Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Use cases for NBO #127

Open
DitchingIt opened this issue Feb 9, 2023 · 32 comments
Open

Use cases for NBO #127

DitchingIt opened this issue Feb 9, 2023 · 32 comments
Assignees
@DitchingIt
Labels
Structure

Comments

@DitchingIt
Copy link
Collaborator

Background

We know some of the original aims for merging the Animal Behavior Ontology (ABO) with the NBO, but things seem to have come off the rails: Leadership has evaporated, there is no paid curation, just occasional 'hacking', some of those originally involved in ABO have gone back to using it 'off-line' instead of NBO, and others who once appeared commited appear to have given up on it. @matentzn is working on a process which should lead to the deprecation of the phenotype branch from NBO. Being realistic, NBO is almost defunct - indeed @cmungall has suggested it might be time to call time on NBO. I remain hopeful that something viable can emerge from NBO.

What next?

@dosumis has suggested collecting use cases before a radical overhaul of NBO. I agree with a proviso: Nothing theoretical and no direct return to the original aims, because they have essentially failed. Starting afresh, what would any potential user realistically want from NBO?

From discussions I have been part of since December 2022, the only real hope I have seen expressed by anyone has been along the lines of:

  1. Ethology - ABO approach

Is there any other use case that anyone feels passionate about - not just hope that someone else would do, or possibly use if it was there? Because I think that there would be commitment to work on an ethology-based ontology, IF it had much of the purely human or invisible stuff stripped out.

Any thoughts?
@DitchingIt DitchingIt pinned this issue Feb 9, 2023
@dosumis
Copy link
Collaborator

dosumis commented Feb 9, 2023

I suspect that the ABO approach may also be useful for annotating observational behavioural data such as FlyBowl home cage experiments (e.g. https://www.noldus.com/blog/high-throughput-method-natural-behavior-mice). I planned to use it for FlyBowl annotation as part of Virtual Fly Brain but some of the required collaborative effort broke down.

Use case 2: Behavioural studies of model organisms centred on behavioural testing paradigms.

  • see use of NBO and related GO terms in model organism phenotype ontologies. We can generate a report of these. NBO is doing dome useful work here. We should not break this. Lots of potentially relevant refs from usage of these terms in annotation.

Use case 3: Clinical

The only use I'm aware of is in the Human Phenotype Ontology (but there may be others). We can get a report of these.

The promise of an ontology like this is that it can bridge between animal and human data. It does some useful work under use cases 2 and 3 which we should not break. The merge with ABO was never done properly. I think it could be done without breaking use cases 2 and 3. That seems like a reasonable aim.

In the end, fixing it more comprehensively will need proper funding.

@DitchingIt
Copy link
Collaborator Author

Thanks @dosumis

We can generate a report of these.

How do we generate these two suggested reports (model organism phenotype, clinical)?

Lots of potentially relevant refs from usage of these terms in annotation.

Would these reports show usage in annotation?

@dosumis
Copy link
Collaborator

dosumis commented Feb 10, 2023

Would these reports show usage in annotation?

We can do that.

@matentzn - my first thought was to use an UberGraph query to find usage in phenotype ontology terms, but thinking about it, we should be able to get that + usage in annotation --> Pubs from a query of the Monarch graph. Can you do that - if not who should we ask?

@matentzn
Copy link
Collaborator

I asked in slack how we could go about this, but in the meantime, here is a draft query you can work with:

Note NBO is not in Ubergraph, but I have requested to add it here: INCATools/ubergraph#118

@matentzn
Copy link
Collaborator

Sorry, I forgot the first query cannot work because ubergraph does not have NBO. Use this instead:

https://api.triplydb.com/s/8ewjIiZ3g

You can download the results as a TSV file.

@DitchingIt
Copy link
Collaborator Author

DitchingIt commented Feb 10, 2023 •
edited

You can download the results as a TSV file.

@matentzn @dosumis This is great! It's a big missing piece of jigsaw and helps me understand better what will break if/when we try to change things. But before I do it manually:

  1. is there a way to separate these easily into phenotype branch and process branch queries?
  2. and if possible, bring up the NBO behaviour labels as well as the IDs?

@matentzn
Copy link
Collaborator

It is very easy to separate these and add labels, but only after INCATools/ubergraph#118 is merged and Ubergraph rebuilt. This may take up to a week, but usually less.

