

I have similar concerns. As I demonstrated in an earlier email, these
sets can generally be reconstituted from queries in terms of the class
of the members. There may be reasonable exceptions, if the choice of
members is the result of something that doesn't have to do with body
structure (for instance as an entity denoted in a surgical procedure).
From the comments of Onard, Some of the use seems to arise from
developing in an environment with no significant reasoning capability.
It would be good to review the use of Anatomical set with a mind to
removing those that don't have a clear justification. A first stop
would be the short list I identified where anatomical sets have
members that are sets as well as subclasses

-Alan

[mej...@u.washington.edu]

There's a lot of good stuff in the exchanges. let me start with this one:

On Thu, 15 Dec 2011, Alan Ruttenberg wrote:
>
> BTW, in another email you asked for another example where the member/subclass confusion occurred. Here is a query I used to fish for them
>

> (map-fmaￜ
> ￜ(lambda (e)ￜ
> ￜ ￜ(when (and (someￜ
> (lambda(m)ￜ
> ￜ(find fma::|Anatomical set| (fma-superclasses m)))

> (slot-values e fma::|member|))

> ￜ ￜ ￜ (fma-children e)) (print e))))

>
> Which means: Find me sets that have sets as members and also subclasses.
>

> Musculature of trunkￜ
> Set of joints of footￜ
> Set of basal gangliaￜ
> Set of all visceraￜ
> Dentitionￜ
> Musculature of perineumￜ
> Set of arteries of footￜ
> Musculature of trunk properￜ
> Skeleton of free lower limbￜ
> Set of viscera of abdomenￜ
> Set of phalanges of footￜ
> Skeleton of footￜ
> Skeleton of upper limb
>
> If you look at Skeleton of foot ï¿œyou see, for example, that it has member Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply that the right foot has parts all of the tarsal bones of
> both feet.

I think this is a decision you made with your program where you implied that the right foot has parts all of the tarsal bones of both feet. In the FMA, the laterality differentia has been declared for the subclasses of Skeleton of foot, Skeleton of right foot and Skeleton of left foot. And their corresponding properties, e.g. part membership, have been declared at that level. Why would I go back to the superclass and get the part membership of that superclass and apply those to its subclasses? In frames, the subclasses inherits the template slots of the superclass, but not the values of the the superclass' own slots. I think this is a rule that you have to implement in your program.

>
> -Alan
>
> <I'll take the below up in another message, but essentially: there's no time like the present (heh, dual use phrase :) ). Hopefully Barry and Cornelius will chime in, but in their absence I think there's enough
> grownups around for us to make progress>
>
>
>
> Otherwise I propose we just decide to drop "Entire dentition" if it
> will mean that set.
> If it would have the meaning "at any given time having members all the
> teeth in a mouth" that might be a useful class.
>
> Maybe analogous to this
> is the class "Portion of blood". It subsumes both "portion of blood in right

> ventricle" and "entire portion of blood". The constituents of ï¿œall types of

> "portion of blood" also change over time (faster, of course) and so when we
> say "portion of blood" do we refer only to a portion present during
> childhood or during the individual's entire life?
>
>
> Here we are not talking about a set, but rather a continuant and its
> parts. If we were to talk about something set-like, it would be a kind
> of aggregate that acts similar to our bodies losing and gaining parts
> in time. I don't think it is analogous to teeth.
>
>

> Sorry I was not clear on this one. I didn't mean that portion of blood is a set. ï¿œWhat I wanted to say is that a class subsumes both portions (parts) and the entire whole. Portion of substance ï¿œmay be a

[Alan Ruttenberg]

>> If you look at Skeleton of foot  you see, for example, that it has member

>> Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply
>> that the right foot has parts all of the tarsal bones of
>> both feet.
>
>
> I think this is a decision you made with your program where you implied that
> the right foot has parts all of the tarsal bones of both feet. In the FMA,
> the laterality differentia has been declared for the subclasses of Skeleton
> of foot, Skeleton of right foot and Skeleton of left foot. And their
> corresponding properties, e.g. part membership, have been declared at that
> level. Why would I go back to the superclass and get the part membership of
> that superclass and apply those to its subclasses? In frames, the subclasses
> inherits the template slots of the superclass, but not the values of the the
> superclass' own slots. I think this is a rule that you have to implement in
> your program.

