comparison region_motif_compare.r @ 35:4ce22698acb0 draft

Uploaded

34:fe8a884363df

# Description: Reads in two count files and determines enriched and depleted
# motifs (or any location based feature) based on poisson tests and gc
# corrections. All enrichment ratios relative to overall count / gc ratios.
# Author: Jeremy liu
# Email: [e-mail]
# Date: 15/02/04
# Note: This script is meant to be invoked with the following command
# R --slave --vanilla -f ./region_motif_compare.r --args <workingdir> <db> <intab1> <intab2> 
#   <enriched_tab> <depleted_tab> <plots_png>
# <workingdir> is working directory of galaxy installation
# <db> types: "t" test, "p" pouya, "j" jaspar jolma, "m" mouse, "c" combined
# Dependencies: plotting.r

# Auxiliary function to concatenate multiple strings
concat <- function(...) {
	input_list <- list(...)
	return(paste(input_list, sep="", collapse=""))

options(warn=-1)

# Set common and data directories
args <- commandArgs()
workingDir = args[7]
dbCode = args[8]
# dbCode "c" implemented when pwmFile is loaded
if (dbCode == "t" | dbCode == "p") {
	pwmFile = concat(workingDir, "/pwms/pouya.pwms.from.seq.RData")
} else if (dbCode == "j") {
	pwmFile = concat(workingDir, "/pwms/jaspar.jolma.pwms.from.seq.RData")
} else if (dbCode == "m") {
	pwmFile = concat(workingDir, "/pwms/mm9.pwms.from.seq.RData")
} else if (dbCode == "c") { # rest of dbCode "c" implemented when pwmFile loaded
	pwmFile = concat(workingDir, "/pwms/pouya.pwms.from.seq.RData")
	pwmFile2 = concat(workingDir, "/pwms/jaspar.jolma.pwms.from.seq.RData")
} else {
	pwmFile = concat(workingDir, "/pwms/pouya.pwms.from.seq.RData")
}

# Set input and reference files
inTab1 = args[9]
inTab2 = args[10]
enrichTab = args[11]

cat("Running ... Started at:", format(startTime, "%a %b %d %X %Y"), "...\n")

# Loading motif position weight matrix (pwm) file and input tab file
cat("Loading and reading input region motif count files...\n")
load(pwmFile) # pwms data structure
if (dbCode == "c") { # Remaining implementation of dbCode "c" combined 
	temp = pwms
	load(pwmFile2)
	pwms = append(temp, pwms)
}
region1DF = read_tsv(inTab1)
region2DF = read_tsv(inTab2)
region1Counts = region1DF$counts
region2Counts = region2DF$counts
names(region1Counts) = region1DF$motif

35:4ce22698acb0

# Description: Reads in two count files and determines enriched and depleted
# motifs (or any location based feature) based on poisson tests and gc
# corrections. All enrichment ratios relative to overall count / gc ratios.
# Author: Jeremy liu
# Email: [e-mail]
# Date: 15/02/11
# Note: This script is meant to be invoked with the following command
# R --slave --vanilla -f ./region_motif_compare.r --args <workingdir> <pwm_file> 
#        <intab1> <intab2> <enriched_tab> <depleted_tab> <plots_png>
# <workingdir> is working directory of galaxy installation
# Dependencies: region_motif_data_manager, plotting.r

# Auxiliary function to concatenate multiple strings
concat <- function(...) {
	input_list <- list(...)
	return(paste(input_list, sep="", collapse=""))

options(warn=-1)

# Set common and data directories
args <- commandArgs()
workingDir = args[7]
pwmFile = args[8].split(',')[0] # If duplicate entires, take first one

# Set input and reference files
inTab1 = args[9]
inTab2 = args[10]
enrichTab = args[11]

cat("Running ... Started at:", format(startTime, "%a %b %d %X %Y"), "...\n")

# Loading motif position weight matrix (pwm) file and input tab file
cat("Loading and reading input region motif count files...\n")
load(pwmFile) # pwms data structure
#if (dbCode == "c") { # Remaining implementation of dbCode "c" combined 
#	temp = pwms
#	load(pwmFile2)
#	pwms = append(temp, pwms)
#}
region1DF = read_tsv(inTab1)
region2DF = read_tsv(inTab2)
region1Counts = region1DF$counts
region2Counts = region2DF$counts
names(region1Counts) = region1DF$motif