comparison region_motif_intersect.r @ 31:0e031bf8c5b1 draft

Uploaded

30:da0bc6dead0d

# Name: region_motif_intersect.r
# Description: Takes a bed file of target regions and counts intersections
# of each motif (built in rdata database) and target regions.
# Author: Jeremy liu
# Email: [e-mail]
# Date: 14/07/02
# Note: This script is meant to be invoked with the following command
# R --slave --vanilla -f ./region_motif_intersect.r --args <workingdir> <db> <inbed> <outtab>
# <workingdir> is working directory of galaxy installation
# <db> types: "t" test, "p" pouya, "j" jaspar jolma, "m" mouse
# Dependencies: none

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

# Set common and data directories
args <- commandArgs()
workingDir = args[7]
dbDir = concat(workingDir, "/region_motif_db")
dbCode = args[8]
if (dbCode == "t") {
  motifDB = concat(dbDir, "/pouya_test_motifs.bed.bgz")
} else if (dbCode == "p") {
  motifDB = concat(dbDir, "/pouya_motifs.bed.bgz")
} else if (dbCode == "j") {
  motifDB = concat(dbDir, "/jaspar_jolma_motifs.bed.bgz")
} else if (dbCode == "m") {
  motifDB = concat(dbDir, "/mm9_motifs.bed.bgz")
} else {
  motifDB = concat(dbDir, "/pouya_motifs.bed.bgz")
}

# Set input and reference files, comment to toggle commmand line arguments
inBed = args[9]
outTab = args[10]

suppressPackageStartupMessages(library(Rsamtools, quietly=TRUE))

# Initializing hash table (as env) with motif names and loading tabix file
cat("Loading motif database and initializing hash table...\n")
motifTable = new.env()
motifTbx <- TabixFile(motifDB)

# Loading input bed file, convert integer columns to numeric, name columns
cat("Loading region file...\n")
regionsDF = read_bed(inBed)
dfTemp = sapply(regionsDF, is.integer)

31:0e031bf8c5b1

# Name: region_motif_intersect.r
# Description: Takes a bed file of target regions and counts intersections
# of each motif (in separately installed tabix database) and target regions.
# 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_intersect.r --args <db_bgz> <db_tbi> <inbed> <outtab>
# Dependencies: region_motif_data_manager

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

# Retrive motif database path
args <- commandArgs()
motifDB_bgz = args[7]
motifDB_tbi = args[8]

# Set input and reference files, comment to toggle commmand line arguments
inBed = args[9]
outTab = args[10]

suppressPackageStartupMessages(library(Rsamtools, quietly=TRUE))

# Initializing hash table (as env) with motif names and loading tabix file
cat("Loading motif database and initializing hash table...\n")
motifTable = new.env()
motifTbx <- TabixFile(motifDB_bgz)

# Loading input bed file, convert integer columns to numeric, name columns
cat("Loading region file...\n")
regionsDF = read_bed(inBed)
dfTemp = sapply(regionsDF, is.integer)