support multiple pubmed IDs for a single evidence record

[malachig]

Use case from the VHL project. A group reports on 10 patients with VHL variants. A year later the same 10 patients plus 3 more appear in an update paper. There are some additional details on the same patients from this paper. We don't want to create a redundant record for this case/variant. But we want a record of duplicate reports in the published literature on what is essentially the exact same data.

[kkrysiak]

Some questions about implementation. How best to do this:

• show both papers at the evidence level?
• show links between publications at the publication level?
• have a primary and secondary source at the evidence level?

[kkrysiak]

Let's go with a "secondary source" which can be many. You can only add these to an existing evidence item. The big thing is to show evidence from a secondary publication. If an evidence item has a primary and a secondary source, that evidence will be visible on the source page for both.

[arpaddanos]

An alternate way to address the problem of a patient appearing in different studies would be to have patient tracking IDs. These could then be linked to associated phenotypes in some manner, and a grid view implemented to see patients as rows, and EIDs as columns, with associated phenotype (AP) in the cells (along with potentially other information that could be useful). Are there instances where an ACMG evidence code would be supported by necessarily more than one pubmed ID? Or could one usually look to the publication that gives the required amount of information to achieve the code (meet the criteria for the code)?

[arpaddanos]

Note that since ACMG criteria cannot be used more than once, then a redundant occurrence of a patient meeting that criteria does not add "fake" weight to the equation in the direction of attaining a pathogenicity valuation. So the danger of double counting resulting in false pathogenicity calls is not an issue here. But tracking patients interally with an optional Patient ID can save from the danger of adding "fake" weight to a linking between a given CIViC Variant and Associated Phenotype (AP) from the HPO. A type of star rating or weighting for an individual patient's association with an AP might be strengthened with multiple papers reaffirming this association (as long as they are not just directly quoting a previous report of that association). Also the depth of detail described in the paper regarding the patient's AP would add to the weighting.