From 1ddb55bb4227d9673d23271aced5e215c3e1c5e7 Mon Sep 17 00:00:00 2001
From: npklein <niekdeklein@gamail.com>
Date: Mon, 10 Oct 2016 17:14:18 +0200
Subject: [PATCH 1/4] ConvertPLtoGL use gzipped vcf instead of vcf

---
 .../protocols/MergeFastq.sh                   | 67 +++++++++++++++++++
 .../workflowKallisto.csv                      |  3 +
 .../BIOS_phasing/protocols/ConvertPLtoGL.sh   | 52 ++++++++++++++
 3 files changed, 122 insertions(+)
 create mode 100755 compute5/Public_RNA-seq_quantification/protocols/MergeFastq.sh
 create mode 100644 compute5/Public_RNA-seq_quantification/workflowKallisto.csv
 create mode 100755 molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertPLtoGL.sh

diff --git a/compute5/Public_RNA-seq_quantification/protocols/MergeFastq.sh b/compute5/Public_RNA-seq_quantification/protocols/MergeFastq.sh
new file mode 100755
index 00000000..e48f943a
--- /dev/null
+++ b/compute5/Public_RNA-seq_quantification/protocols/MergeFastq.sh
@@ -0,0 +1,67 @@
+#MOLGENIS nodes=1 ppn=1 mem=10gb walltime=01:00:00
+
+### variables to help adding to database (have to use weave)
+#string sampleName
+#string project
+###
+#list reads1FqGz,reads2FqGz
+
+
+#Function to check if array contains value
+array_contains () {
+    local array="$1[@]"
+    local seeking=$2
+    local in=1
+    for element in "${!array-}"; do
+        if [[ "$element" == "$seeking" ]]; then
+            in=0
+            break
+        fi
+    done
+    return $in
+}
+
+
+echo "## "$(date)" Start $0"
+echo "ID (project-sampleName): ${${project}-${sampleName}"
+
+#check modules
+module list
+
+for file in "${reads1FqGz[@]}" "${reads2FqGz[@]}"; do
+    echo "getFile file='$file'"
+    getFile $file
+done
+
+
+#Create string with input fastq files to merge
+#This check needs to be performed because Compute generates duplicate values in array
+INPUTFQ1=()
+INPUTFQ2=()
+
+echo "merging"
+for fq in "${reads1FqGz[@]}"
+do
+   echo $fq
+   array_contains INPUTFQ1 "$fq" || INPUTFQ1+=("$fq")    # If fqFile does not exist in array add it
+done
+echo "done"
+for fq in "${reads2FqGz[@]}"
+do
+  echo $fq
+  array_contains INPUTFQ1 "$fq" || INPUTFQ2+=("$fq")    # If fqFile does not exist in array add it
+done
+
+echo "writing to $(dirname reads1FqGz[@])/${sampleName}_R2.fq.gz"
+
+if cat ${INPUTFQ1[*]} > $(dirname reads1FqGz[@])/${sampleName}_R1.fq.gz && cat ${INPUTFQ2[*]} > $(dirname reads1FqGz[@])/${sampleName}_R2.fq.gz
+then 
+  echo "returncode: $?"; putFile $(dirname reads1FqGz[@])/${sampleName}
+  putFile $(dirname reads1FqGz[@])/${sampleName}_R1.fq.gz
+  echo "succes moving files";
+else
+  echo "returncode: $?";
+  echo "fail";
+fi
+
+echo "## "$(date)" ##  $0 Done "
diff --git a/compute5/Public_RNA-seq_quantification/workflowKallisto.csv b/compute5/Public_RNA-seq_quantification/workflowKallisto.csv
new file mode 100644
index 00000000..18e02536
--- /dev/null
+++ b/compute5/Public_RNA-seq_quantification/workflowKallisto.csv
@@ -0,0 +1,3 @@
+step,protocol,dependencies
+MergeFastq,protocols/MergeFastq.sh,
+Kallisto,protocols/Kallisto.sh,MergeFastq
diff --git a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertPLtoGL.sh b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertPLtoGL.sh
new file mode 100755
index 00000000..205df01d
--- /dev/null
+++ b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertPLtoGL.sh
@@ -0,0 +1,52 @@
+#MOLGENIS walltime=23:59:00 mem=8gb nodes=1 ppn=2
+
+### variables to help adding to database (have to use weave)
+#string project
+###
+#string stage
+#string checkStage
+
+#string WORKDIR
+#string projectDir
+#string genotypedChrVcfGLDir
+#string genotypedChrVcfGL
+#string vcf
+#string biopythonVersion
+#string genotypedChrVcfGL
+#string ngsutilsVersion
+
+echo "## "$(date)" Start $0"
+
+getFile ${vcf}
+
+${stage} Biopython/${biopythonVersion}
+${stage} ngs-utils/${ngsutilsVersion}
+${checkStage}
+
+mkdir -p ${genotypedChrVcfGLDir}
+
+echo "Starting conversion."
+
+
+#Run conversion script beagle vcf to shapeit format
+if python $EBROOTNGSMINUTILS/PL_to_GL_reorder.py \
+    --vcf ${vcf} \
+    --out ${genotypedChrVcfGL}
+
+then
+ echo "returncode: $?";
+ putFile ${genotypedChrVcfGL}
+ cd ${genotypedChrVcfGLDir}
+ bname=$(basename ${genotypedChrVcfGL})
+ md5sum ${bname} > ${bname}.md5
+ cd -
+ echo "succes moving files";
+else
+ echo "returncode: $?";
+ echo "fail";
+fi
+
+echo "Finished conversion."
+
+echo "## "$(date)" ##  $0 Done "
+

