Question about cluster idents in Seurat object

[Seoyeon Kim]

Hello authors. Thanks for sharing the valuable data and code.
While reproducing the data analysis using Seurat, I had a question about the differences between three idents that can be used to annotate the clusters: cell_type, cell_type_major, and cell_type_tumor.
Sometimes, the same cluster is named differently depending on which ident we used for annotating the cluster.
For example, a cluster annotated with "type I/II pneumocytes" with cell_type ident is differently annotated with "Alveolar cell type 1/2" when using the cell_type_major ident.
Could you explain the differences between these three idents, and what should I use as the principal object for the annotation?
Thank you so much.

[Gregor Sturm]

Hi @ssun1116,

thanks for your question!
Assuming you downloaded the dataset from cell-x-gene:

• cell_type contains standardized cell-type annotations as defined by the cell-ontology. This is one of the fields controlled by cell-x-gene, and in general, I'd recommend to use this annotation unless you need a higher resolution or want to reproduce results from our paper.

The other cell-type annotations are provided by us and the labels just are those that we used for the paper:

• cell_type_tumor is the most fine-grained resolution (higher resolution for than cell_type, as we could not map all labels to the cell ontology)
• cell_type_neutro(_coarse) is the same as cell_type_tumor, but additionally contains the neutrophil cluster labels
• cell_type_major is a medium-resolution annotation that we used for most analyses in the paper.

Hope that helps,
Gregor

[Seoyeon Kim]

Got it. Thank you for the clarification!