@matentzn matentzn mentioned this issue Feb 10, 2023
Merged
@DitchingIt
Copy link
Collaborator Author

DitchingIt commented Feb 10, 2023 •
edited

Summary report stats

NBO has 929 of its own classes. Of these:

  • Five ontologies (FBCV, HP, MP, OBA, WBP) make use of 187 NBO classes
    • These are used 3,714 times
    • 2%, 45%, 46%, 3%, 5% respectively
  • 322 are in the phenotype branch (35%)
    • 23 of these are used by other ontologies (it has never been their intended role)
      • HP has used 19 of them 98 times
      • MP has used 6 of them 33 times
    • Moves are under way to deprecate all of them from NBO
  • 607 are in the process branch (65%)
    • One of these (agoraphobia) is misplaced and (correctly) duplicated in the phenotype branch connection between behavior process and phenotype #36
    • 164 are used in the other five ontologies (27% of the process branch)
      • These are used 3,582 times
      • The root behavior process NBO:0000313 superclass represents 37% of these uses and is the only class used by all five
      • Voluntary movement behavior NBO:0000403 is next most used (15% of the remainder), followed by body part movement NBO:0000001 (9%), and eye movement NBO:0000444 (8%)
    • 73% of the process branch's classes are unused by other ontologies

@DitchingIt
Copy link
Collaborator Author

DitchingIt commented Feb 11, 2023 •
edited

The merge with ABO was never done properly.

@dosumis can you outline the key things which need to be done properly, or are they all captured in the open Issues? This may be something I can give time to.

@pmidford
Copy link
Collaborator

@DitchingIt, You should definitely check with @aclark-binghamton-edu about the ABO merge. We had started an incremental merge, starting with adding some terms and definitions, but there were some major issues related to agonism and aggression that Anne was concerned about but the previous owners didn't want to budge on.

@DitchingIt
Copy link
Collaborator Author

@aclark-binghamton-edu I'd really appreciate some more information about where the ABO merge seems to have come unstuck.

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 11, 2023 via email

@DitchingIt
Copy link
Collaborator Author

I am happy to share and explain my spreadsheets and ABO with its definitions.

Yes please - that would be great.

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 11, 2023 via email

@DitchingIt
Copy link
Collaborator Author

Give me until tomorrow and I will start!

No rush! Take your time.

@DitchingIt DitchingIt mentioned this issue Feb 11, 2023
Closed
17 tasks
@dosumis
Copy link
Collaborator

dosumis commented Feb 12, 2023 •
edited

Really great to see progress here. @DitchingIt - the summary stats look great. (@cmungall - do you think we could go further and use the Monarch graph to pull back some references for usage of phenotype terms?)

My main concern is that we avoid churn for the phenotype ontologies already using NBO. I worry that a focus on upper ontology issues will be a major cause of this - Phenotypes and functions are out-of-scope; x is a phenotype! Y is a function! They must go! But I also think that looking at actual use cases and using some standard patterns to cross between upper ontology branches can help.

My sense is that phenotype ontologies need terms that refer to behaviours, but sometimes these need to be abstracted to general behaviour terms that group behaviours with a particular function or cause. Additional NBO branches could help here (e.g. a separate behavioural function branch as in the original ABO - see below) as we could use them compositionally - we just need to be very clear about where/when they should be used. The write up from the 2016 meeting has some potential compositional patterns.

ABO top level

image

Practically - I'd say be very conservative about obsoleting terms - especially those in use - and try to find practical solutions for upper ontology issues using patterns to bridge between branches. Concentrate on improving the hierarchy (this may mean some flattening and we should have reports for cases where hierarchy that is being used by Phenotype ontologies is changed). I've added @rays22 and @sbello to this repo so that they can give feedback on any proposed changes that affect phenotype ontologies. @rays22 can potentially help with patterns & their implementation.

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 12, 2023 via email

@DitchingIt
Copy link
Collaborator Author

@aclark-binghamton-edu I can't see the attachments, which sound tantalising...

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 12, 2023 via email

@DitchingIt
Copy link
Collaborator Author

Let me go into GitHub and see how to pass them on

There are some options offered at the bottom of a new comment box.

@rays22
Copy link
Collaborator

rays22 commented Feb 17, 2023

NBO is almost defunct - indeed @cmungall has suggested it might be time to call time on NBO. I remain hopeful that something viable can emerge from NBO.

@DitchingIt , I do not think @cmungall meant a call an end to NBO. In my reading, he just meant that it should be transparent on the OBO site that NBO development has been inactive for a while. However, I may have misunderstood you both.

