Workflow 3 Module 1 - Delineation of Symptom/Phenotype Polarity

[karafecho]

Module Overview This module focuses on polarity. Polarity refers to symptoms/phenotypes/adverse events that might be considered to be opposites (e.g., high vs low, absent vs present, normal vs abnormal). Much of medicine boils down to this general concept. Greek/Latin prefixes/suffixes/roots are useful in this regard and great for training the prototype Translator system (e.g., NLP), and developing ontologies, algorithms, etc.

Related Modules

Implementation TO BE ADDED AFTER BID MATRIX IS FINALIZED

Comment This ticket is for general considerations about Workflow 3, Module 1 as a whole. Separate tickets may be created for issues specific to a particular submodule.

[karafecho]

Thoughts from Jim Cimino (Alpha):

From: "Cimino, James" jamescimino@uabmc.edu To: 'William Byrd' webyrd@gmail.com; 'Chris Bizon' bizon@renci.org; 'David Koslicki' david.koslicki@math.oregonstate.edu; 'Greg Rosenblatt' greg.weiqi@gmail.com; "'Karamarie Fecho, PhD'" kfecho@copperlineprofessionalsolutions.com; "Carballo, Amanda" acarball@uabmc.edu Sent: Thursday, October 11, 2018 9:56 AM Subject: RE: [External Sender] UMLS, opposite terms, etc

William,

NLP lexicon’s might have such information, including the UMLS’s SPECIALIST lexicon.

In some cases, you could look for “hypo[x]” and “hyper[x]”, also “presence of [x]” (or simple (“[x]”) and “absence of [x]”

Is the hyponatremia the opposite of hypernatremia or is it no hyponatremia?

Then there is hypovolemia and euvolemia.

NegEx might be helpful as well.

Happy to talk – no time this week but Amanda Carballo (cc’ed) can set something up for next week.

--Jim

[karafecho]

Additional thoughts/feedback from Jim Cimino (Alpha):

There is a great community of scholars in informatics. See below.

Wendy Chapman told me that NegEx does not have antonyms but posted the question to an AMIA discussion board.

--Jim

Good to hear from you both.

At this point, the SPECIALIST Lexicon does not have antonym relations. However, Dina Demner-Fushman, who currently oversees the Lexicon (and MetaMap), says the addition of antonym information has been considered.

Instead of (or in addition to) lexical resources, I would suggest you consider logical definitions (i.e., semantics) as a possible source of information.

It looked promising in SNOMED CT, there 302215000 | Thrombocytopenic disorder (disorder) | is defined as:

=== 762656009 |Abnormal blood cell count (finding)| + 442686002 |Measurement finding below reference range (finding)| + 106200001 |Hematopoietic system finding (finding)| + 22716005 |Platelet disorder (disorder)| : { 363713009 |Has interpretation (attribute)| = 281300000 |Below reference range (qualifier value)|, 363714003 |Interprets (attribute)| = 61928009 |Platelet count (procedure)| }

We could easily find the “opposite” disorder by substituting Above reference range.

Unfortunately, no logical definition is provided for 234467004 | Thrombophilia (disorder) | in the current version of SNOMED CT.

This strategy might work in other cases, though (and will work better in the future as SNOMED CT provides makes an effort to adding definitions).

Other vocabularies based on DL could be considered as well, e.g., the NCI Thesaurus.

This could be a great project for a summer student.

Dr. Olivier Bodenreider

Senior Scientist

Acting Director, Lister Hill National Center for Biomedical Communications

National Library of Medicine, NIH

8600 Rockville Pike - MS 3828 (Bldg 38A, Rm 7N707)

Bethesda, MD 20894 - USA

office: 301 827-4982 - cell: 240 271-3195

From: McCray, Alexa T. Alexa_McCray@hms.harvard.edu Sent: Tuesday, November 20, 2018 07:59 To: Cimino, James jamescimino@uabmc.edu Cc: Bodenreider, Olivier (NIH/NLM/LHC) [E] obodenreider@mail.nih.gov Subject: Re: Question about SPECIALIST

Dear Jim,

I am forwarding your question to Olivier, who is now overseeing SPECIALIST. I see that there is also active discussion on the ACMI list!

