In smaegol/raceseq_scripts: Race-Seq analysis

#knitr::opts_chunk$set(echo = TRUE)
library(dplyr)
library(ggplot2)
library(PMCMRplus)
library(data.table)
library(ggsignif)
#melt_data_localization <- "/home/smaegol/storage/analyses/tail_seq_5/ALL/processing_out/all_samples_2_3_4_5_anal.tsv"
#data_melt<-fread(melt_data_localization)

LEAP

predefined transcript ends - tail lengths

LEAP_data <- all_data %>% filter(transcript=="LEAP_AU")
print(cond)
LEAP_data$condition <- as.character(LEAP_data$condition)
LEAP_data$condition <- as.factor(LEAP_data$condition)
for (cond in levels(LEAP_data$condition)) {
  plot_data <- LEAP_data %>% filter(condition==cond) %>% group_by(condition,tail_length) %>% count() %>% ungroup() %>% mutate(all_reads = sum(n)) %>% mutate(fraction=n/all_reads) 
  max_fraction = max(plot_data[-c(1,2,3),]$fraction)
  max_fraction_length = plot_data[which.max(plot_data[-c(1,2,3),]$fraction)+3,]$tail_length
  print(paste("most frequent tail length:",max_fraction_length,"(",max_fraction,")"))
  plot <- plot_data %>% ggplot(aes(x=tail_length,y=fraction)) + geom_line() + ggtitle(paste("LEAP",cond))
  print(plot)

}

predefined transcript ends - tail lengths (A tails only)

LEAP_data <- all_data %>% filter(transcript=="LEAP_AU")
print(cond)
LEAP_data$condition <- as.character(LEAP_data$condition)
LEAP_data$condition <- as.factor(LEAP_data$condition)
for (cond in levels(LEAP_data$condition)) {
  plot_data <- LEAP_data %>% filter(condition==cond) %>% group_by(condition,Atail_length) %>% count() %>% ungroup() %>% mutate(all_reads = sum(n)) %>% mutate(fraction=n/all_reads) 
  max_fraction = max(plot_data[-c(1,2,3),]$fraction)
  max_fraction_length = plot_data[which.max(plot_data[-c(1,2,3),]$fraction)+3,]$Atail_length
  print(paste("most frequent tail length:",max_fraction_length,"(",max_fraction,")"))
  plot <- plot_data %>% ggplot(aes(x=Atail_length,y=fraction)) + geom_line() + ggtitle(paste("LEAP A-tails only",cond))
  print(plot)

}

predefined transcript ends - tail lengths (U tails only)

LEAP_data <- all_data %>% filter(transcript=="LEAP_AU")
print(cond)
LEAP_data$condition <- as.character(LEAP_data$condition)
LEAP_data$condition <- as.factor(LEAP_data$condition)
for (cond in levels(LEAP_data$condition)) {
  plot_data <- LEAP_data %>% filter(condition==cond) %>% group_by(condition,Utail_length) %>% count() %>% ungroup() %>% mutate(all_reads = sum(n)) %>% mutate(fraction=n/all_reads) 
  max_fraction = max(plot_data[-c(1,2),]$fraction)
  max_fraction_length = plot_data[which.max(plot_data[-c(1,2),]$fraction)+2,]$Utail_length
  print(paste("most frequent tail length:",max_fraction_length,"(",max_fraction,")"))
  plot <- plot_data %>% ggplot(aes(x=Utail_length,y=fraction)) + geom_line() + ggtitle(paste("LEAP U-tails only",cond))
  print(plot)

}

predefined transcript ends - tail lengths (U tails only in AUtails)

LEAP_data <- all_data %>% filter(transcript=="LEAP_AU")
print(cond)
LEAP_data$condition <- as.character(LEAP_data$condition)
LEAP_data$condition <- as.factor(LEAP_data$condition)
for (cond in levels(LEAP_data$condition)) {
  plot_data <- LEAP_data %>% filter(tail_type=='AU',condition==cond) %>% group_by(condition,Utail_length) %>% count() %>% ungroup() %>% mutate(all_reads = sum(n)) %>% mutate(fraction=n/all_reads) 
  max_fraction = max(plot_data[,]$fraction)
  max_fraction_length = plot_data[which.max(plot_data[,]$fraction),]$Utail_length
  print(paste("most frequent tail length:",max_fraction_length,"(",max_fraction,")"))
  plot <- plot_data %>% ggplot(aes(x=Utail_length,y=fraction)) + geom_line() + ggtitle(paste("LEAP U-tails only",cond))
  print(plot)

}