Mercurial > repos > stef > qdnaseq

changeset 72:fb2dfd8ba417 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author stef
date Fri, 05 Sep 2014 11:10:27 -0400
parents a93649056969
children 4db025c51b6e
files QDNAseq.R QDNAseq.sh QDNAseq.xml static/binannotation/1000kbp_binAnnotations.rds static/binannotation/100kbp_binAnnotations.rds static/binannotation/15kbp_binAnnotations.rds static/binannotation/30kbp_binAnnotations.rds static/images/LGG150_copynumber.png static/images/LGG150_copynumberSegmented.png static/images/LGG150_copynumber_example.png tool_dependencies.xml
diffstat 11 files changed, 176 insertions(+), 17 deletions(-) [+]
line wrap: on
line diff
--- a/QDNAseq.R	Wed Aug 27 04:56:37 2014 -0400
+++ b/QDNAseq.R	Fri Sep 05 11:10:27 2014 -0400
@@ -38,6 +38,8 @@
 	l1 <- length(paths)
 	l2 <- length(names)
 	if ( l1 != l2 ) stop( "Unequal amount of bam-paths (", l1, ") and -names (", l2, ") in makeBamFileList!!!\n" )
+	if ( l1 == 0 ){ return(tmp) } # empty list in debug mode
+
 	for ( i in 1:length(paths) ){
 		path <- paths[i]
 		name <- names[i]
@@ -155,9 +157,6 @@
 
 ## desparate tries to make png text scale well, damn you R...!
 PLOT_RES  <- min( PLOT_WIDTH, PLOT_HEIGHT ) / 6.3 
-#PLOT_PS <- min( PLOT_WIDTH, PLOT_HEIGHT ) / 40
-#TXT_FACTOR <- PLOT_WIDTH/480
-#PAR_SET   <- list( pch=22, cex.lab=TXT_FACTOR, cex.main=TXT_FACTOR )
 PAR_SET   <- list( pch=22 )
 
 systemUser <- system("whoami",T)
@@ -176,8 +175,8 @@
 ## ------------------------
 ## DEBUG
 if ( debug ){
-	catMsg( c("Analysis run under user: ", systemUser ) )
-	catMsg( c("Here comes sessionInfo: " ) )
+	catMsg( c("Analysis run by user: ", systemUser ) )
+	catMsg( c("DEBUG SessionInfo: " ) )
 	print( sessionInfo() )
 }
 ## /DEBUG
@@ -189,7 +188,9 @@
 }
 
 ## copy source config file to output dir to include it in output zip
-file.copy( config, paste(outputPath, 'galaxyConfigFile.R', sep='/') )
+if ( inGalaxy ){
+	file.copy( config, paste(outputPath, 'galaxyConfigFile.R', sep='/') )	
+}
 
 ## setup bam filelist for easy retrieval later
 fileList    <- makeBamFileList( bamsPaths, bamsNames )
@@ -233,7 +234,13 @@
 	igvCopyNrPath  <- paste( outputPath, '/', igvCopyNrName, sep='')
 	noiseImgPath   <- paste( outputPath, '/', noiseImgName, sep='')
 
-	binAnnotations <- getBinAnnotations( binSize=binSize, type=experimentType )
+	binAnnFile <- paste( TOOL_PATH, '/static/binannotation/', binSize, 'kbp_binAnnotations.rds', sep="" )
+	if ( file.exists(binAnnFile) ){
+		binAnnotations <- readRDS( binAnnFile )
+		catMsg( c("Using local binAnnotations file" ) )
+	}else{
+		binAnnotations <- getBinAnnotations( binSize=binSize, type=experimentType )
+	}
 
 	## in case of debug just use inbuilt LGG data for speedup
 	if ( debug ){
@@ -493,8 +500,9 @@
 ## ------------------------
 ## creating main html output for galaxy history
 ## ------------------------
-sink( file = outputHtml, type = "output" )
-		
+if ( inGalaxy ){ # dont create when running outside Galaxy
+	sink( file = outputHtml, type = "output" )
+			
 		cat( "<head>", "\n")
 			cat( "\t", '<link rel="stylesheet" href="', PURE_CSS, '">', "\n", sep='' )
 
@@ -509,7 +517,8 @@
 		cat( '<p>Explore <a href="', htmlOutputName, '" class="button">the results</a> directly within galaxy</p>', "\n", sep="")
 		cat( '<p>Or download a <a href="', gzipOutputName, '" class="button">zipfile</a> with all output (', zippedSize, ')</p>', "\n", sep="" )
 
