Allow oncogenic evidence and assertions

[malachig]

After lengthy discussion it was decided that to support documenting of oncogenic evidence (distinct from biochemical functional evidence) we should do the following:

1.) For functional evidence. Add a new clinical significance value: "oncogenic" 2.) For assertions. Add a new assertion type value: "oncogenic"

[susannasiebert]

RE 2) Are there any clinical significance and assertion direction values associated with an oncogenic assertion?

Please provide short definitions for these new terms to be used for tooltips, help text etc.

[susannasiebert]

We also need to pick an icon for this new term (https://getbootstrap.com/docs/3.3/components/#glyphicons). Do we want the same icon for oncogenic clinical significance and for the oncogenic assertion type?

[arpaddanos]

I would suggest holding off on Oncogenic Assertions for now. The main discussion on Thursday Aug 6 2020 was about adding Oncogenic as one of the Functional EID Clinical Significances. I propose keeping this Issue about Functional Evidence and making a new issue for Oncogenicity Assertions.

[arpaddanos]

The issue of capturing experimental results indicating variant oncogenicity into the CIViC Functional EID data model

The current set of functional clinical significance annotation in CIViC is based around an interpretation of the system of Mueller's Morphs:

It was recently proposed during a Friday morning call in Spring 2020, that Evidence Items describing biological readouts tied to a variant such as increased cell growth (see EID 7935 and 7936) not be given the GOF, LOF, DN or Neomorphic annotations. Previous to the Spring 2020 discussion, for oncogenes, this sort of experimental data was annotated as GOF. (Note that for oncogenes, increased growth was associated with a GOF annotation, but for tumor supressors, increased growth would be associated to a LOF annotation).

The updated data model, adopted after the Friday morning discussion in Spring 2020, now restricts data described in the functional EID to evidence which directly describes a change to protein activity (e.g. enzyme activity assays such as glycogen synthase assays using radiolabeled glycogen as a readout, or phosphatase activity assays using P-32 readout). All of the functional EIDs in CIViC, in addition, are to be labeled exclusively with the disease Cancer (DOID 162). This has resulted in orphaned EIDs such as 7935 and 7936 above, which no longer fit the data model.

Oncogenicity Assertions will require supporting oncogenicity EIDs

Evidence associating a variant to increased growth or other biological effects associated with oncogenicity (see EID 7935 and 7936) are evidence types that are used for the proposed oncogenicity criteria OS2 and BS1 from the KCIS Oncogenicity efforts, in association with whom we intend to create Oncogenicity Assertions:

Oncogenicity (or benign) Assertions have inferential clinical utility in some instances:

It is therefore of interest for CIViC to be able to capture evidence into the data model which demonstrates oncogenic activity associated to a variant. One way to accomplish this is the addition of a new functional annotation Oncogenicity, so Functional EIDs can be structured as Supports Oncogenic or Does not support Oncogenic.

So with the above changes in place, the new set of clinical significance values for Functional EIDs would be

Gain of function Loss of function Neomorphic Dominant negative Oncogenic Unknown

Possible tooltip for the Oncogenic Clinical Significance:

• Sequence variant that induces or increases oncogenic cellular potential

Other points of discussion relevant to Functional EID use cases and data model:

An open question regarding what types of evidence are admissible to use for LOF, GOF, DN or Neomorphic annotation is the use of complex biological readout, which can be used to infer basic functional annotation for a variant. for instance information on neomorphic variant function is inferred from complex biological data in EID7531:

Another instance of this sort of inferential result is obtained from increased growth data in this EID8083:

Note that in both of these instances, the evidence used to obtain the LOF, GOF, DN, Neomorphic or Unaltered Function annotation goes well beyond direct in vitro assay of the protein activity. But if one considers that originally, Mueller developed the morphs working with flies, which are a complex biological system used to infer basic information on protein function, then these sorts of evidence items such as 7531 and 8083, which use a complex biological readout for basic LOF, GOF, DN or Neomorphic annotation, are in the spirit of the original Mueller Morphs, from which CIViC LOF, GOF, DN, Neomorphic and Unaltered Function annotations are derived.

Open question: should all functional EIDs with the oncogenic clinical significance still be restricted to the Cancer DOID 162? Oncogenicity can be a very cancer type specific phenomenon.

Open question: if the CIViC functional EID will be updated to integrate oncogenic information, instead of strictly functional evidence, then should the evidence type be termed biological, as originally proposed?

Open question: Encoding frequency information into a functional/biological EID in CIViC currently uses the "Unknown" annotation. Frequency data is often used to gain insight into variant oncogenicity (in a disease dependent manner as well). Should considerations be made for better integrating this data type into the CIViC functional data model? This data type corresponds to Oncogenicity codes OS3, OM5, and OP3.

[malachig]

@susannasiebert for the icon. How about glyphicon glyphicon-certificate https://getbootstrap.com/docs/3.3/components/#glyphicons.

Tool tip text from Arpad above: Sequence variant that induces or increases oncogenic cellular potential

Only allow for evidence at this time no assertions at this time.

This will be a new clinical significance value for the existing evidence type functional.

[malachig]

Further refinement has been proposed after discussion.

For functional evidence. If a user selects Oncogenic then selecting a Disease is required. However, if they pick one of the other clinical significance values (gain of function, loss of function, etc.) then the Disease must be n/a (empty in the database).

[kkrysiak]

For existing Functional Evidence, we should use civic-bot to auto assign "N/A" as the disease in a suggested change. All existing Function Evidence needs to be reviewed and appropriately moved in some cases to Oncogenic Clinical Significance where disease should not be N/A, so this change will need to be reviewed by Editors and rejected in some cases.

Also, serious help doc updates are needed to accommodate these changes.

[malachig]

After further extensive discussion it seems that we may want to review this decision.

Currently to document Oncogenic evidence a user selects:

• Evidence Type -> Functional
• Clinical Significance -> Oncogenic
• Disease must be entered for Oncogenic but Disease is disabled for all the other Clinical Significance values for Functional Evidence Type.
• Evidence Direction (Supports or Does Not Support) can be set for all Functional evidence regardless of which Clinical Significance a user selects.

The alternative proposal, which several seem to agree would be tidier would look like this:

• Evidence Type -> Oncogenic. Requiring creation of a new distinct Evidence Type. From this point the rest of the flow would look exactly like it does for Predisposing Evidence Type. i.e.:
• Clinical Significance -> N/A
• Disease is required
• Evidence Direction -> N/A

And remember that for:

• Evidence Type -> Functional -> Entering Disease will still not be permitted.

And also remember that for both Oncogenic and Functional evidence types there will not be assertions at this time. Oncogenic assertions will be added when the Oncogenicity SOP is published. We have no plans to allow assertions for Functional evidence.