Human body ontology

classic Classic list List threaded Threaded
15 messages Options
Reply | Threaded
Open this post in threaded view
|

Human body ontology

raza
Hi,
I'm new to ontologies and Protege. I am trying to solve the following
problem:

I have created simple hierarchy of body parts (using subclasses), and a
parallel one for skin of each body part. What is the correct way of linking
say the "skin of hand" to the body part "hand"? I've been reading about
class restrictions but feel that it's too cumbersome to define a separate
class restriction for each pair body part and it's associated skin.

Any help would be appreciated!

- U



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Campbell, James R
Have you looked at the Foundatiobnal Model of Anatomy?

James R. Campbell MD
[hidden email]
cell: 402-290-5154
Secretary: 402-559-7502

> On Mar 12, 2019, at 7:02 PM, raza <[hidden email]> wrote:
>
> Non-UNMC email
>
>
>
>
> Hi,
> I'm new to ontologies and Protege. I am trying to solve the following
> problem:
>
> I have created simple hierarchy of body parts (using subclasses), and a
> parallel one for skin of each body part. What is the correct way of linking
> say the "skin of hand" to the body part "hand"? I've been reading about
> class restrictions but feel that it's too cumbersome to define a separate
> class restriction for each pair body part and it's associated skin.
>
> Any help would be appreciated!
>
> - U
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> protege-user mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-user
>

The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender.
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Matthew Horridge-2
Administrator
In reply to this post by raza
Hi,

You can (should) class expressions (restrictions to do this).  For example, 

’skin of hand’ SubClassOf partOf ‘hand’

You would add an expression like this in Protege


This is rather tedious to do by hand.  It sounds like you have the pattern, for

‘x’

and

‘skin of x' 

you would want to add

’skin of x’ SubClassOf partOf some ‘x’


If this is the case, I recommend using Excel (or similar) to generate to columns.  You would have ‘x’ in the first column and generate the second column with ’skin of x’.  i.e.,

 

If you already have the x values in Protege then you can copy and paste them from the hierarchy into Excel.  Once you have this spreadsheet you can generate OWL axioms from it using Cellfie (or a similar tool/toolchain, like DOS-DPs and ROBOT).  Choose, “Tools” -> “Generate axioms from Excel workbook”.

The rule for Cellfie would look like this:


Then, generate the axioms and you should have all of the connections as required.  

If you have “View” -> “Display relationship is class hierarchy” checked then you’ll see something like this.


Cheers,

Matthew


On Mar 12, 2019, at 18:01, raza <[hidden email]> wrote:

Hi,
I'm new to ontologies and Protege. I am trying to solve the following
problem:

I have created simple hierarchy of body parts (using subclasses), and a
parallel one for skin of each body part. What is the correct way of linking
say the "skin of hand" to the body part "hand"? I've been reading about
class restrictions but feel that it's too cumbersome to define a separate
class restriction for each pair body part and it's associated skin.

Any help would be appreciated!

- U



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

raza
In reply to this post by Campbell, James R
Thank you James. I have explored FMA a while back. I believe that ontology
follows are very detailed anatomical scheme and also uses system-based
classification. My use case is more targeted towards dealing with how a lay
person talks about body parts, and I have to my surprise found no such
ontology out there. FMA is perhaps too complicated for this purpose and
orthogonal in some sense. But thanks for suggestion.

- Usman



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

raza
In reply to this post by Matthew Horridge-2
Thanks Mathew for the detailed explanation. I guess that's the way to go.

- Usman



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

jsluka
In reply to this post by raza
I agree that while FMA contains a huge amount of information it is so large
as to be very difficult to use. Indeed downloading FMA from BioPortal gives
a 250MB file that Protege (Protege_550b9) can't open.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Campbell, James R
It might be reasonable to subset FMA for your purposes rather than to embark on a one-off
Jim

-----Original Message-----
From: protege-user <[hidden email]> On Behalf Of jsluka
Sent: Wednesday, March 13, 2019 11:43 AM
To: [hidden email]
Subject: Re: [protege-user] Human body ontology

Non-UNMC email




I agree that while FMA contains a huge amount of information it is so large as to be very difficult to use. Indeed downloading FMA from BioPortal gives a 250MB file that Protege (Protege_550b9) can't open.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender.
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

brunil
Hi! 

