Calculating average ssGSEA score for easy-to-see heatmap?

[ncborcherding]

Calculating average ssGSEA score for easy-to-see heatmap?

[ncborcherding]

Hey SimoniMD,

This is a good suggestion and actually the way I generally summarise my enrichment results - I will add it to the to do list and have it released in the next version.

In the interim, a method for this using tidyr would be:

seuratObj <- addMetaData(seuratObj, ES)
seurat.meta <- [email protected]
summary <- seurat.meta %>%
                            group_by(variable) %>%
                            summarise(across(colnames(ES), median))
#You can make this ggplot friendly with
summary.melt <- reshape2::melt(summary)

Thanks,
Nick

[Dragonmasterx87]

Hi @ncborcherding ,

Sorry for the stupid question, but I havent gotten this to work.
If I have a seurat object called data and an erichIT obj called ES where I have performed the following:

seurat.meta <- data$meta.data
summary <- seurat.meta %>%
  group_by("celltype") %>% #celltype being the slot of metadata I would want to create averages for
  summarise(across(colnames(ES), median))
summary.melt <- reshape2::melt(summary)

The output doesnt make sense, its just some median value of all cells, not seggregated data based on a meta data slot i.e. celltype. If we dont do this somehow we wont be able to plot an average heatmap of certain strings of GO types..... any suggestions?

The goal is to make a heatmap possibly using ditto similar to what we do natively for all cells, but showing averages instead of all cell values.

Cheers and thanks again!
🐲

[Dragonmasterx87]

For now this is what im doing....would be gerat to have the heatmap function work as well!!!

[ncborcherding]

Hey @Dragonmasterx87

I think you might have an error in your tidy grammar - instead of "celltype", it should be celltype with no quotes. With quotes, it ends up summarizing across all cells instead of looking for a variable.

So it should read like this:

seurat.meta <- data$meta.data
summary <- seurat.meta %>%
  group_by(celltype) %>% #here is the change
  summarise(across(colnames(ES), median))
summary.melt <- reshape2::melt(summary)

Hope that helps,
Nick