The question is how to make a logically consistent version of the FMA.
What we try to do first is to look at the relations and try to make
sense of them. The member_of relation implies has_part (spoken by
Onard) and assertions on classes, following the theory in the
Relations in Biomedical Ontology paper, bottom out as quantified
expressions over individuals. The inheritance follows that.

But basically, I'd be happy with any documented and coherent story on
how to interpret the frames, which suffer from, unfortunately, the
defect of not having such a story.

Here's how I approach it: We need to be able to evaluate the
sensibility of the representation according to the general ontological
principles that we use everywhere. The *only* theory of relations
between classes is the one in the Relations paper.

Now the language in which you discussed the frames semantics made use
of the idea that there are slots on classes with values. Let's take
one of them:

Foot (class) member Set of tarsal bones (class)
and from Onard we infer
Foot (class) part_of Set of tarsal bones (class)

The thing is, I don't understand what it means for a class to have
parts without further explanation. The relation paper, however, says
how to do that and I applied the results there to this case.

Below is the interpretation. Do you have an interpretation having it
mean anything other than:

Every instance of foot has some part that is an instance of Set of
tarsal bone?

If you agree that this is the only interpretation, then the inheritance follows.

Since the Set of tarsal bones itself has members/parts, I have to also
conclude that instances of those are parts of each instance of foot.
Since every right foot is a foot, I would conclude that every instance
of right foot also has an instance of the set as part, and therefore
also has instances of the tarsal bones as part.

Frames, and OWL, are servants to meaning, not creators. When their
semantics differ with meaning, sense needs to win.

But if you disagree, that's also fine, but then the rules of the game
have it that you need to say what it *does* mean (in terms of the
body), and how to discern that meaning from the representation.
Following that, I and others would apply those rules for the other
parts of the FMA and see what happens.

Make sense?

-Alan

[Christine Golbreich]

2011/12/16 Alan Ruttenberg <alanrut...@gmail.com>:

>>> If you look at Skeleton of foot  you see, for example, that it has member
>>> Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply
>>> that the right foot has parts all of the tarsal bones of
>>> both feet.
>>
>>
>> I think this is a decision you made with your program where you implied that
>> the right foot has parts all of the tarsal bones of both feet. In the FMA,
>> the laterality differentia has been declared for the subclasses of Skeleton
>> of foot, Skeleton of right foot and Skeleton of left foot. And their
>> corresponding properties, e.g. part membership, have been declared at that
>> level. Why would I go back to the superclass and get the part membership of
>> that superclass and apply those to its subclasses? In frames, the subclasses
>> inherits the template slots of the superclass, but not the values of the the
>> superclass' own slots. I think this is a rule that you have to implement in
>> your program.
>
> The question is how to make a logically consistent version of the FMA.
> What we try to do first is to look at the relations and try to make
> sense of them. The member_of relation implies has_part (spoken by
> Onard)

indeed, in the FMA artefact member is represented as a subproperty of
part and member_of as a subproperty of part_of
I dont' understand this given the semantics of member in the original
FMA paper, see my comment below

> and assertions on classes, following the theory in the
> Relations in Biomedical Ontology paper, bottom out as quantified
> expressions over individuals. The inheritance follows that.

From Cornelius & al 2003 p. 489 I understood that
- the member and member_of relation were initially specified jointly
with the notion of Anatomical_set : member_of is the relationship
that relates a member to a set [1]
- that the meaning of the member relation is closer to the notion of
'element' (instance) than to the notion of 'part' [1] (though
asserted being different [2] )
[1] "Anatomical sets have members, rather than parts (e.g., Oculomotor
nerve is a member of Set of cranial nerves)." [Cornelius & al 2003 p.
489]
[2] "Members of an anatomical set, as defined in the FMA, are distinct
from elements of a mathematical set"

Would it be possible to first clarify/agree the definition of Anatomical_set ?

What do you mean by [3] "the members do not define an anatomical set
(which is a class), whereas a mathematical set is defined by its
members."
why the members don't define an anatomical set ?
in your view is it because of an order? because they are classes? or
another reason ?

