CODEWITHSUNDEEP
rplotggplot2
neversaint
neversaint

Reputation: 64004

How to enable `geom_text` to recognize `aes` in QPLOT (R programming)

I have a data that looks like this

ensg mirna_hgc time value perc id
ENSG00000211521 MIR665 x 89 2.07612456747405 1
ENSG00000207787 MIR98 x 73 1.73010380622837 2
...
ENSG00000207827 MIR30A y 99 21.4532871972318 288
ENSG00000207757 MIR93 y 94 1.73010380622837 289

What I'm trying to do is to create a facet plot with label on top of it. The label can be easily called from the perc column.

Using this code:

dat.m <- read.delim("http://dpaste.com/1271039/plain/",header=TRUE,sep=" ")

qplot(value, data=dat.m,facets=time~.,binwidth=1,main="")+
xlab("Value")+ ylab("Count")+
theme(legend.position="none")+
stat_bin(aes(value,label=sprintf("%.01f",perc)),geom="text")

But it gave me this error:

Error: geom_text requires the following missing aesthetics: label

What I'm trying to do is to generate this plot: enter image description here

Upvotes: 2

Views: 2256

Answers (1)

mnel
mnel

Reputation: 115392

Your issue arises in part

  • because you are using qplot (which makes things more confusing than you need).

  • You have set binwidth = 2 which means each histogram bar is based (possibly) on 2 values of value. Your 1-to-1 mapping of perc to value is now 2 -to -1 mapping of perc to the binned values

Using plyr to continue the hadleyesque implementation we can aggregated before we plot

library(plyr)
agg.data <-  ddply(dat.m, .(value,time), summarize, p = unique(perc), n = length(perc))

ggplot(agg.data, aes(x= value)) + 
 geom_bar(aes(y = n),stat='identity') + 
 facet_grid(time~.) + 
 geom_text(aes(y=n, label = sprintf('%.01f',p)),vjust=-1,size=3)

enter image description here

If you want to use geom_histogram and display the proportions associated with each bar, this can be done quite easily, but you will need to use the values created by stat_bin (namely ..density.., which is proportion allocated to each bar)

ggplot(dat.m, aes(x= value)) + 
  geom_histogram(binwidth=2) +  
  facet_grid(time~.) + 
  stat_bin(aes(y = ..count.., label = sprintf('%.01f', 100 * ..density..)),
           binwidth=2,geom='text',vjust = -1)

enter image description here

and to replace the 0.0 with no labels use ifelse (with some definition trickery to avoid replicating the test.

ggplot(dat.m, aes(x= value)) + 
   geom_histogram(binwidth=2) +  facet_grid(time~.) + 
   stat_bin(aes(y = ..count.., 
     label = ifelse(test = (xxx <- sprintf('%.01f', 100 * ..density..))=='0.0','',xxx)),
     binwidth=2,geom='text',vjust = -0.7)

enter image description here

And note that you could get the same thing using qplot using

qplot(value, data=dat.m,facets=time~.,binwidth=2,main="")+
     xlab("Value")+ ylab("Count")+
     theme(legend.position="none")+
     stat_bin(aes(value,label=ifelse(test = (xxx <-  
       sprintf("%.01f",..density..*100))=='0.0','',xxx)),geom="text",binwidth=2,vjust = -0.7)

Upvotes: 4

Related Questions