library(tidyverse)

mat = read.table("HISAT2_splice_exon/map_rate.txt", sep="\t", header=F)
colnames(mat) = c("QC", "Sample", "Group", "MapRate")
mat = mat %>% filter(grepl("UT", Group)) %>% select(Sample, MapRate)
mat

##        Sample MapRate
## 1  201411_Is1  48.46%
## 2  201411_Is2  47.71%
## 3  201411_Is3  39.21%
## 4  201411_Ky1  52.22%
## 5  201411_Ky2  48.03%
## 6  201411_Ky3  47.34%
## 7  201411_Mi1  41.90%
## 8  201411_Mi2  44.07%
## 9  201411_Mi3  51.68%
## 10 201502_Is1  64.34%
## 11 201502_Is2  68.98%
## 12 201502_Is3  64.64%
## 13 201502_Ky1  59.74%
## 14 201502_Ky2  57.25%
## 15 201502_Ky3  62.53%
## 16 201502_Mi1  49.81%
## 17 201502_Mi2  49.84%
## 18 201502_Mi3  52.88%
## 19 201505_Is1  61.65%
## 20 201505_Is2  61.24%
## 21 201505_Is3  66.21%
## 22 201505_Mi1  62.41%
## 23 201505_Mi2  59.64%
## 24 201505_Mi3  57.21%
## 25 201508_Is1  52.84%
## 26 201508_Is2  52.56%
## 27 201508_Is3  49.57%
## 28 201508_Mi1  41.65%
## 29 201508_Mi2  39.81%
## 30 201508_Mi3  50.22%
## 31 201511_Is1  50.44%
## 32 201511_Is2  57.19%
## 33 201511_Is3  48.11%
## 34 201511_Mi1  43.43%
## 35 201511_Mi2  46.68%
## 36 201511_Mi3  48.44%
## 37 201602_Is1  54.43%
## 38 201602_Is2  58.04%
## 39 201602_Is3  62.02%
## 40 201602_Mi1  65.11%
## 41 201602_Mi2  53.94%
## 42 201602_Mi3  54.47%
## 43 201603_Is1  49.79%
## 44 201603_Is2  55.70%
## 45 201603_Is3  51.34%
## 46 201603_Mi1  49.07%
## 47 201603_Mi2  52.45%
## 48 201603_Mi3  51.61%
## 49 201605_Is1  53.62%
## 50 201605_Is2  57.66%
## 51 201605_Is3  54.34%
## 52 201605_Mi1  61.27%
## 53 201605_Mi2  61.60%
## 54 201605_Mi3  59.05%
## 55 201608_Is1  51.35%
## 56 201608_Is2  60.69%
## 57 201608_Is3  45.52%
## 58 201608_Mi1  39.61%
## 59 201608_Mi2  43.00%
## 60 201608_Mi3  33.37%
## 61 201609_Is1  34.86%
## 62 201609_Is2  51.61%
## 63 201609_Is3  50.35%

figmat = mat %>% 
  mutate(MapRate = as.numeric(gsub("%","",MapRate)))

summary(figmat$MapRate)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   33.37   48.27   52.22   52.63   58.55   68.98

fig1 = ggplot(figmat) + 
  geom_histogram(aes(x=MapRate, y=..count..), bins=20, fill="grey80") +
  geom_vline(xintercept = mean(figmat$MapRate), color="red", size=1) + 
  scale_x_continuous(limits = c(20, 80)) +
  scale_y_continuous(limits = c(0, 14), breaks=seq(0, 14, by=2)) +
  theme_bw() + 
  xlab("Mapping rate [%]") + ylab("Count")
fig1