Merge pull request #10 from sb43/dev

Dev

Author: sb43

.gitignore

@@ -0,0 +1 @@
+cgp_downsampler

cgp_downsampler/bpipe.config

@@ -0,0 +1,31 @@
+//read config file
+def config = new ConfigSlurper().parse(new File('cgp_downsampler.prm').toURL())
+props = config.toProperties()
+
+LSF_JOB_NAME=props.LSF_JOB_NAME
+
+LSF_Q_DODWONSAMPLING=props.LSF_Q_DODWONSAMPLING
+LSF_MEMORY_DODWONSAMPLING=props.LSF_MEMORY_DODWONSAMPLING
+LSF_ARRAY_START_DODWONSAMPLING=props.LSF_ARRAY_START_DODWONSAMPLING
+LSF_ARRAY_END_DODWONSAMPLING=props.LSF_ARRAY_END_DODWONSAMPLING
+LSF_THREADS_DODWONSAMPLING=props.LSF_THREADS_DODWONSAMPLING
+
+LSF_Q_BWAMEM=props.LSF_Q_BWAMEM
+LSF_MEMORY_BWAMEM=props.LSF_MEMORY_BWAMEM
+LSF_ARRAY_START_BWAMEM=props.LSF_ARRAY_START_BWAMEM
+LSF_ARRAY_END_BWAMEM=props.LSF_ARRAY_END_BWAMEM
+LSF_THREADS_BWAMEM=props.LSF_THREADS_BWAMEM
+//def JOBID = System.getenv('LSB_JOBID');
+executor="lsf"
+procs="1"
+
+commands {
+  lsf_DODWONSAMPLING {
+  procs="$LSF_THREADS_DODWONSAMPLING"
+    lsf_request_options="  -P analysis-cgp -q $LSF_Q_DODWONSAMPLING -R'select[mem>$LSF_MEMORY_DODWONSAMPLING]  span[hosts=1] rusage[mem=$LSF_MEMORY_DODWONSAMPLING]' -M$LSF_MEMORY_DODWONSAMPLING -o ${LSF_JOB_NAME}_DODWONSAMPLING.%I.canPipe.log -J ${LSF_JOB_NAME}_DODWONSAMPLING'[$LSF_ARRAY_START_DODWONSAMPLING-$LSF_ARRAY_END_DODWONSAMPLING]'"
+  }
+  lsf_BWAMEM {
+    procs="$LSF_THREADS_BWAMEM"
+    lsf_request_options=" -P analysis-cgp  -q $LSF_Q_BWAMEM -R'select[mem>=$LSF_MEMORY_BWAMEM] span[hosts=1] rusage[mem=$LSF_MEMORY_BWAMEM]' -M$LSF_MEMORY_BWAMEM -o ${LSF_JOB_NAME}_BWAMEM.%I.canPipe.log -J ${LSF_JOB_NAME}_BWAMEM'[$LSF_ARRAY_START_BWAMEM-$LSF_ARRAY_END_BWAMEM]'"
+  }
+}

cgp_downsampler/cgp_downsampler.prm

@@ -0,0 +1,20 @@
+CANPIPE_LIVE_PATH="/software/CGP/canpipe/test/bin/canpipe_test"
+OUTDIR="~/sb43_scratch116/rt_user/downsampling_564469/scripts/cgp_downsampler/downsampledBams"
+GENOME_SIZE="3137454505"
+COVERAGE_X="17"
+SEED="${GENOME_SIZE}"
+REFERENCE="/lustre/scratch112/sanger/cgppipe/canpipe/live/ref/human/GRCh37d5/genome.fa"
+LSF_JOB_NAME="downsampler"
+
+LSF_Q_DODWONSAMPLING="normal"
+LSF_MEMORY_DODWONSAMPLING="12000"
+LSF_ARRAY_START_DODWONSAMPLING="1"
+LSF_ARRAY_END_DODWONSAMPLING="1"
+LSF_THREADS_DODWONSAMPLING="1"
+
+LSF_Q_BWAMEM="long"
+LSF_MEMORY_BWAMEM="12000"
+LSF_ARRAY_START_BWAMEM="1"
+LSF_ARRAY_END_BWAMEM="1"
+LSF_THREADS_BWAMEM="6"
+

cgp_downsampler/cgp_downsampler.run

@@ -0,0 +1,5 @@
+load 'cgp_downsampler.wfl'
+
+Bpipe.run{
+ '%.bam' * [do_downsampling+bwa_mem]
+}

cgp_downsampler/cgp_downsampler.wfl

@@ -0,0 +1,29 @@
+load 'cgp_downsampler.prm'
+about title: "CGP downsampling workflow"
+
+do_downsampling = {
+  doc title: " downsample data ",
+  desc:  " downsample bam to specified depth x",
+  constraints: "N/A",
+  author: "[email protected]"
+	exec """ mkdir -p $OUTDIR && 
+           DS_FACTOR=\$( awk -v bases=${GENOME_SIZE} '(\$0!~/#/) {sum=sum+\$9} END { print (${COVERAGE_X}+1)/(sum/bases)}' ${input}.bas) &&
+					 SAMPLE=\$(awk '(NR==2) {print \$2}' ${input}.bas) &&
+					 $CANPIPE_LIVE_PATH bamdownsamplerandom seed=$SEED p=$DS_FACTOR I=${input} level=1 O=$OUTDIR/${SAMPLE}.bam && 
+					 date >$output.track
+			""", """lsf_DODWONSAMPLING"""
+			forward input
+}
+
+bwa_mem = {
+        doc title: "bwa_mem - bwa_mem alignment - mapping",
+            desc:  "bwa_mem  step for bwa_mem.pl",
+                        constraints: "N/A",
+                        author: "[email protected]"
+exec """ SAMPLE=\$(awk '(NR==2) {print \$2}' ${input}.bas) &&
+         mkdir -p $OUTDIR/$SAMPLE &&
+				 bwa_mem.pl -b '-T 30 -Y' -mt $LSF_THREADS_BWAMEM -o $OUTDIR/$SAMPLE -r $REFERENCE -t $LSF_THREADS_BWAMEM  -s ${SAMPLE} $OUTDIR/${SAMPLE}.bam && 
+				 rm $OUTDIR/${SAMPLE}.bam && 
+				 date >$output.track
+""", """lsf_BWAMEM"""
+}