jashapiro
commented
3 hours ago Just looking at ontology tools, My main suggestion is probably going to be ontologyX https://academic.oup.com/bioinformatics/article/33/7/1104/2843897, which you may have already found. This should make it pretty easy to see if one term is an "ancestor" of another, which seems likely to be the main kind of relationship we are interested in?
Please link to the GitHub Discussion for this proposed analysis.
853
Describe the goals of this analysis module.
The goal of this module is to assign a consensus cell type label to cells when both SingleR and CellAssign cell type assignments agree. We would also like these consensus cell types to be ontology aware.
To do this we will need to identify a set of rules that can be used to identify cells that are in agreement. E.g., if two cell types share a parent term, use the parent term as the consensus label. This module will focus on developing these rules and testing that these rules work to assign consensus cell types on a subset of samples.
What software will you require?
I plan on doing most of this in R if possible, but we will need a way to identify shared parent terms with cell ontology IDs. TBD on what tool we will use for that and what software will be required.
What will your first pull request contain?
I'm first going to initate the module in a PR and then will work on assigning cell ontology IDs to the cell type labels in the CellAssign reference (separate issue for this forthcoming).
What computational resources will you require?
TBD
If known, when do you expect to file the first pull request?
Week of 11/18