Map ClinVar significance terms to numbers

[moonso]

This is a suggested mapping, please give your opinion if you have any other idea @henrikstranneheim @dnil . It would be nice to see what mapping you use in Lund @bjhall

Term	Number
Pathogenic	5
Likely pathogenic	4
Conflicting_interpretations_of_pathogenecity	3
Uncertain significance	0
likely benign	2
benign	1
affects	20
association	21
drug response	22
other	23
protective	24
risk_factor	25
association_not_found	26
drug_response	27
not_provided	100

[moonso]

And similar with the review status:

Term	Number
reviewed_by_expert_panel	9
practice_guidelines	8
criteria_provided	7
multiple_submitters	6
no_conflicts	5
conflicting_interpretations	4
single_submitter	3
no_assertion_criteria_provided	2
no_assertion_provided	1

[dnil]

Looks good to me: Swap conflicting and Uncertain significance,

[moonso]

What do you mean @dnil ? They are in different tables...

[moonso]

Aha you mean 3 and 4 in the first table! What numbers should we use on those?

[dnil]

At a meeting..

Conflicting_interpretations_of_pathogenecity | 3 Uncertain significance | 0

is weird. Uncertain (VUS) is 3, conflict is typically 1.5, 4.5 or possibly 2.5 or 3.5.

[moonso]

Is it important to use same numbers like acmg? Or could we just assign any number. So least interesting would be 1 and pathogenic 12 or something? Otherwise I suggest this:

Term	Number
pathogenic	15
likely_pathogenic	14
uncertain_significance	13
conflicting_interpretations_of_pathogenecity	12
likely_benign	11
benign	10
affects	9
association	8
drug_response	7
other	6
protective	5
risk_factor	4
association_not_found	3
drug_response	2
not_provided	1

In this way it might be easier to ask the database if a variant has climbed the hierarchy between ClinVar releases.(Suggestion from @henrikstranneheim )

[moonso]

Ok here is the final solution that I have chosen. It has the advantage that it will work with existing values in the database.

Term	Number
other	255
not_provided	1
drug_response	6
association_not_found	13
risk_factor	12
protective	11
association	10
affects	9
histocompatibility	7
benign	2
likely_benign	3
conflicting_interpretations_of_pathogenicity	8
uncertain_significance	0
likely_pathogenic	4
pathogenic	5

[henrikstranneheim]

Looks good! It is nice for now to have backwards compatibility. We can then provide another map of how each term should be viewed in clinical significance when updating clinvar and recalculating the status for each variant i.e. if they moved up or down in pathogenicity.