view bubbles_v9_NSAF_natural_log.R @ 27:ab9ace68e3b5 draft

Uploaded

rm(list = ls()) 
###################################################################################################
# R-code: Multi-bubble graph generation from SAINTexpress output
# Author: Brent Kuenzi
###################################################################################################
# This Script generates the bubble graphs based upon Saint output. 
###################################################################################################
# Copyright (C)  Brent Kuenzi.
# Permission is granted to copy, distribute and/or modify this document
# under the terms of the GNU Free Documentation License, Version 1.3
# or any later version published by the Free Software Foundation;
# with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.
# A copy of the license is included in the section entitled "GNU
# Free Documentation License".
###################################################################################################
## REQUIRED INPUT ##

# 1) listfile: SAINTexpress generated "list.txt" file
# 2) preyfile: SAINT pre-processing generated "prey.txt" file used to run SAINTexpress

## OPTIONAL INPUT ##

# 3) crapome: raw output from crapome Workflow 1 query (http://www.crapome.org)
# 4) color: bubble color (default = "red")
#     - color = "crapome": color bubbles based on Crapome(%)
#     - Also recognizes any color within R's built-in colors() vector
# 5) label: Adds gene name labels to bubbles within the "zoomed in" graphs (default = FALSE)
# 6) cutoff: Saintscore cutoff to be assigned for filtering the "zoomed in" graphs (default = 0.8)
# 7) type: Specifies if the data is MaxQuant (MQ) or Scaffold (SC) data (default = "SC")
# 8) inc_file: Selects only the uniprot ids in the provided list (default ="None")
# 9) exc_file: Removes the proteins in the list (default = "None")
###################################################################################################


ins_check_run <- function(){
  if ('dplyr' %in% rownames(installed.packages())){}
  else {
    install.packages('dplyr', repos = 'http://cran.us.r-project.org')
  }
  if ('tidyr' %in% rownames(installed.packages())){}
  else {
    install.packages('tidyr', repos = 'http://cran.us.r-project.org')
  }
  if ('ggplot2' %in% rownames(installed.packages())){}
  else {
    install.packages('ggplot2', repos = 'http://cran.us.r-project.org')
  }
}

ins_check_run()
library(dplyr); library(tidyr); library(ggplot2)


main <- function(listfile, preyfile, crapome = FALSE, color = "red", label = FALSE, cutoff = 0.8, type = "SC", inc_file = "None", exc_file = "None" ) {
  cutoff_check(cutoff)
  listfile <- list_type(listfile, inc_file, exc_file)
  if(type == "SC") {
    df <- merge_files_sc(listfile, preyfile, crapome)
  }
  if(type == "MQ") {
    df <- merge_files_mq(listfile, preyfile, crapome)
  }
  bubble_NSAF(df, color)
  bubble_SAINT(df, color)
  bubble_zoom_SAINT(df, color, label, cutoff)
  bubble_zoom_NSAF(df, color, label, cutoff)
  write.table(df, "output.txt", sep = "\t", quote = FALSE, row.names = FALSE)
}
###################################################################################################
# Include and Exclude list filtering
###################################################################################################
list_type <- function(df, inc_file, exc_file) {
  Saint <- read.delim(df, stringsAsFactors = FALSE)
  if (inc_file != "None") {
    if (exc_file == "None") {
      inc_prots <- read.delim(inc_file, sep = '\t', header = FALSE, stringsAsFactors = FALSE)
      print(inc_prots[, 1])
      print(Saint$Prey)
      filtered_df = subset(Saint, Saint$Prey == inc_prots[, 1])
    }
    else {
      inc_prots <- read.delim(inc_file, sep = '\t', header = FALSE, stringsAsFactors = FALSE)
      exc_prots <- read.delim(exc_file, sep = '\t', header = FALSE, stringsAsFactors = FALSE)
      filtered_df = subset(Saint, Saint$Prey == inc_prots[, 1])
      filtered_df = subset(filtered_df, filtered_df$Prey != exc_prots[, 1])
    }
  }
  else if (exc_file != "None") {
    exc_prots <- read.delim(exc_file, sep = '\t', header = FALSE, stringsAsFactors = FALSE)
    filtered_df = subset(Saint, Saint$Prey != exc_prots[, 1])
  }
  else {
    filtered_df = Saint
  }
  return(filtered_df)
  