Regarding use cases, I think NBO is/would be a good place for
GO:0007610 behavior-type biological processes that may not be in scope for the Gene Ontology.
An up-to-date NBO:0000313 behavior process branch would be very useful for modelling traits and phenotypes.
This is not a radical overhaul use case, but there is plenty of room for useful improvement.

@DitchingIt
Copy link
Collaborator Author

There are some options offered at the bottom of a new comment box.

@aclark-binghamton-edu I am very much looking forward to seeing these documents :-)

@DitchingIt
Copy link
Collaborator Author

I do not think @cmungall meant a call an end to NBO.

@rays22 Thankyou for clarifying: I should have used the the term 'implied' instead of 'suggested'.

On that score, there is now a Team working on behaviour ontologies (particularly NBO) whose membership is currently open. If you are interested in supporting 'useful improvement', please fire off a request to @matentzn to join it.

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 18, 2023 via email

@DitchingIt DitchingIt mentioned this issue Feb 21, 2023
Closed
2 tasks
@aclark-binghamton-edu
Copy link

@DitchingIt Working files, including an old pdf of ABO hierarchy and updates of NBO terms that @pmidford looked at and may have been made.
ABO_hierarchy.pdf
NBO_definition_updates6 Aug17PM.xlsx
NBO-ABO merging Hunger-Nutrients per-ABCmodified28July17copy2.pdf
NBO-ABO allignment updated 7-29-17-Download15Oct19.xlsx

@aclark-binghamton-edu
Copy link

@DitchingIt One last excel file as I was coming to end of what I thought I could do.
NBO Flat List of Terms with ABO synonyms-ABC-3Sep17B.xlsx

@aclark-binghamton-edu
Copy link

@DitchingIt I foresee that some attachments will be confusing! I will try to interpret their whys and wherefores. Some of changes in definitions specified as needed may well have been done. I did not keep up after about Sept 2017. Thanks for your efforts!

@DitchingIt
Copy link
Collaborator Author

Thanks for your efforts!

@aclark-binghamton-edu Not a problem - I'm enjoying it :-)

This is not my feedback yet, but just adding one missing document from your uploads at #136 to this conversation so we can all start from the same place: ABO leaf terms-add to NBO.docx

@aclark-binghamton-edu
Copy link

aclark-binghamton-edu commented Feb 22, 2023 via email

@DitchingIt
Copy link
Collaborator Author

@aclark-binghamton-edu It's fine; Nico is tutoring me in some basic curation tasks so it's not like I am very far ahead of you on the learning curve :-)

@DitchingIt DitchingIt mentioned this issue Feb 24, 2023
Open
@DitchingIt DitchingIt self-assigned this Mar 2, 2023
@DitchingIt DitchingIt unpinned this issue Jul 4, 2023
@DitchingIt DitchingIt mentioned this issue Jul 23, 2023
Open
@DitchingIt
Copy link
Collaborator Author

DitchingIt commented Aug 7, 2023 •
edited

Embedded is my proposal for editing a section of the NBO I'm calling the Ethology module. I hope to start submitting some significant pull requests from mid-September.
Feedback will be much appreciated.
Ethology module.pdf
@dosumis @pmidford @aclark-binghamton-edu @jannahastings @matentzn

@DitchingIt DitchingIt mentioned this issue Aug 8, 2023
Closed
@DitchingIt DitchingIt mentioned this issue Oct 9, 2023
Open
@DitchingIt DitchingIt mentioned this issue Nov 30, 2023
Open
@DitchingIt
Copy link
Collaborator Author

My attempts over the last year to change the NBO have not made enough progress to justify pursuing it further.

Unlike anatomy or biochemistry, I doubt behaviour above the morpho-physiological level will (in the near future at least) be more than a set of more or less disputed opinions, although doubtless there will be individuals or 'schools' who believe in the universality of their own systematics.

Perhaps a more viable approach to cataloguing behaviour would be to map these opinions as a knowledge graph of behavioural assertions (something I have already begun to explore) rather than to think there is a universally acceptable set of fixed behavioural entities which can be pinned down in a fixed ontology.

Even if I am wrong, at the very least, involvement, let alone consensus, has virtually disappeared over the last 12 years from the construction of the NBO. I now support @cmungall in his suggestion that the NBO be labelled functionally inactive, and declared as such on the OBO site (see#126).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@DitchingIt DitchingIt

Labels
Structure
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
6 participants
@dosumis @pmidford @matentzn @aclark-binghamton-edu @rays22 @DitchingIt