My interpretation in reading 1-2-3 and the whole paragraph about
Anatomical_set (as said in my previous mail) is that sets were classes
originally designed to specify different "species" among the
anatomical entities and their elements (which are themselves classes,
therefore so called 'member' ). A set is an individual that denotes a
group (species) of different classes that share a common property
for example the class named Set_of_cranial_nerves refers to the
individual representing the species of cranial nerves belonging to
the (meta)class of all anatomical species (that is the species of
nerves that emerge directly from the brain), hence the list of its
elements (member) is the list of classes below :
Olfactory+nerve Optic+nerve-tract+complex Oculomotor+nerve
Trochlear+nerve Trigeminal+nerve Abducens+nerve Facial+nerve
Vestibulocochlear+nerve Glossopharyngeal+nerve Vagus+nerve
Hypoglossal+nerve), the species "Set of fingers", Set of toes, etc.

NB As said in my previous mail, if this interpretation is correct, it
is possible to use OWL 2 metamodelling to represent metaclasses such
as set_of_xxx, and even use the same term for the class Cranial_nerve
and the individual Cranial_nerve denoting the species (set)

I did not have the opportunity to discuss that, is this
interpretation of sets correct or erroneous ?

Regards

> But basically, I'd be happy with any documented and coherent story on
> how to interpret the frames, which suffer from, unfortunately, the
> defect of not having such a story.
>
> Here's how I approach it: We need to be able to evaluate the
> sensibility of the representation according to the general ontological
> principles that we use everywhere. The *only* theory of relations
> between classes is the one in the Relations paper.
>
> Now the language in which you discussed the frames semantics made use
> of the idea that there are slots on classes with values. Let's take
> one of them:
>
> Foot (class)  member Set of tarsal bones (class)
> and from Onard we infer
> Foot (class)  part_of Set of tarsal bones (class)

is the assertion Foot (class) member Set of tarsal bones (class) correct ?

> The thing is, I don't understand what it means for a class to have
> parts without further explanation. The relation paper, however, says
> how to do that and I applied the results there to this case.
>
> Below is the interpretation. Do you have an interpretation having  it
> mean anything other than:
>
>    Every instance of foot has some part that is an instance of Set of
> tarsal bone?
>
> If you agree that this is the only interpretation, then the inheritance follows.
>
> Since the Set of tarsal bones itself has members/parts, I have to also
> conclude that instances of those are parts of each instance of foot.
> Since every right foot is a foot, I would conclude that every instance
> of right foot also has an instance of the set as part, and therefore
> also has instances of the tarsal bones as part.
>
> Frames, and OWL, are servants to meaning, not creators. When their
> semantics differ with meaning, sense needs to win.
>
> But if you disagree, that's also fine, but then the rules of the game
> have it that you need to say what it *does* mean (in terms of the
> body), and how to discern that meaning from the representation.
> Following that, I and others would apply those rules for the other
> parts of the FMA and see what happens.
>
> Make sense?
>
> -Alan

--
Christine

Please do not print this e-mail unless you really need to. Wasting
paper costs trees and releases carbon into the atmosphere.

[mej...@u.washington.edu]

I'd like to try again and explain our approach in the FMA. As I have mentioned in the previous message the differentia between anatomical cluster and anatomical set is the property connectivity. I guess the question is what benefit do we get in separating the two instead of having a class called "group" which encompasses both set and cluster? A cluster represents one fully formed unit such as the root of lung or joint. On the other hand, a set consists of two or more separate fully formed units, e.g. set of 10 fingers. We have the class "set" so that we are consistent with representing only singular units in the ontology but in essence a set of 10 fingers (10 entities) is very different from a cluster "joint" (1 entity). So why is the distinction important? I think the most obvious reason is our fidelity to the definitions we declared for anatomical classes. We define "Anatomical structure"
as "Material anatomical entity which is generated by coordinated expression of the organism's own genes that guide its morphogenesis; has inherent 3D shape; ITS PARTS ARE CONNECTED and spatially related to one another in patterns determined by coordinated gene expression". A set does not satisfy or meet the said definition which specifies connectivity of parts. Members of a set are discontinuous. And so "Anatomical structure" (which includes "Anatomical cluster") and "Anatomical set" are siblings in our hierarchy.

"Joint" qualifies as a cluster because it is a single unit with connected parts. For example, a suture is a joint which consists of parts of adjacent articulating bones CONNECTED by collagenous dense connective tissue.

