FMA in OWL 2013

[Chris Mungall]

It's been 3 or 4 years since we met in Stanford, from the outside it looked as if there hasn't been much progress in terms of getting a usable DL version of FMA generated as part of the FMA release pipeline.

Congratulations are in order then to Christine for producing a working tool the FMA team can use to generate OWL2 DL versions of the FMA, described here:
>>> The Foundational Model of Anatomy in OWL 2 and its use
>>> http://dx.doi.org/10.1016/j.artmed.2012.11.002

(Unfortunately it's not open access)

The tool produces two versions:

no metaclasses (suitable for the majority of end-users):
http://gforge-lirmm.lirmm.fr/gf/download/docmanfileversion/ 214/747/FMA 3.0 noMTC 100702.owl.zip

with metaclasses:
http://gforge-lirmm.lirmm.fr/gf/download/ docmanfileversion/215/748/FMA 3.0 MTC 100701.owl.zip

It would be great to see this incorporated into the FMA release process, such that there are regular releases of the FMA in both frames and OWL2. I'd also encourage everyone involved with the FMA to incorporate this tool into the development process, to use reasoners to check for errors, etc.

Unfortunately the parallel work that Alan has been doing is unpublished, so there is no comparison of approaches. My understanding (based on a combination of guesswork and hazy memory of the 2009 meeting) is that both approaches produce semantically identical 'core axioms' (e.g. subclass and existential restrictions for parthood relationships - what the majority of the DL consumers want) but produce different sets of 'value-added axioms' (which are extremely useful for assisting maintenance and QC of the FMA). In particular the Golbreich approach uses lexical methods to generate equivalence axioms (v useful for automatic classification) and the Ruttenberg approach makes use of a lot of buried frame information to generate disjointness axioms for spatial partitions. Ideally these could be combined a single tool, with the user selecting which axiom generation heuristics to employ.

We used to use this list to discuss some of the issues Alan found using his tool, but this seems to have tailed off over the years.

Some questions for Christine:

- do you intend to keep extending this tool and making regular releases, or do you consider this project done?
- would you be willing to make some changes to make the OWL compliant with OBO Foundry guidelines? This would include using the numeric identifier in the IRI, reusing standard annotation properties and object properties.
- are you aware of the OCDM project, and have you tried using the tool to generate an OWL version of OCDM?

--
Chris

[mej...@comcast.net]

Chris,

Many thanks for bringing this up! This is good news for me and I'm very much encouraged to move forward with the suggestions. We have been aware of the tool for some time now but I have not had the chance to follow up with Christine on this. As you know I'm technically challenged on this matter and so I have to consult with Todd and Jim on what our local group should do next.

Christine, is the tool available for us to try out? We would like to work with you on this. Perhaps I can call you directly and talk more in detail on how we can pursue possible collaboration?

I very much look forward to any progress we can achieve here. FMA in OWL is long due!

Thanks!
Onard

---

From: "Chris Mungall" <cjmu...@lbl.gov>
To: fma-ow...@googlegroups.com, "Alan Ruttenberg" <alanrut...@gmail.com>, "Christine Golbreich" <Christine...@univ-rennes1.fr>
Cc: "Harry Stewart Hochheiser" <har...@pitt.edu>, "Melissa Haendel" <hae...@ohsu.edu>, "Todd Detwiler" <d...@u.washington.edu>, "Jim Brinkley" <brin...@u.washington.edu>, "mej...@u.washington.edu Mejino" <mej...@u.washington.edu>, "zooinfo...@gmail.com Travillian" <zooinfo...@gmail.com>
Sent: Thursday, January 3, 2013 6:00:38 PM
Subject: FMA in OWL 2013

[Jim Brinkley]

Chris et al,

This does indeed look like a great opportunity to make some progress on the FMA-OWL conversion we all want. As both Chris and Onard point out, its seems like the logical next step would   be for the FMA team to work with Christine to try out the FMA-OWLizer tool.  So Christine, perhaps you could let us  know (or perhaps Onard separately as he suggests) the status of the tool, and whether it is available for us to try out.

