search

obi-ontology / obi

The Ontology for Biomedical Investigations
http://obi-ontology.org
Creative Commons Attribution 4.0 International
75 stars 27 forks source link

chemical solution needs different ancestors in OBI #986

Closed gaviganet closed 5 years ago

gaviganet commented 5 years ago

The placement of chemical solution and all of its children under material entity in OBI seems misplaced and or problematic. Let me explain the chebi ontology places chemical and material entity at the same level. Chemical solution is not a direct descendant of chemical entity it requires several predecessors. I suggest that OBI use these predecessors. The flow would be: material entity to chemical substance (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A59999) to mixture (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A60004) to solution (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A75958 which is equivalent to chemical solution) . Under solution there could be phosphate-buffered saline etc..

If you now run the reasoner in Protege this arrangement does not provide for a "red flag" and seems to work.

thank you

Brenda

bpeters42 commented 5 years ago

Brenda, could you please explain what you consider problematic? Is there any wrong inference resulting from what OBI is doing? In general, we (and others) often import partial hierarchies from external ontologies, which typically reflects that as consumers we are only interested in a limited part. To give a different example, the NCBI taxonomy defines 40+ parent classes of 'homo sapiens' that have limited utility when not doing evolutionary biology. Importantly, unless we did something wrong, importing a subset of classes and the relationships between them should not lead to inconsistencies.

As a completely separate matter, we would be interested and excited to work with the Chebi team and others on a re-organization of material entities - there are different levels of granularity that need different treatments, and doing that consistently across OBO ontologies would be great.

On Mon, Nov 12, 2018 at 10:55 AM gaviganet notifications@github.com wrote:

The placement of chemical solution and all of its children under material entity in OBI seems misplaced and or problematic. Let me explain the chebi ontology places chemical and material entity at the same level. Chemical solution is not a direct descendant of chemical entity it requires several predecessors. I suggest that OBI use these predecessors. The flow would be: material entity to chemical substance ( https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A59999) to mixture (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A60004) to solution (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A75958 which is equivalent to chemical solution) . Under solution there could be phosphate-buffered saline etc..

If you now run the reasoner in Protege this arrangement does not provide for a "red flag" and seems to work.

thank you

Brenda

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9IqravVkSrx7UWaZ0JD-eeyY2xXu6ks5uucQXgaJpZM4YaPTr .

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

gaviganet commented 5 years ago

I am trained chemist but sorry I am not part of the CHEBI team. Let me try and explain.

When I take the current OBI (May 2018) and placed a new named solution for an assay as a child of chemical solution and run the HERMIT reasoner it delivered a red flag for chemical solution and all children underneath. Indeed defining solutions in OBI is not a trivial matter. Now this may be a problem with HERMIT (imperfect reasoner) or a problem with some logic used within OBI or a problem with me. I then looked closely at the CHEBI logic here which is quite clear as I see it, they have a large staff running this tool which is fantastic. Chemical entity and material entity (are equivalent) in the sense that all chemical entities will descend from chemical entity and all material (not just chemical) will descend from material entity. In order to catch those “chemical” entities (which maybe biological or clinical as well (e.g., mixtures of solutions or dispersions or colloidal solutions, for example “drinks” dispensed to patients before some scanning) then we need to be able to place the correct descendants in the path. I then added on my OWL file of OBI a new path: material entity to chemical substance to mixture to solution to (my solution). I then ran the HERMIT reasoner and I did not get a red flag which suggested to me that this maybe an improvement over the existing logic using HERMIT as the test.

As to your argument about the 40+ classes of homo sapiens I get it, but in this case there are only two descendant classes- chemical substance and mixture which is not a problem, and captures the essence of what bioassays may need the various different kinds of solutions or substances to perform assays. The same comparative logic could be undertaken for some other chemical-based classes in OBI like molecular entity, but this would need more work. I think it is doable. The hard bit is deciding upon the level of granularity needed as you state. In the solution case I think we need more or we ignore the HERMIT red flags. However, if we ignore them it is confusing for users like me who want to use OBI to describe their data, but do not know whether I am “breaking” rules or whether there are inherent inconsistencies in the OBI logic.

thank you

Brenda

On Mon, Nov 12, 2018 at 7:29 PM bpeters42 notifications@github.com wrote:

Brenda, could you please explain what you consider problematic? Is there any wrong inference resulting from what OBI is doing? In general, we (and others) often import partial hierarchies from external ontologies, which typically reflects that as consumers we are only interested in a limited part. To give a different example, the NCBI taxonomy defines 40+ parent classes of 'homo sapiens' that have limited utility when not doing evolutionary biology. Importantly, unless we did something wrong, importing a subset of classes and the relationships between them should not lead to inconsistencies.

