continued star wrapper

dge_workflow/dge_utils.sh

@@ -165,7 +165,7 @@ wait4jobs .tophatjobs
 
 
 
-dge_bam_correlate .
+dge_bam_correlate . &
 
 ## create tophat mapping report
 spin.R ${NGS_TOOLS}/dge_workflow/tophat_qc.R .

dge_workflow/star_align.sh

@@ -5,14 +5,23 @@
 
 usage='
 Use star to align fastq files against a genome
-Usage: star_align.sh <igenome> <fastq_files>...
+Usage: star_align.sh <igenome> <fastq_files>
 
 Options:
+-c        Cache results
 '
-#-c        Cache results
 
-eval $(echo "$usage" | ~/bin/docopts/docopts -h - -A dopts : "$@")
-#eval $(echo "$usage" | ~/bin/docopts/docopts -h -  : /projects/bioinfo/igenomes/Mus_musculus/Ensembl/GRCm38 /projects/bioinfo/holger/projects/florio_11b_2nd_batch/lanereps_pooled/arhgap11b_1.fastq.gz)
+#eval $(echo "$usage" | ~/bin/docopts/docopts -h - -A dopts : "$@")
+#echo "$usage" | ~/bin/docopts/docopts -h -  : "$@"
+#echo "$usage" | ~/bin/docopts/docopts -h - : "hallo dfds"
+#eval $(echo  $usage | ~/bin/docopts/docopts -h - : "$@")
+#eval(exit 64)
+eval "$(echo  "$usage" | ~/bin/docopts/docopts -h - : "$@")"
+
+# v0.7 style
+#eval $(echo  "$usage" | /home/brandl/bin/docopts_v0.7/docopts "hallo")
+
+#eval "$(echo "$usage" | ~/bin/docopts/docopts -h -  : /projects/bioinfo/igenomes/Mus_musculus/Ensembl/GRCm38 /projects/bioinfo/holger/projects/florio_11b_2nd_batch/lanereps_pooled/arhgap11b_1.fastq.gz)"
 
 #for fastqFile in ${fastq_files[@]} ; do
 #    echo processing $fastqFile
@@ -21,34 +30,33 @@ eval $(echo "$usage" | ~/bin/docopts/docopts -h - -A dopts : "$@")
 #echo $igenome
 
 
-## build index if not present
-
+fastqFiles=${fastq_files[@]}
+#echo $fastqFiles
 
-
-export star_index="$igenome/Sequence/StarIndex/genome"
+export star_index="${igenome}/Sequence/StarIndex"
 export gtfFile="$igenome/Annotation/Genes/genes.gtf"
-head $gtfFile
+#head $gtfFile
 
 if [ ! -f $gtfFile ]; then
     >&2 echo "gtf '$gtfFile' does not exis"; exit 1;
 fi
 
-## basic 
+## basic usage tutorial
 #http://www.homolog.us/blogs/blog/2012/11/02/star-really-kick-ass-rna-seq-aligner/
 
-if ! -d "${igenome}/Sequence/StarIndex" ]; then
+## build index if not present
+if ! -d "${star_index}" ]; then
 
 mailme "${project}: creating STAR index for $igenome"
-mkdir ${igenome}/Sequence/StarIndex
+mkdir ${star_index}
 
-cmd="STAR --runMode genomeGenerate --genomeDir ${igenome}/Sequence/StarIndex --genomeFastaFiles ${igenome}/Sequence/WholeGenomeFasta/genome.fa --runThreadN 10"
+cmd="STAR --runMode genomeGenerate --genomeDir ${star_index} --genomeFastaFiles ${igenome}/Sequence/WholeGenomeFasta/genome.fa --runThreadN 10"
 #eval $cmd
-#STAR --runMode genomeGenerate --genomeDir ${igenome}/Sequence/StarIndex --genomeFastaFiles ${igenome}/Sequence/Chromosomes/*.fa --runThreadN 10
+#STAR --runMode genomeGenerate --genomeDir ${star_index} --genomeFastaFiles ${igenome}/Sequence/Chromosomes/*.fa --runThreadN 10
 mysub "${project}_star_index" "$cmd" -n 5 -R span[hosts=1] -q medium  | blockScript
 
 ## prevent modification
-chmod -R -w ${igenome}/Sequence/StarIndex
-
+chmod -R -w ${star_index}
 fi
 
 
@@ -70,7 +78,7 @@ ll $fastqFiles
 for fastqFile in $fastqFiles ; do
     echo "submitting tophat job for $fastqFile"
 
-    # DEBUG fastqFile=/projects/bioinfo/holger/projects/helin/mouse/trimmed/mouse_big_cyst_rep4_ca.fastq.gz
+    # DEBUG fastqFile=/projects/bioinfo/holger/projects/florio_11b_2nd_batch/lanereps_pooled/arhgap11b_1.fastq.gz
     fastqBaseName=$(basename ${fastqFile%%.fastq.gz})
     outputdir=$fastqBaseName
 
@@ -80,7 +88,7 @@ for fastqFile in $fastqFiles ; do
     ## note --outSAMstrandField intronMotif is required for cuffdiff compatiblity (xs flag)
     mysub "${project}__star__${fastqBaseName}" "
 #    tophat -p6  -G $gtfFile -g1 -o $outputdir $bowtie_gindex $fastqFile
-    STAR --genomeDir $star_index --readFilesIn /path/to/read1 $fastqFile --runThreadN 6 --outFileNamePrefix $fastqBaseName --outSAMtype BAM SortedByCoordinate --outSAMstrandField intronMotif --sjdbGTFfile
+    STAR --genomeDir $star_index --readFilesIn $fastqFile --runThreadN 6 --readFilesCommand zcat --outFileNamePrefix $fastqBaseName --outSAMtype BAM SortedByCoordinate --outSAMstrandField intronMotif --sjdbGTFfile $gtfFile
     " -n 5 -R span[hosts=1] -q long | joblist .tophatjobs
 done