Mapping: considerations and feedback after first round of review

[sabrinatoro]

After reviewing the lexical matchings here, here are some considerations:

matching on synonyms

When it sounds like a good idea to match based on exact synonyms, this can present some issues:

• Some resources have “exact” synonyms which represent “terms included” though different from the concept. For examples: DOID:1602 (lymphadenitis) groups the concepts of 3 different NCIT (lymphadenitis, acute lymphadenitis, chronic lymphadenitis), and includes all the NCIT term names and synonyms as “exact synonyms”. In Mondo, these 3 NCIT would be represented as 3 different entries OR, if we don’t think that the specific concept should have their own entry, we would add the more specific concept (x-ref and synonyms) as “narrow” or “broad” (as it is appropriate).
• We decided in Mondo to separate concepts grouped in other sources, but not all sources did the same (for example, the “OMIM included” entries from the main OMIM entry). As a consequences, terms at a source might have exact synonyms that are represented in a term of different concept. Example:
  • OMIM:209950 = IMMUNODEFICIENCY 27A + “autosomal recessive Mendelian susceptibility to mycobacterial diseases due to partial IFNgammaR1 deficiency”
  • These concepts are separated in Mondo ( MONDO:0008856; MONDO:0020530; MONDO:0017901)
  • DOID:0111955 ('immunodeficiency 27A') includes all the terms (both disease and susceptibility)
  • → the lexical match suggests DOID:0111955 ('immunodeficiency 27A') should be a match to Mondo “'autosomal recessive Mendelian susceptibility to mycobacterial diseases due to partial IFNgammaR1 deficiency'
• We should ignore mapping to synonym with “abbreviation” type as it is not consistently correct.

Based on these observations, I don’t think that we can trust synonym matching, especially when trying to map sources that “include” other sources. These exact matches based on synonyms lex-matching should be manually reviewed by a curator before being accepted. Question: how should we keep track of the matching that were reviewed and are incorrect?

x-ref already in Mondo

(note that the following issues might have been resolved since the files reviewed were created. But since I have found examples, I am adding here. Maybe we can ignore until the next round and see if this is still a problem)

• Many x-refs were already a “MONDO:equivalent” in Mondo. The mappings suggested another mapping.
• Some x-ref were already in Mondo, but not as MONDO:equivalent. If x-refs exist for a Mondo term with a source that is not MONDO:equivalentTo, these x-refs should not be considered for exact match for these terms.

general questions

• How many terms in each resource do not have an exact mapping in Mondo? Do we have these numbers?
• From these terms with no exact matches, how many of the terms are in Mondo but with another type of match (narrow/broad/related)?
• Generally, we might want to get a strategy to deal with synonyms.

[matentzn]

This is a fabulous analysis, amazing amazing work! Thank you @sabrinatoro - we will discuss this at our next call. There is plenty of stuff @hrshdhgd and I have to fix based on your feedback, and I think we can address most, if not all, of your points. This is exactly what was needed!