Marker management repo for CL and CL-KG

[dosumis]

We will generate a new Cell Marker ODK repo to manage markers for CL & CL-KG : - cellmark

The first aim for this repo will be to generate NS-Forest markers following standard patterns developed for BDSO (see also https://github.com/obophenotype/cell-ontology/pull/2439).

MVP:

• cellmark id space (register)
• Generates reference lung dataset NS-Forest markers for ingestion to CL
• Specifies standard TSVs + location in repo for Submission new NS-Forest markers
• Python populates 2 data files to drive templates
• templates based on BDSO for:
  • marker sets
  • CL term comments with marker details

Challenges:

• Namespace
  • OPTION1: register cellmark ID space with OBO Foundry
  • OPTION1: re-use CL
• Managing Reference gene files
  • For now we will work with reference gene sets from the vars of AnnData files. A script will pull the reference AnnData files & extract reference gene tables (name and ID) from the vars--> TSV. Some slicing mechanism will pull genes that are referenced in template data files --> generate OWL files.
  • Longer term - we will need to develop a robust plan for management, working with Richard's group with advice from Chris and his team.
    • Managing which content goes to CL
  • Option 1: separate products - one for CL and one for KG. CL product becomes component of CL build. We will do this for now.
  • Option 2: Single product used by CL import pipeline. This is appealing, but may hit scaling issues in future for CL import build.

[hkir-dev]

New repo created: https://github.com/Cellular-Semantics/CellMark (development ongoing)

LungMAP is using ensemble genes:

Lung cell atlas as well:

But in the template we have ncbi genes, in the DOSDP template https://github.com/Cellular-Semantics/CellMark/blob/main/src/markers/NSForestMarkersSource.tsv

Should we use ensemble genes in the DOSDP template as well? Who can help me for picking the correct genes?

[dosumis]

What did we get from Renne? She should be able to provide a mapping if she is providing NCBI gene IDs.

[hkir-dev]

I'm uncertain about the origin of these minimal markers. I believed that one of our curators had manually extracted them.

[dosumis]

They come from Renne's analysis. I will dig out the emails. One thing we need for the marker repo is a standard place to put files like this.

[dosumis]

Adding links and file here for now.

Renne's Zenodo pub - notebooks - this gives us a DOI to ref for analysis.

Highlights of this analysis are in HLCA_CellRef_MarkerPerformance_forDOS.xlsx

An older HLCA NS Forest analysis can be found here - exactMatch2CL_definitionAdditions.xlsx + related ticket ( https://github.com/obophenotype/cell-ontology/issues/2313) - but I believe this is superseded by analysis in Renne's pub above.

[hkir-dev]

Source data is at https://github.com/Cellular-Semantics/CellMark/blob/main/src/markers/NSForestMarkersSource.tsv

• CL product is https://github.com/Cellular-Semantics/CellMark/blob/main/clm-base.owl
• KG product is https://github.com/Cellular-Semantics/CellMark/blob/main/clm-kg.owl