Seurat clustering and classification

Question

[HK]

Hi All,

I have a seurat object, but now i want to subset that on the basis of few genes only (almost 250 genes). These 250genes I got from a published paper showing different macrophae annotations (7 annotated clusters).

The idea is to see if our annotated macrophages fit in one (or more) of the 7 macrophaes subtypes as annotated by the published paper. So potentially we could find that our one type of annotation, based on gene expression levels, are most alike any of the defines publihed annotation.

So, I have the subset object, I found almost 200 genes from the published ist in my object. Then , i ran the sclaeData, Run PCA, f FindNeighbors, FindClusters, RunUMAP functions on my subset.

But, now i am confused how to move from here, how to know which of the defined annotated clusters fit in my object.

Kindly i need help!!

Answer 1

[Hamid Ghaedi]

Suppose you have a reference dataset (macrophage dataset) that is already annotated with 7 subtypes, and you want to annotate your query dataset according to this reference. In such cases, you can make use of SingleR.

I also wrote a blog post that you might find helpful. In the post, I used a manually labeled cell dataset and wanted to compare the concordance of my cell labeling with a reference dataset ( bladder tissue dataset from Tabula Sapiens). The figure below demonstrates that my manual cell labeling (y-axis labels) aligns well with the Tabula Sapiens labels (x-axis labels). For example, on the y-axis, there is a label called APCs (Macrophages, B-cells), which exhibits a high level of overlap with two labels on the y-axis: B cell and macrophage.