Jim

On Jan 4, 2013, at 10:59 AM, mej...@comcast.net wrote:

Chris,

Many thanks for bringing this up! This is good news for me and I'm very much encouraged to move forward with the suggestions. We have been aware of the tool for some time now but I have not had the chance to follow up with Christine on this. As you know I'm technically challenged on this matter and so I have to consult with Todd and Jim on what our local group should do next. 

Christine, is the tool available for us to try out? We would like to work with you on this. Perhaps I can call you directly and talk more in detail on how we can pursue possible collaboration?

I very much look forward to any progress we can achieve here. FMA in OWL is long due!

Thanks!
Onard

---

From: "Chris Mungall" <cjmu...@lbl.gov>
To: fma-ow...@googlegroups.com, "Alan Ruttenberg" <alanrut...@gmail.com>, "Christine Golbreich" <Christine...@univ-rennes1.fr>

Cc: "Harry Stewart Hochheiser" <har...@pitt.edu>, "Melissa Haendel" <hae...@ohsu.edu>, "Todd Detwiler" <d...@u.washington.edu>, "Jim Brinkley" <brin...@u.washington.edu>, "mej...@u.washington.edu Mejino" <mej...@u.washington.edu>, "zooinfo...@gmail.comTravillian" <zooinfo...@gmail.com>

[Melissa Haendel]

Hi all,

I just want to say how happy this makes me :-) and that I am looking forward to seeing this come to fruition. I am happy to help too, with QA as needed.

Best,

Melissa

Dr. Melissa Haendel

Assistant Professor
Ontology Development Group, 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

[Stefan Schulz]

Dear all,

I am happy so see some progress here.
@ Christine: can you send a preprint of your paper around?
@ Alan: can you share with us what you have done so far

Best regards,
Stefan



2013/1/5 Melissa Haendel <hae...@ohsu.edu>:

--

Stefan SCHULZ (Univ.-Prof. Dr. med.)

Institut für Medizinische Informatik,
Statistik und Dokumentation
Medizinische Universität Graz
Auenbruggerplatz 2/V
8036 Graz (Austria)

http://www.medunigraz.at/imi
http://g.co/maps/aqedt

+43 (0)316 385 16939
+43 (0)316 385 13201

http://purl.org/steschu
mailto:stefan...@medunigraz.at
Skype: stschulz

[ home: Afritschgasse 32/3
[ 8020 Graz (Austria)
[ mobile: +43 (0)699 150 96270
[ http://g.co/maps/m8rau

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

Hi All,

did you see that Antoine's paper,along with commentaries and editorial has appeared? I received the issue of the journal.

I am very pleased about the news of the FMA in OWL. I hope it means a boost for the FMA. Pleas let me know what happens, particularly whether Christine will play ball. What would happen if she maintains her old attitude?

Cornelius

On Fri, 4 Jan 2013, mej...@comcast.net wrote:

> Chris,
>
> Many thanks for bringing this up! This is good news for me and I'm very much
> encouraged to move forward with the suggestions. We have been aware of the tool for
> some time now but I have not had the chance to follow up with Christine on this. As
> you know I'm technically challenged on this matter and so I have to consult with
> Todd and Jim on what our local group should do next.
>
> Christine, is the tool available for us to try out? We would like to work with you
> on this. Perhaps I can call you directly and talk more in detail on how we can
> pursue possible collaboration?
>
> I very much look forward to any progress we can achieve here. FMA in OWL is long
> due!
>
> Thanks!
> Onard
>
>
>
>
>

> ___________________________________________________________________________________

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

Dear Chris,

You have received several responses welcoming the FMA OWL 2.I want to thank you personally for all your good work and efforts on making the maintenance and the availability of the FMA easier. I hope that the opportunity you have opened up will be exploited.

Cornelius

[Chris Mungall]

Here's a copy of the paper

[Christine Golbreich]

Dear Cornelius

I'm very happy to hear from you. Happy New Year 2013.

I absolutely miss what these words about me and my attitude refer to :

" ... particularly whether Christine will play the ball. What would happen if she maintains her old attitude" 

Can you please be more explicit ?

