paranasal sinus cancer vs adenocarcinoma

[cmungall]

is the EXACT nature of all syns justified? especially the highlighted

[Term]
id: DOID:0050619
name: paranasal sinus cancer
namespace: disease_ontology
alt_id: DOID:1354
alt_id: DOID:2442
alt_id: DOID:2765
alt_id: DOID:2767
alt_id: DOID:4689
def: "A respiratory system cancer that is located_in the paranasal sinuses." [url:http://www.cancer.gov/cancertopics/pdq/treatment/paranasalsinus/Patient/page1]
synonym: "Epidermoid carcinoma of the paranasal sinus" EXACT [NCI2004_11_17:C8193]
synonym: "Mucoepidermoid carcinoma of Accessory sinus" EXACT [NCI2004_11_17:C6018]
synonym: "adenoid cystic carcinoma of Accessory sinus" EXACT [NCI2004_11_17:C6019]
synonym: "adenoid cystic carcinoma of paranasal sinus" RELATED []
***synonym: "paranasal sinus adenocarcinoma" EXACT []
synonym: "paranasal sinus adenoid cystic carcinoma" EXACT []
synonym: "paranasal sinus mucoepidermoid carcinoma" EXACT []
synonym: "paranasal sinus squamous cell carcinoma" EXACT []
synonym: "squamous cell carcinoma of paranasal sinus" RELATED []
xref: NCI:C8193
xref: UMLS_CUI:C0280334
xref: UMLS_CUI:C0854995
xref: UMLS_CUI:C1335336
xref: UMLS_CUI:C1335337
xref: UMLS_CUI:C1335340
is_a: DOID:0050615  ! respiratory system cancer

If it's always adenocarcinoma/glandular, why are there distinct classes for ethmoid sinus cancer vs adenocarcinoma?

[wakibbe]

This is a question of granularity. Do we want to represent subtypes of paranasal sinus cancer? If we are not representing adenocarcinoma separately from squamous then we can point them all that paranasal sinus cancer. However, I believe we do distinguish ‘tissue of origin’ in Disease Ontology, and since the outcomes are different, I believe that we want to separate them.

Interestingly, "paranasal sinus squamous cell carcinoma” is an exact synonym for "squamous cell carcinoma of paranasal sinus”, not related.

On Apr 7, 2015, at 2:10 AM, Chris Mungall [email protected] wrote:

is the EXACT nature of all syns justified? especially the highlighted

[Term]
id: DOID:0050619
name: paranasal sinus cancer
namespace: disease_ontology
alt_id: DOID:1354
alt_id: DOID:2442
alt_id: DOID:2765
alt_id: DOID:2767
alt_id: DOID:4689
def: "A respiratory system cancer that is located_in the paranasal sinuses." [url:http://www.cancer.gov/cancertopics/pdq/treatment/paranasalsinus/Patient/page1]
synonym: "Epidermoid carcinoma of the paranasal sinus" EXACT [NCI2004_11_17:C8193]
synonym: "Mucoepidermoid carcinoma of Accessory sinus" EXACT [NCI2004_11_17:C6018]
synonym: "adenoid cystic carcinoma of Accessory sinus" EXACT [NCI2004_11_17:C6019]
synonym: "adenoid cystic carcinoma of paranasal sinus" RELATED []
***synonym: "paranasal sinus adenocarcinoma" EXACT []
synonym: "paranasal sinus adenoid cystic carcinoma" EXACT []
synonym: "paranasal sinus mucoepidermoid carcinoma" EXACT []
synonym: "paranasal sinus squamous cell carcinoma" EXACT []
synonym: "squamous cell carcinoma of paranasal sinus" RELATED []
xref: NCI:C8193
xref: UMLS_CUI:C0280334
xref: UMLS_CUI:C0854995
xref: UMLS_CUI:C1335336
xref: UMLS_CUI:C1335337
xref: UMLS_CUI:C1335340
is_a: DOID:0050615 ! respiratory system cancer
If it's always adenocarcinoma/glandular, why are there distinct classes for ethmoid sinus cancer vs adenocarcinoma?

—
Reply to this email directly or view it on GitHub #35.

[cmungall]

It seems useful to be granular here. I found this high level description useful http://www.cancer.org/acs/groups/cid/documents/webcontent/003123-pdf.pdf

Design patterns here should probably be aligned with NCIt

But I see this sort of thing both in the cancer subset of DO and outside. There is ambiguity in the terms, with graph position, exact synonyms and textual definition not quite agreeing about the level of specificity. Having general groupings is fine, but it should be clear if the intent is to have a general grouping, or something else.

[cmungall]

Just for comparison, a bit of NCIt:

NCIt makes granular distinctions.

I find the equivalence axiomatization pattern in NCIt unusual though. If we are to follow other OBO ontologies we would have a simple genus-differentia axiom, with specific patterns set up for branches like cancer. The disease findings would go into a separate subClassOf R some axiom (hidden GCI) rather than overloading the definition (which inhibits useful auto-classification).