Searching for a way to plot Seurat style equivalent dotplot for gene expression with DESeq on RNA-seq data

Question

[Ondina]

Hello,

I'm searching for a way to plot something like this but for RNA-seq data: so having for exemple as x a set of genes, y the different groups and the expression plotted by the rlog transformed values for example of the DESeq RNA-seq data.

I kind of found a way to do it using plotCounts and then ggplot but for a single gene ... I would like something that allows me to plot multiple genes and their counts number in a single plot.

Do you have any idea of something like that already existing as a package or if there's an easy way to do it?

[Papyrus]

You mean something like this?

library(ggplot2)

df <- data.frame(
  groups = c(rep("group1",3),rep("group2",3),rep("group3",3)),
  genes = rep(c("gene1","gene2","gene3"),3),
  exp = rnorm(9,0,10)
)

ggplot(df, aes(x = genes, y = groups, color = exp, size = exp)) + 
  geom_point(stat = 'identity') + 
  xlab("") + ylab("") + ggtitle("your data") + 
  theme_bw()

[ATpoint]

It's just ordinary ggplot, basically all of these plotting functions you find in Seurat or Bioconductor are just ggplot wrappers. Can you provide some data that can be copy/pasted into R, maybe with dput()?

[Ondina]

Yeah I know all of this, it's just easier to use something that's already coded out and pretty, especially when I want to show some basic stuff to use in R to colleagues who don't know how to code or use ggplot.

I can't really use dput() now as it's confidential data but the data just looks like any other RNA-seq data, treated/untreated, several batches etc ...

Answer 1

[jared.andrews07]

dittoSeq has a dotplot function that will work with a SummarizedExperiment.

Add your rlog counts as an assay and provide it to the function if you want them to be used, something like:

library(dittoSeq)

assay(dds, "rlog") <- rlog.counts
dittoDotPlot(dds, c("your", "genes", "here"), group.by = "your_grouping_variable", assay = "rlog")

There are lots of other options for controlling size, etc. Though for bulk data, I really find a heatmap to be more informative than a dotplot.

[Ondina]

Hmmm, indeed, dittoSeq looks nice, thank you for reminding me of it. Didn't knew it could be used directly with data matrices instead of a Seurat object. Thank you!