Best,

Alexa

On Nov 19, 2018, at 4:44 PM, Cimino, James jamescimino@uabmc.edu wrote:

Alexa,

I am working with some folks (Matt Might and others) on the NCTAS Translator project and they are picking my brain about how to identify antonyms of adverse events (thrombocytopenia vs. thrombophilia). I was wondering if SPECIALIST has antonyms in it, and thought asking you might be the easiest way to find out that would also allow me to ask the more general question of whether you know of a good resource for such information.

--Jim

[cmungall]

We have curated opposites for HPO

https://www.biorxiv.org/content/early/2017/02/16/108977

Credit for this goes to @drseb (who is not on this project), but myself or @pnrobinson should be able to answer any questions.

HPO should have most of the adverse events you need, together with mappings to meddra. If any are missing you can let us know on: https://github.com/obophenotype/human-phenotype-ontology/issues/

[pnrobinson]

I am no longer on the translator project, but I please contact me about this because the plans mentioned above sound a little weird. For instance, there is no general connection between antonyms and adverse events, and adverse events are not generally "opposites", although of course there are many examples where this is the case.

[karafecho]

Thanks for your comments, Chris and Peter. Much appreciated!

A few of us previously explored the work you cite. We think this is great and completely aligned with the intent of Module 1, Workflow 3. We're wondering, however, how complete the "opposite-of" curation is for HPO.

I'm hoping to arrange a remote four-hour mini-hackathon to tackle this module. The current participant list includes Will Byrd, David Koslicki, Greg Rosenblatt, Chris Bizon, Mark Williams, and me. Any chance that you are willing to add your names to the participant list? No pressure, but I think it would be valuable to have you join the group.

[balhoff]

@karafecho the existing HPO opposites relations are loaded into our ubergraph; here's a query for them: http://yasgui.org/short/jJmMbPjgk

[karafecho]

Remote mini-hackathon is scheduled for 1/10, 1-5 pm ET.

Zoom connectivity: click herehttps://renci.zoom.us/j/522932773; Dial by your location +1 877 853 5257 US Toll-free +1 855 880 1246 US Toll-free Meeting ID: 522 932 773 Find your local number: https://zoom.us/u/aNHSbXuvK

What: Data Translator Mini-hackathon: Tackle approaches for Module 1 of Workflow 3https://drive.google.com/open?id=1LR0-Wf_iQRxOaIrV06sawxLBIA9nRPnSvmD_30Ew-1Y; When: Thur, Jan 10, 1-5pm EST Where: Zoom and RENCI-Sandb

[karafecho]

HPO opposites GitHub repo: https://github.com/cmungall/phenopposites

Related manuscript: https://www.biorxiv.org/content/early/2017/02/16/108977

HPO 'opposites' relations are loaded into Green/Gamma's UBERONGRAPH; here's a query for them: http://yasgui.org/short/jJmMbPjgk.

[karafecho]

@cbizon : Please add a link to your Jupyter exploration of the use of AE data (FAERS) to tackle module one. Thanks!

[cmungall]

That's my fork, I suggest using https://github.com/Phenomics/phenopposites

On Fri, Jan 4, 2019 at 1:48 PM karafecho notifications@github.com wrote:

@cbizon https://github.com/cbizon : Please add your Jupyter exploration of the use of AE data (FAERS) to tackle module one. Thanks!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ncats/translator-workflows/issues/8#issuecomment-451578243, or mute the thread https://github.com/notifications/unsubscribe-auth/AADGOZmxHVXK5iSQrJKruP95ibUVjkPrks5u_8wPgaJpZM4WiHO6 .