Conserved markers between clusters (not between stim/exp etc)

Question

[jennifer.bridge8]

Hi Guys,

Is there a way to assess/look for conserved markers between two clusters (not between a varying group etc)?

Currently, I have identified genes that are both within the DE lists, but I don't really think that captures everything.

For background, I have subsetted a small cluster from a larger cluster, I can run DE between the two and find plenty of genes that are expressed differently but I would like to know essentially what the core genes, eg why they were clustered together in the first place?

Hints, tips would be greatly appreciated.

Many thanks,

J

[rpolicastro]

Seurat has a function FindConservedMarkers that will look for genes that are enriched or depleted in each cluster among all samples. Run this with your full seurat object though, and not the one where you subset the cluster.

[jennifer.bridge8]

The FindConservedMarkers function only works with a grouping.var, and by finding the conserved markers within the big cluster doesn't really find what is conversed between the two clusters.

[rpolicastro]

The grouping argument can be any column in the meta-data. In this case all you need to do is make a column where the two clusters you want to find conserved markers for are combined into one cluster, and then leaving all other clusters the same.

[mark]

This doesn't really work for me, too.

I want to use FindConservedMarkers to identify shared markers of 2 different celltypes.

I did everything you explained here and end up with >1,000 DEGs between those 2 celltypes, rather than the shared markers...

Any idea?

[jennifer.bridge8]

Still not having any luck, I subset the cluster for the original ident and then added metadata for the new cluster so that they are A or B then set active ident to the original ident

$ conserved.markers <- FindConservedMarkers(cluster4, ident.1 = "4", grouping.var = "AB")

and get this error.

$ Error in FindConservedMarkers(cluster4, ident.1 = "4", grouping.var = "AB") : Only one identity class present: 4

What am I doing wrong :(

[rpolicastro]

You don't want to do any subsetting. You just want to create a new column that combines the names of the two clusters, and leaves the other cluster names alone.

[jennifer.bridge8]

Where "variable" is the clusters all labelled the same but the cluster is also labelled as a distinct cluster. Eg "seurat_clusters" has 6 clusters and "variable" has 7; I get the same error $ Idents(seurat_object) <- "seurat_clusters" $ conserved.markers <- FindConservedMarkers(seurat_object, ident.1 = "4", grouping.var = "variable") $ Error in FindConservedMarkers(seurat_object, ident.1 = "4", grouping.var = "variable", : Only one identity class present: 4

[mark]

I've got a very similar problem:

I want to use FindConservedMarkers to identify similarities between 2 celltypes.

Therefore I did what was suggested here, using a meta.data column that contains the same value for both celltypes of interest.

However, when I run:

FindConservedMarkers(obj, 
                                   ident.1 = "celltype.1", 
                                   grouping.var = "group", 
                                   verbose = TRUE)

I get >1,000 genes that are obviously the DEGs between the 2 celltypes rather than the shared markers.

Any idea?

P.S. The same is true when I define ident.1 = "celltype.2"