-sink()
+	sink()
+}
 
 ## ------------------------
 ## create final zip and quit with status 0 to tell galaxy all was fine
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/QDNAseq.sh	Fri Sep 05 11:10:27 2014 -0400
@@ -0,0 +1,148 @@
+#!/bin/sh
+
+## This script creates a config file for use with the QDNAseq R wrapper
+## For use outside Galaxy
+
+QDNASEQ_R_SCRIPT='/ccagc/home/stef/code/QDNASEQ/QDNAseq.R'
+R_LOCATION='/ccagc/lib/R/R-3.1.0/bin/'
+OUTPUT_DIR='QDNASEQ'
+SCRIPT_DIR=$(dirname $0)
+SCRIPT_NAME=$(basename $0)
+
+binSizesString='1000,100,15'
+outputName=`basename $PWD`
+doSegment='TRUE'
+debug='FALSE'
+
+usage()
+{
+cat<<EOF
+  This script creates a QDNAseq wrapper config.
+  
+  Experiment name and/or output dir are optional
+    $SCRIPT_NAME [-n myExperiment] [-o outputDir] *.bam
+
+  Debug mode for working with build in LGG150 data
+    $SCRIPT_NAME -d
+
+  OPTIONS:
+    -n [s]  provide analysis name [$outputName]
+    -o [s]  output dir [$OUTPUT_DIR]
+    -d      debug mode (only create config)
+    
+EOF
+}
+
+PARAM_COUNT=0
+if [ $# -eq 0 ] || [ "$1" == "-h" ]
+then
+	usage
+	exit 1
+else
+	while getopts "n:o:d" opt; do
+	  case $opt in
+	    n)
+	      #echo "$opt was triggered, Parameter: $OPTARG" >&2
+	      outputName=$OPTARG
+	      PARAM_COUNT=`bc <<< $PARAM_COUNT+2`
+	      ;;
+	    o)
+	      OUTPUT_DIR=$OPTARG
+	      PARAM_COUNT=`bc <<< $PARAM_COUNT+2`
+	      ;;
+	    d)
+	      debug='TRUE'
+	      PARAM_COUNT=`bc <<< $PARAM_COUNT+1`
+	      ;;
+	    \?)
+	      echo "Invalid option: -$OPTARG" >&2
+	      exit 1
+	      ;;
+	    :)
+	      echo "Option -$OPTARG requires an argument." >&2
+	      exit 1
+	      ;;
+	  esac
+	done
+fi
+
+## set param dependent variables
+printf '%s\n' "[INFO] --- QDNAseq wrapper config creator ---"
+CONFIG_FILE=$OUTPUT_DIR'/qdnaseqConfig.R'
+
+## remove the params from input string leaving only the BAMs
+shift $PARAM_COUNT
+
+## sanity checks
+if [ -e $CONFIG_FILE ]; then
+	echo "[ERR] Config file ($CONFIG_FILE) already exists, exit.."; 
+	exit 0
+fi
+if [ -d $OUTPUT_DIR ]; then
+	echo "[ERR] Output dir ($OUTPUT_DIR) already exists, exit.."; 
+	exit 0
+else
+	mkdir $OUTPUT_DIR
+fi
+
+## variables setup
+CONFIG_TXT="## ==========
+## QDNAseq pipeline config file
+## ==========
+
+## ----------
+inGalaxy <- FALSE
+binSizesString <- '$binSizesString'
+experimentType <- 'SR50'
+outputName <- '$outputName'
+
+## ----------
+outputHtml  <- 'galaxyIndex.html'
+outputId    <- NA
+newFilePath <- '$OUTPUT_DIR'
+outputPath  <- '$OUTPUT_DIR'
+doSegment   <- as.logical( $doSegment )
+debug       <- as.logical( $debug )
+undoSD      <- as.double( 1.0 )
+binAnnotations <- ''
+
+## ----------
+filterBlacklistedBins <- as.logical( 'TRUE' )
+mappabilityCutoff <- as.integer( 0 )
+undoSplits <- 'sdundo'
+doOutputCopynumbersIgv <- FALSE
+
+## ----------
+PLOT_WIDTH <- as.integer( 1440 )
+PLOT_HEIGHT <- as.integer( 720 )
+excludeChrsString <- 'X,Y'
+
+## ----------
+bamsPaths <- c()
+bamsNames <- c()
+"
+
+## create config file
+printf '%s\n' "$CONFIG_TXT" > $CONFIG_FILE
+
+## add bam files to config file
+BAMS=$@
+for bam_path in $BAMS; do
+	
+	if [ ! -e $bam_path  ]; then
+		echo "[ERR] Bam file does not exist ($bam_path), exit.."; 
+		exit 0
+	fi
+
+	bam_name=`basename "$bam_path"`
+	printf '%s\n' "bamsPaths <- c( bamsPaths, \"$bam_path\" )" >> $CONFIG_FILE
+	printf '%s\n' "bamsNames <- c( bamsNames, \"$bam_name\" )" >> $CONFIG_FILE
+	printf '%s\n' "[INFO] BAM: $bam_name"
+done
+
+## done
+printf '%s\n' "[INFO] OUTPUT DIR:  $OUTPUT_DIR"
+printf '%s\n' "[INFO] OUTPUT NAME: $outputName"
+printf '%s\n' "[INFO] CONFIG FILE can be found in $CONFIG_FILE"
+printf '%s\n' "[INFO] Run with: $R_LOCATION/Rscript $QDNASEQ_R_SCRIPT $CONFIG_FILE"
+printf '%s\n' "[INFO] Done"
\ No newline at end of file
--- a/QDNAseq.xml	Wed Aug 27 04:56:37 2014 -0400
+++ b/QDNAseq.xml	Fri Sep 05 11:10:27 2014 -0400
@@ -10,7 +10,6 @@
 