Best Regards

Christine 

[Christine Golbreich]

Hi Chris and all

Chris, thank you very much for your email to the list and also for sending the paper. I'm happy to get it :-) , I even did not have it.

There is a 2 weeks Christmas break here in France. I'm waiting for talking with my French colleagues, as soon as everybody is back (hopefully from tomorrow). I hope to answer  in more details if possible next. Meanwhile, see below my first impressions/reactions 

Some questions for Christine:

- do you intend to keep extending this tool and making regular releases, or do you consider this project done?

do you mean regular releases of the tool or of the FMA in OWL ?

- the project is absolutely not done, it's a first (second) step, there is still much to do and the tool might / should be improved and extended

- it's also still necessary to debug the FMA before using it in real applications. Reasoning with OWL may help to fix some errors but it's a hard and time consuming work which needs means.

- a loop and regular feedback should be achieved

- alignment between FMA and SNOMED-CT is another important issue

etc.

All this requires means. 

Unfortunately, on my side further development is being stuck nearly since  2 or 3 years, mainly because of lack of fundings and people availability. I already said it several times.

- would you be willing to make some changes to make the OWL compliant with OBO Foundry guidelines? This would include using the numeric identifier in the IRI, reusing standard annotation properties and object properties.

- I'm open to all changes that are possible and make sense. A possibility might be perhaps to add new options to the tool, for example to choose using the numeric identifier. 

- What do you mean "reusing standard annotation properties and object properties" ?

- are you aware of the OCDM project, and have you tried using the tool to generate an OWL version of OCDM?

I have not tried. What is the exact connection between OCDL and FMA ? Are they concretely  linked ?

the FMA-OWLizer tool is specific to FMA

- Regarding the other questions about the tool, of course I want to make it available to everybody to try it, it  was the initial goal/intent of this tool.

- Regarding collaboration, as always, I'm personally completely open to any collaboration of * any * type that may help to progress (as much as I can).  But for years now the recurrent problem is to have strengths. I'm still looking for a concrete means to make it in practice.

As some of you remember,  in the past,  aiming at collaborating for improving the FMA, I  had submitted (as PI) several NIH R01 collaborative proposals for grants. The last  one was “Relating Neuroimaging Data to Human Diseases”, submitted together with NCBO, BIRN, Drexel University, Manchester University (I. Horrocks) and others. Onard Mejino and Pr. Cornelius Ross supported the project and were involved as consultants for University of Washington. Unfortunately, all these attempts were unsuccessful.

After our 2009 meeting I had also tried to launch the idea to submit again together a new proposal, if I remember  well mainly focused on

- FMA revision and conversion to OWL 2 or extension, 

- Align, integrate the FMA and SNOMED-CT OWL 2 ontologies (relevant anatomical modules).

- Use reasoners for KE of FMA and SNOMED-CT

- Indexing and answering queries on EMR 

I contacted then several persons ( University of Washington, if I remember well) asking them to be the PI, but nobody was available and nothing happened next.

Today, things have progressed in many areas, and there are many new pressing things to do. I'm still available to contribute.

All ideas, suggestions and proposals are welcome !

Best regards,

Christine

2013/1/6 Chris Mungall <cjmu...@lbl.gov>

Here's a copy of the paper

--

Christine

 

 

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

 

 

[Christine Golbreich]

2013/1/4 <mej...@comcast.net>

Chris,

Many thanks for bringing this up! This is good news for me and I'm very much encouraged to move forward with the suggestions. We have been aware of the tool for some time now but I have not had the chance to follow up with Christine on this. As you know I'm technically challenged on this matter and so I have to consult with Todd and Jim on what our local group should do next. 

 

Christine, is the tool available for us to try out? We would like to work with you on this. Perhaps I can call you directly and talk more in detail on how we can pursue possible collaboration?

Onard, it will be my pleasure to talk with you by phone or Skype as you prefer.

Regards

Christine

[Christine Golbreich]

Hi Chris, Alan

> Unfortunately the parallel work that Alan has been doing [...]