}
###################################################################################################
# Merge input files and caculate Crapome(%) and NSAF for each protein for each bait
###################################################################################################
merge_files_mq <- function(SAINT, prey_DF, crapome = FALSE) {
  #SAINT <- read.table(SAINT_DF, sep = '\t', header = TRUE)
  #Some of these read.table()'s don't use stringsAsFactors = FALSE. Is this on purpose? Factors give rise to some really weird and unpredictable behavior; suggest always using stringsAsFactors = FALSE
  prey <- read.table(prey_DF, sep = '\t', header = FALSE); colnames(prey) <- c("Prey", "Length", "PreyGene")
  DF <- merge(SAINT, prey)
  DF$SpecSum <- log2(DF$SpecSum)
  
  if(crapome != FALSE) {
    crapome <- read.table(crapome, sep = '\t', header = TRUE)
    colnames(crapome) <- c("Prey", "Symbol", "Num.of.Exp", "Ave.SC", "Max.SC")
    DF1 <- merge(DF, crapome); as.character(DF1$Num.of.Exp); DF1$Symbol <- NULL;
                    DF1$Ave.SC <- NULL; DF1$Max.SC <- NULL # Removes unnecessary columns.
    DF1$Num.of.Exp <- sub("^$", "0 / 1", DF1$Num.of.Exp ) # Replace blank values with 0 / 1.
    DF <- DF1 %>% separate(Num.of.Exp, c("NumExp", "TotalExp"), " / ") # Split into 2 columns.
    DF$CrapomePCT <- 100 - (as.integer(DF$NumExp) / as.integer(DF$TotalExp) * 100) # Calculate the crapome %.
  }
  DF$SAF <- DF$AvgSpec / DF$Length
  DF2 = DF %>% group_by(Bait) %>% mutate(NSAF = SAF/sum(SAF))
  DF$NSAF = DF2$NSAF
  return(DF)
}

merge_files_sc <- function(SAINT, prey_DF, crapome = FALSE) {
  #SAINT <- read.table(SAINT_DF, sep = '\t', header = TRUE)
  prey <- read.table(prey_DF, sep = '\t', header = FALSE); colnames(prey) <- c("Prey", "Length", "PreyGene")
  DF <- merge(SAINT, prey)
  