   <description>Quantitative copy number abberation detection</description>
 
-  <!-- command block to call script with default R installation -->
   <!-- change to /full/path/to/Rscript if required (eg /ccagc/lib/R/R-3.1.0/bin/Rscript) -->
   <command interpreter="Rscript"> 
     QDNAseq.R 
@@ -38,9 +37,9 @@
     <!-- Bin Size: only certain sizes are supported by QDNAseq package -->
     <param name="binSizes" type="select" optional="false" multiple="true" label="Select bin-sizes to use (kb)" help="Larger bin sizes provide faster analysis but lower resolution">
       <option value="1000" selected="true">1Mb</option>
-      <option value="100">100kb</option>
+      <option value="100" selected="true">100kb</option>
       <option value="30">30kb</option>
-      <option value="15">15kb</option>
+      <option value="15" selected="true">15kb</option>
       <option value="5">5kb</option>
       <option value="1">1kb</option>
     </param>
@@ -166,6 +165,7 @@
 ## -----
 ## required params
 ## -----
+TRUE -> inGalaxy 
 "${binSizes}" -> binSizesString
 "${experimentType}" -> experimentType
 "${jobName}" -> outputName
Binary file static/binannotation/1000kbp_binAnnotations.rds has changed
Binary file static/binannotation/100kbp_binAnnotations.rds has changed
Binary file static/binannotation/15kbp_binAnnotations.rds has changed
Binary file static/binannotation/30kbp_binAnnotations.rds has changed
Binary file static/images/LGG150_copynumber.png has changed
Binary file static/images/LGG150_copynumberSegmented.png has changed
Binary file static/images/LGG150_copynumber_example.png has changed
--- a/tool_dependencies.xml	Wed Aug 27 04:56:37 2014 -0400
+++ b/tool_dependencies.xml	Fri Sep 05 11:10:27 2014 -0400
@@ -5,19 +5,21 @@
         <repository changeset_revision="c0f72bdba484" name="package_samtools_0_1_18" owner="devteam" toolshed="https://testtoolshed.g2.bx.psu.edu" prior_installation_required="TRUE" />
     </package>
 
-    <!-- As soon as R 3_1_0 package works, all QDNAseq dependencies will be set via "package_qdnaseq_1_0_5" -->
-    <!-- Now this depency has to be manually installed by installing Bioconductor package "QDNAseq" -->
+    <!-- As soon as R 3_1_0 package works good, all dependencies will be set via "package_qdnaseq_1_0_5" -->
+    <!-- Currently these have to be manually installed by installing Bioconductor package "QDNAseq" -->
+    
     <!--
     <package name="R" version="3.1.0">
         <repository changeset_revision="a6cc7706ea14" name="package_r_3_1_0" owner="iuc" toolshed="https://testtoolshed.g2.bx.psu.edu" prior_installation_required="TRUE" />
     </package>
-    
+    -->
+
+    <!--
     <package name="qdnaseq" version="1.0.5">
         <repository name="package_qdnaseq_1_0_5" owner="stef"/>
     </package>
     -->
 
-    <!-- environment variable not in use now -->
     <!--
     <action type="set_environment">
         <environment_variable name="PATH" action="prepend_to">$INSTALL_DIR</environment_variable>