Reviewing disease mappings for Biocurators and Clinicians

Description

This guide provides guidelines on how to rapidly review large scale efforts to create mappings between ontology classes.

Mapping can be created using tools like OAK to generate a bunch of mapping candidates. A reviewer then needs to determines whether the mapping is correct or not.

Refer to Are these two entities the same? A guide. for a comprehensive guide on determining a specific mapping.

Guidelines

1. Depending on the format of the table you are reviewing, it is recommended that you open the table in a google spreadsheet, to allow for collaborative work and to allow yourself to add notes in a new column, or add formatting or color coding.
2. Hide any columns that are not relevant, to make it easier to quickly compare the mappings. For example, in the table below, you may want to hide the mapping_justification column, mapping_tool. You can also move the columns around so the labels are next to each other.
3. Look at the confidence score for the mapping and look at the lower confidence scores first, as these are more likely to have issues.
4. Review the label of the mappings. Note, matching on the label is not always sufficient to conclude two terms are equivalent. Ideally, reviewers should compare the definitions of each term to ensure they have the same meaning.
5. Review how the match was made. If it is a lexical match, matching on labels, it is likely the mapping is correct but this should ideally be confirmed by reviewing the definitions as mentioned above.
6. If matches are made based on shared xrefs, these should be carefully reviewed because not all mapping between terminologies are intended to be equivalent xrefs. For example: MONDO:0000509 non-syndromic intellectual disability was mapped to DOID:0081098 autosomal recessive intellectual developmental disorder 13 based on the shared xref OMIM:613192. The Mondo class is broader than this DO class and this is not an exact mapping.
7. Watch out for matching on acronyms. Acronyms can mean a lot of different things and sometimes the mapping tools will incorrectly match on acronyms.

Example mappings

subject_id	subject_label	predicate_id	object_id	object_label	mapping_justification	mapping_tool	confidence	subject_match_field	object_match_field	match_string	comment
ID		A oboInOwl:hasDbXref	>A oboInOwl:source	>A sssom:object_label
MONDO:0000159	bone marrow failure syndrome	MONDO:equivalentTo	NCIT:C165614	Bone Marrow Failure Syndrome	semapv:LexicalMatching	oaklib	0.849778895	rdfs:label	rdfs:label	bone marrow failure syndrome	LEXMATCH
MONDO:0000376	respiratory system cancer	MONDO:equivalentTo	NCIT:C4571	Malignant Respiratory System Neoplasm	semapv:LexicalMatching	oaklib	0.8	oio:hasExactSynonym	rdfs:label	malignant respiratory system neoplasm	LEXMATCH
MONDO:0000437	cerebellar ataxia	MONDO:equivalentTo	NCIT:C26702	Ataxia	semapv:LexicalMatching	oaklib	0.8	oio:hasExactSynonym	rdfs:label	ataxia	LEXMATCH
MONDO:0000541	jejunal adenocarcinoma	MONDO:equivalentTo	NCIT:C181158	Jejunal Adenocarcinoma	semapv:LexicalMatching	oaklib	0.849778895	rdfs:label	rdfs:label	jejunal adenocarcinoma	LEXMATCH
MONDO:0000543	ovarian melanoma	MONDO:equivalentTo	NCIT:C178441	Ovarian Melanoma	semapv:LexicalMatching	oaklib	0.849778895	rdfs:label	rdfs:label	ovarian melanoma	LEXMATCH
MONDO:0000665	apraxia	MONDO:equivalentTo	NCIT:C180557	Apraxia	semapv:LexicalMatching	oaklib	0.849778895	rdfs:label	rdfs:label	apraxia	LEXMATCH
MONDO:0000705	Clostridium difficile colitis	MONDO:equivalentTo	NCIT:C180523	Clostridium difficile Infection	semapv:LexicalMatching	oaklib	0.8	oio:hasExactSynonym	rdfs:label	clostridium difficile infection	LEXMATCH
MONDO:0000736	dyschromatosis universalis hereditaria	MONDO:equivalentTo	NCIT:C173131	Dyschromatosis Universalis Hereditaria	semapv:LexicalMatching	oaklib	0.849778895	rdfs:label	rdfs:label	dyschromatosis universalis hereditaria	LEXMATCH