  if(crapome != FALSE) {
    crapome <- read.table(crapome, sep = '\t', header = TRUE)
    colnames(crapome) <- c("Prey", "Symbol", "Num.of.Exp", "Ave.SC", "Max.SC")
    DF1 <- merge(DF, crapome); as.character(DF1$Num.of.Exp); DF1$Symbol <- NULL;
                    DF1$Ave.SC <- NULL; DF1$Max.SC <- NULL # Removes unnecessary columns.
    DF1$Num.of.Exp <- sub("^$", "0 / 1", DF1$Num.of.Exp ) # Replace blank values with 0 / 1.
    DF <- DF1 %>% separate(Num.of.Exp, c("NumExp", "TotalExp"), " / ") # Split into 2 columns.
    DF$CrapomePCT <- 100 - (as.integer(DF$NumExp) / as.integer(DF$TotalExp) * 100) # Calculate the crapome %.
  }
  DF$SAF <- DF$AvgSpec / DF$Length
  DF2 = DF %>% group_by(Bait) %>% mutate(NSAF = SAF/sum(SAF))
  DF$NSAF = DF2$NSAF
  return(DF)
}
###################################################################################################
# Plot all proteins for each bait by x = ln(NSAF), y = Log2(FoldChange)
###################################################################################################
bubble_NSAF <- function(data, color) {
    if(color == "crapome") {
      a <- subset(data, CrapomePCT < 80, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait))
      b <- subset(data, CrapomePCT >= 80, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait))
      p <- qplot(x = log(NSAF), y = log2(FoldChange), data = a, colour = I("tan"), size = SpecSum) + scale_size(range = c(1, 10)) + 
        geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a)
      if(length(levels(a$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")} # multiple graphs if multiple baits
      #The text says ln() which is log base e, but the code uses log base 10. Fix code or the axis label. 
      p <- p + geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum, color = CrapomePCT), data = b) + 
        scale_colour_gradient(limits = c(80, 100), low = "tan", high = "red") + 
        labs(colour = "CRAPome Probability \nof Specific Interaction (%)", x = "ln(NSAF)") + 
        geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = b)
      return(ggsave(p, width = 8, height = 4, filename = "bubble_NSAF.png"))
    }
   if(color != "crapome") {
      p <- qplot(x = log(NSAF), y = log2(FoldChange), data = data, colour = I(color), size = SpecSum) + scale_size(range = c(1, 10)) + 
        geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = data) + # add bubble outlines
          labs(x = "ln(NSAF)")
        if(length(levels(data$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
      return(ggsave(p, width = 8, height = 4, filename = "bubble_NSAF.png"))
    }
  }
###################################################################################################
# Plot all proteins for each bait by x = Saintscore, y = Log2(FoldChange)
###################################################################################################
bubble_SAINT <- function(data, color) {
    if(color == "crapome") {
      a <- subset(data, CrapomePCT < 80, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait)) #filter on CRAPome
      b <- subset(data, CrapomePCT >= 80, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait))
      p <- qplot(x = SaintScore, y = log2(FoldChange), data = a, colour = I("tan"), size = SpecSum) + 
        scale_size(range = c(1, 10)) + geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a)
      if(length(levels(a$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
      p <- p + geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum, color = CrapomePCT), data = b) + 
        scale_colour_gradient(limits = c(80, 100), low = "tan", high = "red") + 
        labs(colour = "CRAPome Probability \nof Specific Interaction (%)") +
        geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = b)
      return(ggsave(p, width = 8, height = 4, filename = "bubble_SAINT.png"))
    }
    if(color != "crapome") {
      p <- qplot(x = SaintScore, y = log2(FoldChange), data = data, colour = I(color), size = SpecSum) +
        scale_size(range = c(1, 10)) + geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = data)
      if(length(levels(data$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
      return(ggsave(p, width = 8, height = 4, filename = "bubble_SAINT.png"))
    }
  }
###################################################################################################
# Filter proteins on Saintscore cutoff and plot for each bait x = Saintscore, y = Log2(FoldChange)
###################################################################################################
bubble_zoom_SAINT <- function(data, color, label = FALSE, cutoff = 0.8) {
  if(color == "crapome") {
    a <- subset(data, CrapomePCT < 80 & SaintScore >= cutoff, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait, PreyGene))
    b <- subset(data, CrapomePCT >= 80 & SaintScore >= cutoff, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait, PreyGene))