May be this would be helpful. Take a look at body structures/ skin and related structure


Regards

Brunil


El mié., 13 mar. 2019 a las 18:34, Campbell, James R (<[hidden email]>) escribió:
It might be reasonable to subset FMA for your purposes rather than to embark on a one-off
Jim

-----Original Message-----
From: protege-user <[hidden email]> On Behalf Of jsluka
Sent: Wednesday, March 13, 2019 11:43 AM
To: [hidden email]
Subject: Re: [protege-user] Human body ontology

Non-UNMC email




I agree that while FMA contains a huge amount of information it is so large as to be very difficult to use. Indeed downloading FMA from BioPortal gives a 250MB file that Protege (Protege_550b9) can't open.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender.
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


--

Profa. Brunil D. Romero M.
Departamento de Procesos y Sistemas
Universidad Simón Bolívar
Caracas - Venezuela
MYS Piso 3, Oficina 309 - ext 3314
Tel: (+58) 412-5387805

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

jsluka
In reply to this post by Campbell, James R
James R.: Yes, absolutely, subset FMA, or at least use the subset as a
starting point and retain the links to the full FMA. Add whatever is needed,
which wouldn't necessarily have a corollary in FMA.

I looked at the WHO ICF browser and it doesn't have much detail. No term for
"skin of hand" and no terms for sub-parts of organs (like "lobule partOf
liver" or "nephron partOf kidney").



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

raza
Thanks for suggestions. The ICF browser is interesting even with less detail, but it doesn't link skin to the body parts, which is what I'm trying to do.

I've explored FMA using the browser, and it seems that it does have the right kind of content (although I've taken somewhat abstracted approach to naming parts given the different objective I have). Does anyone know how to get access to OWL files for FMA? I could not find that.

Another related question I have is, how to deal with scenarios where a patient points to a fuzzy area e.g. I currently have upper left abdomen, upper middle abdomen as two disjoint parts. If patient says, I have pain sort of to the right, I want to capture the data that it's upper-right but may also be upper-right + upper middle. Is that something to be captured outside of the ontology?

Thanks
- Usman

___________
Usman Raza
MBBS, MS, MIMS
Product Manager
HealthTap Inc.


On Thu, Mar 14, 2019 at 9:07 AM jsluka <[hidden email]> wrote:
James R.: Yes, absolutely, subset FMA, or at least use the subset as a
starting point and retain the links to the full FMA. Add whatever is needed,
which wouldn't necessarily have a corollary in FMA.

I looked at the WHO ICF browser and it doesn't have much detail. No term for
"skin of hand" and no terms for sub-parts of organs (like "lobule partOf
liver" or "nephron partOf kidney").



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user

The information in this message may be confidential and is intended only for the individual to whom it is addressed. If you are not the intended recipient of this message, you are hereby notified that any use, review, retransmission, dissemination, distribution, or reproduction of this communication is strictly prohibited. If this message was received in error, please immediately notify the sender and delete or destroy any copy of it.
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Csongor Nyulas
Administrator
In reply to this post by jsluka
It is not true that Protege can't open FMA!

I just downloaded the latest version 245MB. Gave Protege 5.5GB memory, close other applications so that I have enough free memory, and opened it in 35 seconds, without any issues.
(I was using the latest beta myself on an Ubuntu machine)



Csongor


On 3/13/19 10:43 AM, jsluka wrote:
I agree that while FMA contains a huge amount of information it is so large
as to be very difficult to use. Indeed downloading FMA from BioPortal gives
a 250MB file that Protege (Protege_550b9) can't open.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

jsluka
So an expert can get it to open. :)

When I tried Protege (Protege_550b9) hangs with no message about what the
problem is. Heap size is certainly the problem but if an ontology won't open
in the standard configuration of a program is that a concern?

5.5GB LOL, it is really a memory pig isn't it.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Csongor Nyulas
Administrator
On 3/14/19 12:47 PM, jsluka wrote:
So an expert can get it to open. :)

You do not need to be an expert. You just need to give Protege enough memory if you are working with huge ontologies. Thousands of Protege users have managed to do that.