Onard

On Thu, 15 Dec 2011, Melissa Haendel wrote:

> Hi David,Yes I agree, we want to avoid the duplication. I think the question is though, do we need the set class if it can be accommodated as a group or cluster? Do we really gain anything by having the set

> class?
>
> cheers
> melissa
>
> On Dec 15, 2011, at 9:34 AM, David Osumi-Sutherland wrote:
>
> On 15 Dec 2011, at 15/Dec/2011 14:08:01, Melissa Haendel wrote:
>
> > Sorry I've been out of the loop this week and have had super spotty internet (government agencies for you).
> >
> > Anyway, while we did not include sets in caro for the reason that David mentions, we do in fact have things that really are sets in our ontologies - we just don't have the "set" supertype- call it
> whatever you like. For example, we have the the chondrocranium, which is the set of cranial cartilage elements or cranial bone-that-has-replaced cartilage elements. The chondroranium is currently
> classified as an anatomical cluster in uberon: "Anatomical cluster that is part of the cranium and composed entirely of either cartilage, cartilage replacement bones or a mixture of both[ZFA]. that
> part of the neurocranium formed by endochondral ossification and comprising the bones of the base of the skull[TFD]." comment: note the definition here specifically mentions cartilage or cartilage
> replacement bones. we make chondrocranium part_of neurocranium. The cartilage elements and endochondral bones all have a part_of relation to the chondrocranium.
> >
> > The point here is that the dentition is much the same issue- it changes its composition as a "set" over time.
>

> These seem like cases where you both need an anatomical group term (not all the parts are connected to each other) and where the parts might sensibly be considered to all belong to the same class.ￜ

> So, I can see the justification for an anatomical set term here. But the FMA seems to use anatomical set much more broadly - to the degree that I have a general worry about the duplication of

> classification as partonomy.ￜ This is what which and I were keen to avoid.ￜ For example, see the FMA explorer screenshot below:

>
> <PastedGraphic-1.png>
>
>
>
> - - David
>
> >
> > As for joint, it is defined (in VAO) as "anatomical cluster that consists of two or more adjacent bones or cartilages, which may be interconnected by various types of tissues". I recall a lot of
> discussion about this, I think perhaps because they are not always connected it was defined this way? This would go against the current definition in the FMA so should probably check on this (in that
> parts must be connected).
> >
> > -Melissa
> >
> >
> >
>
>
> Dr. Melissa Haendel
>
> eagle-i Networking Research Resources

> OHSU Libraryï¿œhttp://www.ohsu.edu/library/

[mej...@u.washington.edu]

Alan,

This is more complicated than I thought. My response to your example using classes, slots and values did not address the real issue. I have largely dealt with classes and only quite recently did I begin to think in terms of instances (tokens, individuals). In Protege frames, we (past authors and myself) were satisfied in representing the "Foot" as having parts "Set of tarsal bones" without realizing that the class "Foot" is really instantiated by all the canonical right and left feet in the world and that the class "Set of tarsal bones" is instantiated by all sets of right and left tarsal bones. So our assertion in frames, that "Foot" has part "Set of tarsal bones" was not right. No foot has as parts sets of left and right tarsal bones but rather the assertion should have been class "Foot" has part "Set of right tarsal bones OR Set of left tarsal bones". And then carry out the next level of
assertions for the subclasses "Right foot" and "Left foot" where "Right foot" has parts "Set of right tarsal bones". Protege frames did not (does not) have the capability to declare "OR". I am assuming this can this be achieved in OWL?

I also have a few comments in-line below.

Onard

On Thu, 15 Dec 2011, Alan Ruttenberg wrote:

>>> If you look at Skeleton of foot ï¿œyou see, for example, that it has member

>>> Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply
>>> that the right foot has parts all of the tarsal bones of
>>> both feet.
>>
>>
>> I think this is a decision you made with your program where you implied that
>> the right foot has parts all of the tarsal bones of both feet. In the FMA,
>> the laterality differentia has been declared for the subclasses of Skeleton
>> of foot, Skeleton of right foot and Skeleton of left foot. And their
>> corresponding properties, e.g. part membership, have been declared at that
>> level. Why would I go back to the superclass and get the part membership of
>> that superclass and apply those to its subclasses? In frames, the subclasses
>> inherits the template slots of the superclass, but not the values of the the
>> superclass' own slots. I think this is a rule that you have to implement in
>> your program.
>
> The question is how to make a logically consistent version of the FMA.

Yes, that's the goal.