    p <- qplot(x = SaintScore, y = log2(FoldChange), data = a, colour = I("tan"), size = SpecSum) + 
      scale_size(range = c(1, 10)) + ggtitle("Filtered on SAINT score")+geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a)
    if(label == TRUE & length(a$NSAF != 0)) {
      p <- p + geom_text(data = a, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black")
    }
    if(length(levels(a$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
    p <- p + geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum, color = CrapomePCT), data = b) + 
      scale_colour_gradient(limits = c(80, 100), low = "tan", high = "red") + 
      labs(colour = "CRAPome Probability \nof Specific Interaction (%)") + 
      geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = b)
    if(label == TRUE & length(b$NSAF != 0)) {
      p <- p + geom_text(data = b, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black", show_guide = FALSE)
    }
    return(ggsave(p, width = 8, height = 4, filename = "bubble_zoom_SAINT.png"))
  }
  if(color != "crapome") {
    a <- subset(data, SaintScore >= cutoff, select = c(NSAF, SpecSum, FoldChange, SaintScore, Bait, PreyGene))
    p <- qplot(x = SaintScore, y = log2(FoldChange), data = a, colour = I(color), size = SpecSum) +
      scale_size(range = c(1, 10)) + ggtitle("Filtered on SAINT score") + 
      geom_point(aes(x = SaintScore, y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a)
    if(label == TRUE & length(a$NSAF != 0)) {
      p <- p + geom_text(data = a, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black", show_guide = FALSE)
    }
    if(length(levels(data$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
    return(ggsave(p, width = 8, height = 4, filename = "bubble_zoom_SAINT.png"))
  }
}
###################################################################################################
# Filter proteins on Saintscore cutoff and plot for each bait x = log(NSAF), y = Log2(FoldChange)
###################################################################################################
bubble_zoom_NSAF <- function(data, color, label = FALSE, cutoff = 0.8) {
  if(color == "crapome") {
    a <- subset(data, CrapomePCT < 80 & SaintScore >= cutoff, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait, PreyGene))
    b <- subset(data, CrapomePCT >= 80 & SaintScore >= cutoff, select = c(NSAF, SpecSum, CrapomePCT, FoldChange, SaintScore, Bait, PreyGene))
    p <- qplot(x = log(NSAF), y = log2(FoldChange), data = a, colour = I("tan"), size = SpecSum) + 
      scale_size(range = c(1, 10)) + ggtitle("Filtered on SAINT score") + 
      geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a)
    if(label == TRUE & length(a$NSAF != 0)) {
      p <- p + geom_text(data = a, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black")
    }
    if(length(levels(a$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
    p <- p + geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum, color = CrapomePCT), data = b) + 
      scale_colour_gradient(limits = c(80, 100), low = "tan", high = "red") + 
      labs(colour = "CRAPome Probability \nof Specific Interaction (%)", x = "ln(NSAF)") + 
      geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = b)
    if(label == TRUE & length(b$NSAF != 0)) {
      p <- p + geom_text(data = b, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black", show_guide = FALSE)
    }
    return(ggsave(p, width = 8, height = 4, filename = "bubble_zoom_NSAF.png"))
  }
  if(color != "crapome") {
    a <- subset(data, SaintScore >= cutoff, select = c(NSAF, SpecSum, FoldChange, SaintScore, Bait, PreyGene))
    p <- qplot(x = log(NSAF), y = log2(FoldChange), data = a, colour = I(color), size = SpecSum) +
      scale_size(range = c(1, 10)) + ggtitle("Filtered on SAINT score") + 
      geom_point(aes(x = log(NSAF), y = log2(FoldChange), size = SpecSum), colour = "black", shape = 21, data = a) + 
      labs(x = "ln(NSAF)")
    if(label == TRUE & length(a$NSAF != 0)) {
      p <- p + geom_text(data = a, aes(label = PreyGene, size = 10, vjust = 0, hjust = 0), colour = "black", show_guide = FALSE)
    }
    if(length(levels(data$Bait) > 1)) {p <- p + facet_wrap(~Bait, scales = "free_y")}
    return(ggsave(p, width = 8, height = 4, filename = "bubble_zoom_NSAF.png"))
  }
}
###################################################################################################
# Check Saintscore cutoff and stop program if not between 0 and 1
###################################################################################################
cutoff_check <- function(cutoff){
  if( any(cutoff < 0 | cutoff > 1) ) stop('SAINT score cutoff not between 0 and 1. Please correct and try again')
}

args <- commandArgs(trailingOnly = TRUE)
main(args[1], args[2], args[3], args[4], args[5], args[6], args[7], args[8], args[9])