dot_plot function

[FelixTheStudent]

Here's code to generate a dot plot of marker genes, given an identity_dataframe. Perhaps add this function to cellpypes package?

Usage example:

Tsubsets_seurat = data.frame(
  Th = malt$seurat_clusters %in% c(2, 5, 7),
  Ttox=malt$seurat_clusters == 3
)

# The difference in purity is invisible in dot plots:
genes = c("CD4", "CD8B", "CD8A")
dot_plot(FetchData(malt, genes, slot="counts"),
         malt$nCount_RNA,
         Tsubsets_seurat)

Function code:

# stuff to develop the function:
# obj = malt %>% pype_from_seurat() 
# identity_dataframe = data.frame(Bcell=malt$seurat_clusters==0,
#                                 Tcell=malt$seurat_clusters==7)
# counts   = FetchData(malt, c("CD3E", "CD79B"), slot = "counts")
# totalUMI = malt$nCount_RNA

dot_plot <- function(counts, totalUMI, identity_dataframe, return_data=FALSE){

  res =data.frame(
    celltype = NULL,
    gene     = NULL,
    percent_expressing = NULL,
    average_expression = NULL
  )

  for(celltype in colnames(identity_dataframe)) {
    for(i in 1:ncol(counts)) {
      umi = counts[, i]
      relevant <-identity_dataframe[,celltype] 
      res = rbind(res, 
                  data.frame(
                    celltype = celltype,
                    gene     = colnames(counts)[i],
                    percent_expressing = 100 * mean(umi[relevant] > 0),
                    average_expression = 1e4*mean(umi[relevant]/totalUMI[relevant]) 
                  ))
    }
  }

  plot <- res %>%
    ggplot(aes(celltype, gene, size=percent_expressing, col=average_expression)) +
    geom_point() +
    scale_size_continuous(name = "Percent Expressed") +
    viridis::scale_color_viridis(name="Average Expression") +
    cowplot::theme_cowplot()

  if(return_data) return(res)
  return(plot)

}`