NCIt and Oncotree

[larrybabb]

Oncotree is a favorite of some major clinical testing labs. While most of the diseases listed in oncotree have a reference to NCI ids, a number do not. Also, oncotree has an ontological connection to a seemingly simplified set of tissue concepts, which help organize it in a way that is favorable to onco-users.

I'm curious if there is or has been any consideration for pulling in or demonstrating how one would provide a robust integration between monarch's ontology and oncotree?

Any insights or info on if would be helpful for those making design and technical decisions with new tool development focused on clinical testing labs as well as for helping understand what may aid with standards for sharing data between labs and EHRs.

[mellybelly]

read this blog and then write back:
https://medium.com/@MonarchInit/tailoring-the-nci-thesaurus-for-semantic-interoperability-21305ccfe3a6

[balhoff]

We have a reduced version of NCIt based on the terms mapped by Oncotree here: https://github.com/NCI-Thesaurus/thesaurus-obo-edition/wiki/Downloads#oncotree-slim

The resulting hierarchy doesn't match the Oncotree structure, since it's just an extract from NCIt. Something that would be nice to do would be to take the tissues they use to group tree levels, and create expression-based grouping terms that reference those tissues and which could automatically subsume the leaf nodes they include. That would allow the tree to be generated from NCIt directly.

[larrybabb]

Thanks @mellybelly and @balhoff. Unfortunately, I'm not sure I understand all the details. My practical issue is that oncotree is being "required" by the users of our tools. So we have to provide it "as is". Some of the oncotree terms don't map to any NCI concepts. I get that you guys have "corrected" or "improved" the mapping of the oncotree terms to NCI which we will work to leverage.
However, if there are terms that are not in the oncotree-slim then we will have to still provide these because the users are always right.
In terms of having monarch provide the tissue grouping terms as @balhoff proposes, that is quite interesting and would provide the opportunity to show folks what the "improved" monarch oncotree version looks like head-to-head with MSKs oncotree including the inherent benefits of the improved NCI term mappings.

All good stuff, love everything you guys do. I'm just letting you know about the battles going on in the trenches in case there's stuff coming we can plan for.