Subsetting cells based on expression-level conditions

[jessielam415]

Hello! Thank you for creating this amazing package. I was wondering whether BayesSpace provides a way to subset cells in enhanced SCE based on expression-level conditions of multiple genes/features. In seurat, this can be done using their subset function. Eg. subset(x = pbmc, subset = MS4A1 > 3 & PC1 > 5) If this cannot be done using BayesSpace, would you have an idea on how I would be able to create a seurat object from enhanced SCE such that I can use the subset function in seurat? I tried using the following code below from #29 , but am encountering the error "Error: No cells found" when I use the subset function on the created Seurat object.

  seurat <- Seurat::CreateSeuratObject(counts=logcounts(sce),
                                       assay='Spatial',
                                       meta.data=as.data.frame(colData(sce)))
  seurat <- Seurat::SetIdent(seurat, value = "spatial.cluster")

Thank you!

[edward130603]

Try this:

MS4A1 = logcounts(sce)["MS4A1",] #if this doesn't work, check the rownames of your data
PC1 = logcounts(sce)["PC1",]
sce.subsetted = sce[, MS4A1 > 3 & PC1 > 5]

[jessielam415]

It works! Thank you so much! Would you know if there is a simple way to visualize the location of these subsetted spots using a plot?

[edward130603]

BayesSpace has its own plotting functions in clusterPlot and featurePlot. It will need your colData to have the columns row, col, imagerow, and imagecol which are in the tissue_positions_list.csv file from SpaceRanger output. These are the grid and image coordinates of each spot. You may also need to add in some metadata using the spatialPreprocess function with skip.PCA = TRUE and log.normalize = FALSE if you just need the metadata and not the preprocessing.

Otherwise, Seurat or SpatialExperiment have alternative ways to make spatial plots.

[jessielam415]

Thank you!