When I tried Protege (Protege_550b9) hangs with no message about what the
problem is. Heap size is certainly the problem but if an ontology won't open
in the standard configuration of a program is that a concern? 

This is a very common problem, which was address countless times on our mailing list and is documented on our wiki. So, if someone has this problem, and looks for solution, he or she should find a solution within minutes. No need to be an expert.
We can't predict what ontologies will people try to open with Protege. Most people, I think, would not work with such huge ontologies from the start. They would first try to build or work with smaller ontologies. What would be a good default value for the heap size in your opinion? Should we aim that people be able to load FMA? Perhaps SNOMED? Or NCIt? Perhaps we should guess what is the biggest ontology in the world and set the heap size for that? Never mind that people may not have that amount of physical memory on their machine.


5.5GB LOL, it is really a memory pig isn't it.

My point earlier was that you can't just boldly state that Protege can't open FMA. Of course it can, if you give it enough resources. My goal was not to find out how much memory you need to give Protege to open FMA, so I just set a big enough amount that I could allocate on my machine. But since you were so shocked by the number, I did some experiments: I can open FMA easily with 3GB, 2GB and even with 1.5GB heap size. With 1GB it hangs, probably cycling between allocating memory while parsing and calling the garbage collector, until finally it exhaust all the available heap and will give an OutOfMemoryError (which it did for me after twenty minutes). So basically you need somewhere between 1GB and 1.5GB heap size, to open FMA. That seems totally reasonable to me, for such a large ontolgy. A good rule of thumb is that you should try to allocate a heap size of about 10 times the size of the OWL file, and go from there. Of course the amount of memory you really need depends on many factors, including the serialization of the ontology and the type of axioms contained.

Csongor







--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

jsluka
No, it really isn't all that easy. I'm aware of the heap size problem. But
since Protege crashed without any info on *why*, then increasing the heap
size is, at best, a WAG. Perhaps instead I should have rolled back Protege
to an earlier version, or done something else. Waiting for tens of minutes
to see if the program will eventually return an error code really isn't all
that practical.

/"Perhaps we should guess what is the biggest ontology in the world and set
the heap size for that?"/

Maybe. I've found that sarcasm is */always/* helpful.

Or perhaps something that is *doable*; such as checking the size of the
ontology file being imported (you know, maybe ask the operating system how
big the file is?) and giving an informative message that the heap size is
likely too small. Not all that hard really. You've even given a rule of
thumb that could be used. If this is a common problem, addressed "countless
times" then why not actually *fix* the problem?

Fact is I tried to open FMA with a vanilla install of Protege (a recent
version as well) and it died. The reason it died is irrelevant. Dead is
dead. FMA couldn't be opened.






--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Human body ontology

Matthew Horridge-2
Administrator
Adding a low memory warning is a good idea.  I’ve file an issue for this here:

https://github.com/protegeproject/protege/issues/861

Cheers,

Matthew


> On Mar 14, 2019, at 14:20, jsluka <[hidden email]> wrote:
>
> No, it really isn't all that easy. I'm aware of the heap size problem. But
> since Protege crashed without any info on *why*, then increasing the heap
> size is, at best, a WAG. Perhaps instead I should have rolled back Protege
> to an earlier version, or done something else. Waiting for tens of minutes
> to see if the program will eventually return an error code really isn't all
> that practical.
>
> /"Perhaps we should guess what is the biggest ontology in the world and set
> the heap size for that?"/
>
> Maybe. I've found that sarcasm is */always/* helpful.
>
> Or perhaps something that is *doable*; such as checking the size of the
> ontology file being imported (you know, maybe ask the operating system how
> big the file is?) and giving an informative message that the heap size is
> likely too small. Not all that hard really. You've even given a rule of
> thumb that could be used. If this is a common problem, addressed "countless
> times" then why not actually *fix* the problem?
>
> Fact is I tried to open FMA with a vanilla install of Protege (a recent
> version as well) and it died. The reason it died is irrelevant. Dead is
> dead. FMA couldn't be opened.
>
>
>
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> protege-user mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-user

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user