> the Ruttenberg approach makes use of a lot of buried frame information to generate disjointness axioms for spatial partitions.

I only remember that we had a discussion at the meeting about the different partitions
I don't know what Alan has been doing.

Questions @ Alan and Chris
- where can I look at the work Alan has been doing  ? is it described or summarized somewhere ?
- has  Alan produced a version of the whole FMA in OWL ?
- is there an open access to it ?
- does an OWL version of the FMA including the disjointness axioms for spatial partitions exist  ?

Thanks

2013/1/4 Chris Mungall <cjmu...@lbl.gov>

[Alan Ruttenberg]

On Thu, Jan 10, 2013 at 2:11 PM, Christine Golbreich <cgol...@gmail.com> wrote:
> Hi Chris, Alan
>
>> Unfortunately the parallel work that Alan has been doing [...]
>
>
>> the Ruttenberg approach makes use of a lot of buried frame information to
>> generate disjointness axioms for spatial partitions.
>
> I only remember that we had a discussion at the meeting about the different
> partitions
> I don't know what Alan has been doing.
>
> Questions @ Alan and Chris
> - where can I look at the work Alan has been doing ? is it described or
> summarized somewhere ?

I haven't written a paper about it yet. Most of the results of the
work have been posted on this list, though some work has been done as
part of the OHD project. The repository is
http://svn.neurocommons.org/trunk/convert/fma which is unfortunately
down at the moment due to a hardware failure. I'll move it to github
and let y'all know when it is available. Apparently the visualizations
I put up and which are pointed to in some of my posts are also broken
:( I will try to fix that too.

> - has Alan produced a version of the whole FMA in OWL ?

No. I focus solely on modules (mouth, brain, kidney, etc) aiming to
generate correct (consistent, satisfiable, anatomically sensible, as
complete as FMA frames is) inferred partonomies. There's been enough
to do to get to that point at the moment. The plan was to move to
other relations only once the partonomy was fixed. I've been trying to
make sure any errors we find are pushed back into the FMA. Most
recently they were going to fix the situation where classes had a
different superclass and metaclass (they should be the same according
to Onard). I haven't had time to work on it recently, but I'll try to
check that.

However the bottom line is that I'm doing a different task than you
have been. I'm not creating a procedure/tool by which any version of
frames FMA can be converted to OWL. Rather I'm trying to get to a
version of the FMA in OWL that is of a quality that it is suitable for
use and further development. That involves also debugging, figuring
out what changes need to be made, lobbying Onard to make changes, etc.

> - is there an open access to it ?

Yes(modulo hardware failure), although as I haven't worked to write it
up it's not the easiest to understand. As I said, I'll post a github
version in another note. Note that "it" is code, some exploratory
modules, and some visualizations.

> - does an OWL version of the FMA including the disjointness axioms for spatial partitions exist?

There are axioms about bilaterally, mostly useful for finding clerical
errors where something left is part of something right. I'm not sure
to what spatial partitions you are referring to here.

A problem is that as long as the FMA is authored in frames, it is
difficult to work efficiently. New versions of the frames ontology
need to be reconciled with old versions of code and FMA patches to
check whether various patches are still necessary. Bulk changes that
are easy to make on the OWL side are not as easy to make on the frames
side. I'm not sure what to do about that situation. Protege is not a
very comfortable or effective place to be doing the sort of work that
needs to be done on the way to having a consistent version of the
frames, either, though I use the explanation features in protege to
help debug issues that come up - it's great for that.

Christine, what is the status of your system? If the code is open
source, it could be something we can all build on (that's what
Cornelius referred to as 'playing ball'). I think your version is
getting better, but in order to use it as a further basis for FMA
development it needs to be something that several of us can work on
together. Since it currently generates ontologies with many
unsatisfiable classes, I don't think it is a tool I would have end
users use. Rather, if it was made an open source project with several
of us, including Onard, contributing, we might be able to make it the
tool of choice for fixing issues in the FMA until the OWL so generated
*was* ready for widespread use.

Regards,
Alan