# Percentage distribution of cells in all clusters based on their treatment condition?

I have 2151 cells, I clustered them by Seurat to 5 clusters. With the code below, I am able to have the number of cells per cluster and per condition:

number_perCluster<- table(object@meta.data$$conditions, object@meta.data$$clusterID)


Could I please get a hint on how I could proceed to produce a figure where I say: X% of cells treated with condition "Y" are located in cluster "C" and so forth?

• Have a look at prop.table, also if you want a figure, please describe it a bit more. This is usually better displayed as a table or heatmap, but perhaps you want something else... – llrs Feb 6 '19 at 13:39
• Please don't also create a github issue on the Seurat page each time you ask a question here, there is no need – TimStuart Feb 6 '19 at 14:40

Here is a solution using dplyr and ggplot2:

library(Seurat)
library(dplyr)
library(ggplot2)

meta.data <- pbmc_small@meta.data

# create random classifications for the sake of this example
meta.data\$condition <- sample(c('A', 'B', 'C'), nrow(meta.data), replace = TRUE)

counts <- group_by(meta.data, condition, res.1) %>% summarise(count = n())

ggplot(counts, aes(res.1, count, fill = condition)) +
geom_bar(stat = 'identity')