Adding 'padding' to DotPlot rows to highlight differences across celltypes?

[RobertWhitener]

Hi, Thank you for this package! Im currently using it to generate DotPlots from our Seurat object because it is giving us much greater flexibility for the dotplots.

I've successfully generated a DotPlot which is grouped by celltype (columns), and then split each group by condition.

I'm wondering, is there a way to 'split' my featurelist (rows in my case) so that I can add some padding between each set of n numbers of features?

That would greatly help with the visualization of the data.

Thanks again!

DotPlot_Ditto <- dittoDotPlot(SUERAT, 
                              vars = rev(Genes_of_interest), 
                              group.by = 'CONDITION', 
                              split.by = 'CELLTYPE', 
                              split.nrow = 1, 
                              adjustment = 'relative.to.max' ,
                              scale = T, 
                              min.color = 'grey',
                              max.color = 'red',
                              y.labels = c('P','N','J')
                              ) + 
    coord_flip() +
    theme_minimal() +
    theme_replace(panel.background = element_blank(), strip.background = element_blank())

DotPlot_Ditto

Genes_of_interest is a vector of 48 features which is sorted by relative enrichment in each CELLTYPE, so adding a little padding between the rows that correspond to different CELLTYPEs would be amazing!

Robert

[dtm2451]

:smile: Thanks, glad you're enjoying the package!

Hmmmm faceting in the genes/vars direction is not a feature of dittoDotPlot currently, but it's certainly something you could do outside of R with a tool like Illustrator. I'll have to think about if there's a good way to implement the user-input of how to 'split' genes into categories... I've also had the desire.

Some scoping of the potential feature add:

What I'm thinking is either a list structure like split.vars = list(celltype1 = c("gene1", "gene2"), celltype2 = c("gene3", "gene4")) or just a string vector (that's the same length as vars) like split.vars = c("celltype1", "celltype1", "celltype2", "celltype2"). Just something that I can have the function 1) interpret, and 2) add to the data.frame that's given to ggplot + facet_wrap() / facet_grid().

(With an SCE, because there's a 'rowData' = Seurat@meta.data equivalent for genes, there could be a simpler way where you could label your genes by celltype within the 'rowData'. Then I'd be able to let you just name that 'rowData' column... But that doesn't help you =p, just me scoping out this add further!)

[j-andrews7]

FYI, this was also requested on Biostars earlier today.

[dtm2451]

Good poke. Lot's on the ditto-docket rn, but this could be a quicker one. Multiple asks for this feature now & I know that I'd use this feature myself too, so this can at least jump to the top of the not-yet-started list!

[dtm2451]

Implementation has begun in #142

Ultimately implemented around use of the normal vars input, where you give it a list to provide groupings. For example:

example("importDittoBulk", echo = FALSE)
logcounts(myRNA)[1:4,1:40] <- 0 # to add dropout to this mock bulk data
dittoDotPlot(
    myRNA,
    vars = list(
        'T cell'=c("gene1", "gene2"),
        'B cell'=c("gene3", "gene4")),
    group.by = "clustering")

Install via remotes::install_github("dtm2451/dittoSeq@dotplot-categories") to test and make use of this functionality for now. I still need to finalize with unit testing, updated documentation, and potentially some automated theme() updates to make the facet labels less intrusive, before deploying into the package.

[dtm2451]

Got a chance to add to this implementation! Almost ready to merge now =).

• There's automatic addition of faceting and theme args to make the categories appear as categories outside of the axis.text now. I added inputs to allow this to be turned off too, categories.split.adjust and categories.theme.adjust but why would you =p.
• And to power keeping those constrained to the vars-axis even when users might want to coord_flip() them to the y-axis, I've decided to build in a vars.dir input as well!

Made from code:

library(dittoSeq)
example(importDittoBulk, echo = FALSE)
logcounts(myRNA)[1:4,1:40] <- 0
dittoDotPlot(
    myRNA, group.by = "clustering",
    vars = list(
        Naive = paste0("gene", 1:5),
        Stimulated = paste0("gene", 6:8),
        Other = paste0("gene", 9:10)),
    vars.dir = "y"
)

[dtm2451]

Merged in! This will make it into bioc-devel in the coming days!