> What we try to do first is to look at the relations and try to make
> sense of them. The member_of relation implies has_part (spoken by
> Onard) and assertions on classes, following the theory in the
> Relations in Biomedical Ontology paper, bottom out as quantified
> expressions over individuals. The inheritance follows that.
>
> But basically, I'd be happy with any documented and coherent story on
> how to interpret the frames, which suffer from, unfortunately, the
> defect of not having such a story.

I'd be happy to tell the stories.

>
> Here's how I approach it: We need to be able to evaluate the
> sensibility of the representation according to the general ontological
> principles that we use everywhere. The *only* theory of relations
> between classes is the one in the Relations paper.

We have in fact covered this to some extent in the paper:
http://sigpubs.biostr.washington.edu/archive/00000204/
In particular, see Table 4.5 on page 116

>
> Now the language in which you discussed the frames semantics made use
> of the idea that there are slots on classes with values. Let's take
> one of them:
>
> Foot (class) member Set of tarsal bones (class)
> and from Onard we infer
> Foot (class) part_of Set of tarsal bones (class)

I don't quite follow the above. Did you mean that "Foot" has member "Set of tarsal bones"? And "Foot" has part (not part_of) "Set of tarsal bone"? The former is not true or not in the FMA and the latter is true via transitivity.

What we have is:
Foot (has constitutional part)
Skeleton of foot (and others like skin, vasculature, lymphatics, etc)
Set of tarsal bones (has members)
Cuboid bone
Talus, etc
Set of metatarsal bones
Set of phalanges
Etc.

>
> The thing is, I don't understand what it means for a class to have
> parts without further explanation. The relation paper, however, says
> how to do that and I applied the results there to this case.
>
> Below is the interpretation. Do you have an interpretation having it
> mean anything other than:
>
> Every instance of foot has some part that is an instance of Set of
> tarsal bone?

Yes.

>
> If you agree that this is the only interpretation, then the inheritance follows.
>
> Since the Set of tarsal bones itself has members/parts, I have to also
> conclude that instances of those are parts of each instance of foot.

You mean parts of some instance of foot?

> Since every right foot is a foot, I would conclude that every instance
> of right foot also has an instance of the set as part, and therefore
> also has instances of the tarsal bones as part.

Sorry, I don't quite follow this. Can you elaborate some more?

[mej...@u.washington.edu]

Hi Christine,

Many thanks for the comments. My response in-line below.

best,
Onard

On Fri, 16 Dec 2011, Christine Golbreich wrote:

> 2011/12/16 Alan Ruttenberg <alanrut...@gmail.com>:
>>>> If you look at Skeleton of foot ï¿œyou see, for example, that it has member

>>>> Set of Tarsal Bones, and subclass Skeleton of right foot, which would imply
>>>> that the right foot has parts all of the tarsal bones of
>>>> both feet.
>>>
>>>
>>> I think this is a decision you made with your program where you implied that
>>> the right foot has parts all of the tarsal bones of both feet. In the FMA,
>>> the laterality differentia has been declared for the subclasses of Skeleton
>>> of foot, Skeleton of right foot and Skeleton of left foot. And their
>>> corresponding properties, e.g. part membership, have been declared at that
>>> level. Why would I go back to the superclass and get the part membership of
>>> that superclass and apply those to its subclasses? In frames, the subclasses
>>> inherits the template slots of the superclass, but not the values of the the
>>> superclass' own slots. I think this is a rule that you have to implement in
>>> your program.
>>
>> The question is how to make a logically consistent version of the FMA.
>> What we try to do first is to look at the relations and try to make
>> sense of them. The member_of relation implies has_part (spoken by
>> Onard)
>
> indeed, in the FMA artefact member is represented as a subproperty of
> part and member_of as a subproperty of part_of
> I dont' understand this given the semantics of member in the original
> FMA paper, see my comment below

We modeled "membership" as a kind of parthood relation based on the categories of meronymic relationships proposed by Winston, et.al.
Winston ME, Chaffin R, Herrman D. A taxonomy of part-whole relations. Cognitive Sci. 1987; 11:417-444.