As a completely separate matter, we would be interested and excited to work with the Chebi team and others on a re-organization of material entities - there are different levels of granularity that need different treatments, and doing that consistently across OBO ontologies would be great.

  • Bjoern

On Mon, Nov 12, 2018 at 10:55 AM gaviganet notifications@github.com wrote:

The placement of chemical solution and all of its children under material entity in OBI seems misplaced and or problematic. Let me explain the chebi ontology places chemical and material entity at the same level. Chemical solution is not a direct descendant of chemical entity it requires several predecessors. I suggest that OBI use these predecessors. The flow would be: material entity to chemical substance ( https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A59999) to mixture (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A60004) to solution (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A75958 which is equivalent to chemical solution) . Under solution there could be phosphate-buffered saline etc..

If you now run the reasoner in Protege this arrangement does not provide for a "red flag" and seems to work.

thank you

Brenda

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986, or mute the thread < https://github.com/notifications/unsubscribe-auth/ANN9IqravVkSrx7UWaZ0JD-eeyY2xXu6ks5uucQXgaJpZM4YaPTr

.

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986#issuecomment-438089321, or mute the thread https://github.com/notifications/unsubscribe-auth/AcUQ2t4j1Vnu35DjFtQPMFQeeXzoJKfRks5uuhsJgaJpZM4YaPTr .

bpeters42 commented 5 years ago

Can you please provide a bit more detail what exactly you did for the following: " When I take the current OBI (May 2018) and placed a new named solution for an assay as a child of chemical solution and run the HERMIT reasoner it delivered a red flag for chemical solution and all children underneath." Specifically, what was the definition of the solution? There should not be a logical inconsistency based on what you describe.

I gave the 40 classes from NCBI taxonomy as an example; we definitely also have a problem with 'mixture' or 'chemical solution', because in the general world of things we are trying to describe, these are very broadly defined, and it is not clear if (for example) a 'cell' is a 'mixture' (defined as : " A mixture is a chemical substance composed of multiple molecules, at least two of which are of a different kind. " ).

On Tue, Nov 13, 2018 at 8:46 AM gaviganet notifications@github.com wrote:

I am trained chemist but sorry I am not part of the CHEBI team. Let me try and explain.

When I take the current OBI (May 2018) and placed a new named solution for an assay as a child of chemical solution and run the HERMIT reasoner it delivered a red flag for chemical solution and all children underneath. Indeed defining solutions in OBI is not a trivial matter. Now this may be a problem with HERMIT (imperfect reasoner) or a problem with some logic used within OBI or a problem with me. I then looked closely at the CHEBI logic here which is quite clear as I see it, they have a large staff running this tool which is fantastic. Chemical entity and material entity (are equivalent) in the sense that all chemical entities will descend from chemical entity and all material (not just chemical) will descend from material entity. In order to catch those “chemical” entities (which maybe biological or clinical as well (e.g., mixtures of solutions or dispersions or colloidal solutions, for example “drinks” dispensed to patients before some scanning) then we need to be able to place the correct descendants in the path. I then added on my OWL file of OBI a new path: material entity to chemical substance to mixture to solution to (my solution). I then ran the HERMIT reasoner and I did not get a red flag which suggested to me that this maybe an improvement over the existing logic using HERMIT as the test.

As to your argument about the 40+ classes of homo sapiens I get it, but in this case there are only two descendant classes- chemical substance and mixture which is not a problem, and captures the essence of what bioassays may need the various different kinds of solutions or substances to perform assays. The same comparative logic could be undertaken for some other chemical-based classes in OBI like molecular entity, but this would need more work. I think it is doable. The hard bit is deciding upon the level of granularity needed as you state. In the solution case I think we need more or we ignore the HERMIT red flags. However, if we ignore them it is confusing for users like me who want to use OBI to describe their data, but do not know whether I am “breaking” rules or whether there are inherent inconsistencies in the OBI logic.

thank you

Brenda

On Mon, Nov 12, 2018 at 7:29 PM bpeters42 notifications@github.com wrote:

Brenda, could you please explain what you consider problematic? Is there any wrong inference resulting from what OBI is doing? In general, we (and others) often import partial hierarchies from external ontologies, which typically reflects that as consumers we are only interested in a limited part. To give a different example, the NCBI taxonomy defines 40+ parent classes of 'homo sapiens' that have limited utility when not doing evolutionary biology. Importantly, unless we did something wrong, importing a subset of classes and the relationships between them should not lead to inconsistencies.

As a completely separate matter, we would be interested and excited to work with the Chebi team and others on a re-organization of material entities

there are different levels of granularity that need different treatments, and doing that consistently across OBO ontologies would be great.

