SE vs StrE vs TE

Preparation

Set up the R environment by including a few necessary R libraries

library(ggplot2)

## Registered S3 methods overwritten by 'ggplot2':
##   method         from 
##   [.quosures     rlang
##   c.quosures     rlang
##   print.quosures rlang

library(cowplot)

## 
## Attaching package: 'cowplot'

## The following object is masked from 'package:ggplot2':
## 
##     ggsave

library(reshape2)
library(RColorBrewer)

Data import and process

Get the midian size and the number of SEs/StrEs/TEs in each cell line, and format the data as a data frame with additional information.

.proc <- function(name) { 
  SE <- read.table(paste0(name,".SE.bed"))
  TE <- read.table(paste0(name,".TE.bed"))
  StrE <- read.table(paste0(name,".StrE.bed"))
  SE.num <- nrow(SE)
  TE.num <- nrow(TE)
  StrE.num <- nrow(StrE)
  SE.median_size <- median(SE[,3]-SE[,2])
  TE.median_size <- median(TE[,3]-TE[,2])
  StrE.median_size <- median(StrE[,3]-StrE[,2])
  data.frame(sample=name, type=c("SE","TE","StrE"), number=c(SE.num, TE.num, StrE.num), median_size=c(SE.median_size, TE.median_size, StrE.median_size))
}
samples <- c("GM12878", "H1", "HepG2", "HMEC", "HSMM", "Huvec", "K562", "NHLF")
data <- do.call(rbind, sapply(samples, .proc, simplify=F))

Median size

Using box-plots to compare the median size of SEs, StrEs, and TEs. Here, the median size is in bp, and the values are plotted on a log scale.

ggplot(data, aes(x=type, y=median_size, color=type)) + geom_boxplot(outlier.size=0.3) + 
  geom_jitter(alpha=0.3, size=0.3) + 
  scale_colour_manual(values=brewer.pal(7, "Set1")[c(2,4,3)]) + 
  scale_y_log10() + theme(legend.title=element_blank()) +
  xlab("") + ylab("Median size (bp)")

Here, the values are plotted in a linear scale.

ggplot(data, aes(x=type, y=median_size, color=type)) + geom_boxplot(outlier.size=0.3) + 
  geom_jitter(alpha=0.3, size=0.3) +
  scale_colour_manual(values=brewer.pal(7, "Set1")[c(2,4,3)]) + 
  theme(legend.title=element_blank()) +
  xlab("") + ylab("Median size (bp)")

Numbers

Using box-plots to compare the number of SEs and TEs, and split the samples into cell lines, primary cells, and tissues. Here, the values are plotted on a log scale.

ggplot(data, aes(x=type, y=number, color=type)) + geom_boxplot(outlier.size=0.3) + 
  geom_jitter(alpha=0.3, size=0.3) + 
  scale_y_log10() + 
  scale_colour_manual(values=brewer.pal(7, "Set1")[c(2,4,3)]) + 
  theme(legend.title=element_blank()) +
  xlab("") + ylab("Number")