>
>> and assertions on classes, following the theory in the
>> Relations in Biomedical Ontology paper, bottom out as quantified
>> expressions over individuals. The inheritance follows that.
>
> From Cornelius & al 2003 p. 489 I understood that
> - the member and member_of relation were initially specified jointly
> with the notion of Anatomical_set : member_of is the relationship
> that relates a member to a set [1]
> - that the meaning of the member relation is closer to the notion of
> 'element' (instance) than to the notion of 'part'

I think there's some confusion in how the term "member" is used, as a class and as an instance. So at the class level, the members of a set class are also classes not instances. For example, the members of the class "Set of fingers" are also classes; "Thumb", "Index finger", "Middle finger", etc. And then there's the instantiation of each of those classes by members that are individuals or tokens (your set of fingers, your thumb, etc.)

[1] (though
> asserted being different [2] )
> [1] "Anatomical sets have members, rather than parts (e.g., Oculomotor
> nerve is a member of Set of cranial nerves)." [Cornelius & al 2003 p.
> 489]
> [2] "Members of an anatomical set, as defined in the FMA, are distinct
> from elements of a mathematical set"

We just want to make the distinction that although members of an anatomical set, such as a set of lymph nodes, are not directly connected with one another, there are structures (lymphatic vessels) that connect them indirectly. But that information is lost or ignored in the discourse. It is interesting that clinicians talk about metastatic spread to lymph nodes and record the findings involving only the set of nodes. And the ensuing interpretation in an ontology captures only "lymph nodes" as a set and not the vessels connecting them. But in actuality they meant the lymphatic chains, consisting of lymph nodes and vessels. How else can tumor spread from one node to another.

>
> Would it be possible to first clarify/agree the definition of Anatomical_set ?

Any collection of discontinuous 3-D anatomical entities (objects and spaces).

>
> What do you mean by [3] "the members do not define an anatomical set
> (which is a class), whereas a mathematical set is defined by its
> members."

I can't quite remember the rationale for this. I defer to Cornelius.

> why the members don't define an anatomical set ?
> in your view is it because of an order? because they are classes? or
> another reason ?
>
> My interpretation in reading 1-2-3 and the whole paragraph about
> Anatomical_set (as said in my previous mail) is that sets were classes
> originally designed to specify different "species" among the
> anatomical entities and their elements (which are themselves classes,
> therefore so called 'member' ). A set is an individual that denotes a
> group (species) of different classes that share a common property
> for example the class named Set_of_cranial_nerves refers to the
> individual representing the species of cranial nerves belonging to
> the (meta)class of all anatomical species (that is the species of
> nerves that emerge directly from the brain), hence the list of its
> elements (member) is the list of classes below :
> Olfactory+nerve Optic+nerve-tract+complex Oculomotor+nerve
> Trochlear+nerve Trigeminal+nerve Abducens+nerve Facial+nerve
> Vestibulocochlear+nerve Glossopharyngeal+nerve Vagus+nerve
> Hypoglossal+nerve), the species "Set of fingers", Set of toes, etc.

I think I follow except for the part the "a set is an individual that denotes a group (species) of different classes". Perhaps you meant a set in the FMA sense is a class which has as members subset classes?

>
> NB As said in my previous mail, if this interpretation is correct, it
> is possible to use OWL 2 metamodelling to represent metaclasses such
> as set_of_xxx, and even use the same term for the class Cranial_nerve
> and the individual Cranial_nerve denoting the species (set)

Sorry I'm not sure I follow this.

>
> I did not have the opportunity to discuss that, is this
> interpretation of sets correct or erroneous ?
>
> Regards
>
>> But basically, I'd be happy with any documented and coherent story on
>> how to interpret the frames, which suffer from, unfortunately, the
>> defect of not having such a story.
>>
>> Here's how I approach it: We need to be able to evaluate the
>> sensibility of the representation according to the general ontological
>> principles that we use everywhere. The *only* theory of relations
>> between classes is the one in the Relations paper.
>>
>> Now the language in which you discussed the frames semantics made use
>> of the idea that there are slots on classes with values. Let's take
>> one of them:
>>

>> Foot (class) ï¿œmember Set of tarsal bones (class)

>> and from Onard we infer

>> Foot (class) ï¿œpart_of Set of tarsal bones (class)

>
> is the assertion Foot (class) member Set of tarsal bones (class) correct ?

No, I don't agree with this assertion.

>
>
>> The thing is, I don't understand what it means for a class to have
>> parts without further explanation. The relation paper, however, says
>> how to do that and I applied the results there to this case.
>>

>> Below is the interpretation. Do you have an interpretation having ï¿œit
>> mean anything other than:
>>
>> ￜ ￜEvery instance of foot has some part that is an instance of Set of