  • Bjoern

On Mon, Nov 12, 2018 at 10:55 AM gaviganet notifications@github.com wrote:

The placement of chemical solution and all of its children under material entity in OBI seems misplaced and or problematic. Let me explain the chebi ontology places chemical and material entity at the same level. Chemical solution is not a direct descendant of chemical entity it requires several predecessors. I suggest that OBI use these predecessors. The flow would be: material entity to chemical substance ( https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A59999) to mixture (https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A60004) to solution ( https://www.ebi.ac.uk/chebi/searchId.do?chebiId=CHEBI%3A75958 which is equivalent to chemical solution) . Under solution there could be phosphate-buffered saline etc..

If you now run the reasoner in Protege this arrangement does not provide for a "red flag" and seems to work.

thank you

Brenda

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986, or mute the thread <

https://github.com/notifications/unsubscribe-auth/ANN9IqravVkSrx7UWaZ0JD-eeyY2xXu6ks5uucQXgaJpZM4YaPTr

.

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986#issuecomment-438089321, or mute the thread < https://github.com/notifications/unsubscribe-auth/AcUQ2t4j1Vnu35DjFtQPMFQeeXzoJKfRks5uuhsJgaJpZM4YaPTr

.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986#issuecomment-438343786, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9IivrBjiHf02z1oDkGE-DLR4bxB8gks5uuvdLgaJpZM4YaPTr .

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

gaviganet commented 5 years ago

  1. I load OBI (May 2018) in Protege on my windows desktop (protege version 5.2.0). When I run the Hermit reasoner several classes come-up with red flags one of which is chemical solution and phosphate buffered saline. If I add a second child to chemical solution (i.e., sister of PBS) e.g. extracellular solution. It also produces a red flag. How do I describe my solutions in OBI used during my assay? Red flag existed without my definition.

  2. I then tried to use CHEBI logic as described. Here I have material entity, chemical substance, mixture, solution, extracellular solution and solution in patch pipette (the last two are my descriptions and siblings) and I repeated the reasoner and I do not detect any red flags. This is when I sent the query. The definitions for solution: A mixture that is homogeneous, made up of at least two scattered molecular aggregates, one playing the role of solute and the other playing the role of solvent (CHEBI) same as chemical solution.. mixture: A mixture is a chemical substance composed of multiple molecules, at least two of which are of a different kind. A chemical substance is a portion of matter of constant composition, composed of molecular entities of the same type or of different types. Now if U examine definition of chemical substance a cell is not a chemical substance as it is not of "constant composition" and hence 'cell" cannot be a mixture or chemical substance. (This of course does not make it full proof).

My definitions: extracellular solution: The solution usually aqueous and buffered that is used to bathe the cells during an intracellular electrophysiology recording. solution in patch pipette: The solution usually aqueous and buffered that is used to back-fill the patch pipette during an intracellular electrophysiology recording.

(Here my definitions still need vetting).

Is this RED FLAG problem just a algorithm glitch then let me know others have mentioned that solutions are difficult to describe.

-thank you

Brenda

bpeters42 commented 5 years ago

I am not sure what is/was going on, but when I download the current version of OBI from http://purl.obolibrary.org/obo/obi.owl (which is from August), and run the current Hermit reasoner that comes with Protege 5.5, I don't get any 'red flags'. So it would be helpful to know what exactly you ran. We do include checking by reasoners in our release process, so I while I didn't dig up the May version, I would be very surprised if there was a problem in the May version that we did not detect, and if so, it would be helpful to know what version of Hermit / Protege / etc. you were running.

I also saw that you closed the ticket - were you saying the problem has been resolved?

On Tue, Nov 13, 2018 at 2:02 PM gaviganet notifications@github.com wrote:

Closed #986 https://github.com/obi-ontology/obi/issues/986.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/obi-ontology/obi/issues/986#event-1964358297, or mute the thread https://github.com/notifications/unsubscribe-auth/ANN9IkwsDFaVLk3WxsnVEfSaMSRudO06ks5uu0GMgaJpZM4YaPTr .

-- Bjoern Peters Professor La Jolla Institute for Allergy and Immunology 9420 Athena Circle La Jolla, CA 92037, USA Tel: 858/752-6914 Fax: 858/752-6987 http://www.liai.org/pages/faculty-peters

gaviganet commented 5 years ago

Sorry I did not mean to close the ticket. I hit the wrong thread when replying. I will download latest version of OBI and repeat this and get back to you. I ran protege on my windows desktop. Protege v. 5.5 with Hermit reasoner.

-thank you

Brenda

gaviganet commented 5 years ago

I ran the latest version (August) from URL on Protege and do not have problem. There were no red flags as you expected obviously something happened when I downloaded it and ran the reasoner very strange. Anyway if I find out I will let you know. I will close this ticket. -thank you Brenda