0

I'm using ComplexUpset in R. I have a results from a Transcription Experiment. It has 4 comparisons. Each gene in in each comparison can be significantly regulated. Furthermore, each can be positively regulated or negatively regulated or opposite.

When producing the plot the intersections are on the genes that are significantly regulated (independent of their direction). I would like to annotate the plot with either another Bar chart on top with stacked proportion of positively regulated or negatively regulated genes. For this I created an additional data set with a factor indicating this for all the 'intersections' shown in the upset plot.

I would like to have something like the examples in th vignette using movies but the MPAA Rating is a factor for all those: enter image description here

This graph shows in principle what I want but not aligned as if it would using annotations=

what I've gotten so far

The sample data and code used for this:

set.seed(45)
example_df <- data.frame(geneid=paste('gene_', 1:15000, sep = ''),
                         comparison_1=sample(c(T, F), size = 15000, replace = T, prob = c(.7, .3)),
                         comparison_2=sample(c(T, F), size = 15000, replace = T, prob = c(.40, .6)),
                         comparison_3=sample(c(T, F), size = 15000, replace = T, prob = c(.25, .75)),
                         comparison_4=sample(c(T, F), size = 15000, replace = T, prob = c(.12, .88)), 
                         fill_comparison_1=sample(c('up', 'down'), size = 15000, replace = T, prob = c(.60,.40)),
                         fill_comparison_2=sample(c('up', 'down'), size = 15000, replace = T, prob = c(.55,.45)), 
                         fill_comparison_3=sample(c('up', 'down'), size = 15000, replace = T, prob = c(.50,.50)), 
                         fill_comparison_4=sample(c('up', 'down'), size = 15000, replace = T, prob = c(.65,.35)))

example_df[example_df$fill_comparison_1=='up' & example_df$fill_comparison_2=='up', "intersection_1"] <- 'up'
example_df[example_df$fill_comparison_1=='down' & example_df$fill_comparison_2=='down', "intersection_1"] <- 'down'
example_df[example_df$fill_comparison_1=='up' & example_df$fill_comparison_2=='down', "intersection_1"] <- 'opposite'
example_df[example_df$fill_comparison_1=='down' & example_df$fill_comparison_2=='up', "intersection_1"] <- 'opposite'

example_df[example_df$fill_comparison_3=='up' & example_df$fill_comparison_4=='up', "intersection_2"] <- 'up'
example_df[example_df$fill_comparison_3=='down' & example_df$fill_comparison_4=='down', "intersection_2"] <- 'down'
example_df[example_df$fill_comparison_3=='up' & example_df$fill_comparison_4=='down', "intersection_2"] <- 'opposite'
example_df[example_df$fill_comparison_3=='down' & example_df$fill_comparison_4=='up', "intersection_2"] <- 'opposite'

example_df[example_df$fill_comparison_1=='up' & example_df$fill_comparison_3=='up', "intersection_3"] <- 'up'
example_df[example_df$fill_comparison_1=='down' & example_df$fill_comparison_3=='down', "intersection_3"] <- 'down'
example_df[example_df$fill_comparison_1=='up' & example_df$fill_comparison_3=='down', "intersection_3"] <- 'opposite'
example_df[example_df$fill_comparison_1=='down' & example_df$fill_comparison_3=='up', "intersection_3"] <- 'opposite'

example_df[example_df$fill_comparison_2=='up' & example_df$fill_comparison_4=='up', "intersection_4"] <- 'up'
example_df[example_df$fill_comparison_2=='down' & example_df$fill_comparison_4=='down', "intersection_4"] <- 'down'
example_df[example_df$fill_comparison_2=='up' & example_df$fill_comparison_4=='down', "intersection_4"] <- 'opposite'
example_df[example_df$fill_comparison_2=='down' & example_df$fill_comparison_4=='up', "intersection_4"] <- 'opposite'

example_fill <- melt(data = example_df[, c(1, 6:ncol(example_df))], id.vars = 'geneid')
example_fill <- na.omit(example_fill)


fill_plot <- 
  ggplot(example_fill, mapping = aes(variable, fill=value))+
  geom_bar(stat='count', position='fill')+
  theme_minimal()+
  theme(axis.text.x = element_text(angle = 90)) 


ex_upset <- upset(example_df, intersect = c(paste('comparison', 1:4, sep='_')), mode = 'inclusive_intersection', max_degree=4, keep_empty_groups=T, sort_sets=F, intersections=list('comparison_1', 'comparison_2', 'comparison_3', 'comparison_4', c(paste('comparison',1:2,sep='_')), c(paste('comparison',3:4,sep='_')), paste('comparison',c(1, 3),sep='_'), paste('comparison',c(2, 4),sep='_'))) 

fill_plot/ex_upset
krassowski
  • 13,598
  • 4
  • 60
  • 92
Panchito
  • 337
  • 1
  • 3
  • 12
  • Why there are four `fill_comparison_`s and four `intersection_`s in the bar plot? How are they supposed to align with the intersections from upset? Would any of the information fit in the set sizes bars maybe? – krassowski Nov 12 '22 at 13:29
  • yes, these would be the same single comparisons and intersections of those comparisons. they would be able to align I guess I didn't ````sort_intersections=F```` – Panchito Nov 14 '22 at 19:49
  • Also, they are essentially from the same data but the in any unlike the ```movies$mpaa``` data set each set can have different levels assigned to each row. When both are "positively regulated" they would have the same level. That's why for each there is a separate column. – Panchito Nov 16 '22 at 18:56

0 Answers0