From 5b198fc39cd91c1882819479f2b85a414d662425 Mon Sep 17 00:00:00 2001
From: npklein <niekdeklein@gamail.com>
Date: Tue, 11 Oct 2016 16:35:51 +0200
Subject: [PATCH 2/4] load GCC version without GCCXX error

---
 .../compute5/BIOS_phasing/parameters.csv      | 63 +++++++++++++++++++
 .../protocols/ConvertBeagleToShapeit.sh       | 62 ++++++++++++++++++
 2 files changed, 125 insertions(+)
 create mode 100755 molgenis-pipelines/compute5/BIOS_phasing/parameters.csv
 create mode 100755 molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh

diff --git a/molgenis-pipelines/compute5/BIOS_phasing/parameters.csv b/molgenis-pipelines/compute5/BIOS_phasing/parameters.csv
new file mode 100755
index 00000000..5cd0e7ad
--- /dev/null
+++ b/molgenis-pipelines/compute5/BIOS_phasing/parameters.csv
@@ -0,0 +1,63 @@
+########################################################################
+## General parameters for SLURM settings and cluster directory paths
+#
+queue,ll
+defaultInterpreter,#!/bin/bash
+stage,module load
+checkStage,module list
+WORKDIR,/groups/
+root,${WORKDIR}
+group,umcg-bios
+tmp,tmp04
+resDir,/groups/umcg-wijmenga/tmp04/resources/
+toolDir,/apps/software/
+projectDir,${root}/${group}/${tmp}/projects/${project}/
+fvdProjectDir,${root}/${group}/${tmp}/projects/umcg-fvandijk/projects/
+########################################################################
+## Software and data versions/builds/paths
+#
+beagleVersion,09Feb16.2b7-Java-1.8.0_45
+beagleJarVersion,09Feb16.2b7
+shapeitVersion,v2.r837-static
+GCCVersion,4.9.3-binutils-2.25
+intervaltreeVersion,2.1.0-foss-2015b-Python-2.7.9
+pyvcfVersion,0.6.7-foss-2015b-Python-2.7.9
+samtoolsVersion,1.2-foss-2015b
+bedtoolsVersion,2.23.0-foss-2015b
+pythonVersion,3.4.1-foss-2015b
+biopythonVersion,1.65-foss-2015b-Python-3.4.1
+ngsutilsVersion,16.06.1
+zlibVersion,1.2.8
+bzip2Version,1.0.6-foss-2015b
+GLibVersion,2.45.2-foss-2015b
+vcftoolsVersion,0.1.12b-goolf-1.7.20-Perl-5.20.2-bare
+RVersion,3.2.1-foss-2015b
+phaserVersion,f085550
+tabixVersion,0.2.6-goolf-1.7.20
+referenceFastaName,human_g1k_v37
+genomeBuild,b37
+onekgGenomeFasta,${resDir}/${genomeBuild}/indices/${referenceFastaName}.fasta
+geneticMapDir,/apps/data/www.shapeit.fr/genetic_map_b37/
+geneticMapChr,${geneticMapDir}/genetic_map_chr${chromosome}_combined_b37.txt
+OneKgPhase3VCF,/apps/data/1000G/release/20130502//ALL.wgs.phase3_shapeit2_mvncall_integrated_v5b.20130502.sites.vcf.gz
+########################################################################
+## Specific tools paths
+#
+## Input Beagle from GATK GenotypeGVF
+genotypedChrVcfGLDir,${projectDir}/genotypeVcfGL/
+genotypedChrVcfGL,${genotypedChrVcfGLDir}/${project}.chr${chromosome}.genotypeGVCF.gg.vcf.gz
+genotypedChrVcfTbi,${genotypedChrVcfGL}.tbi
+## Beagle
+beagleDir,${projectDir}/beagle/
+genotypedChrVcfBeagleGenotypeProbabilities,${beagleDir}/${project}.chr${chromosome}.beagle.genotype.probs.gg
+genotypedChrVcfShapeitInputPrefix,${beagleDir}/${project}.chr${chromosome}.beagle.genotype.probs.gg
+## Shapeit
+shapeitDir,${projectDir}/shapeit/
+phasedScaffoldDir,/groups/umcg-lld/tmp04/projects/genotypingRelease3/selectionLldeep/lldeepPhased/
+shapeitPhasedOutputPrefix,${shapeitDir}/${project}.chr${chromosome}.shapeit.phased
+## phASER
+mapq,0
+baseq,0
+phaserDir,${projectDir}/phASER
+## genotype concordance
+comparisonFileDir,${fvdProjectDir}RNA-seq_rare_variants/comparison_files/
diff --git a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
new file mode 100755
index 00000000..fe003a22
--- /dev/null
+++ b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
@@ -0,0 +1,62 @@
+#MOLGENIS walltime=23:59:00 mem=8gb nodes=1 ppn=2
+
+### variables to help adding to database (have to use weave)
+#string project
+###
+#string stage
+#string checkStage
+
+#string WORKDIR
+#string projectDir
+#string beagleDir
+#string genotypedChrVcfGL
+#string genotypedChrVcfBeagleGenotypeProbabilities
+#string genotypedChrVcfShapeitInputPrefix
+#string GLibVersion
+#string ngsutilsVersion
+#string zlibVersion
+#string bzip2Version
+
+
+echo "## "$(date)" Start $0"
+
+getFile ${genotypedChrVcfGL}
+getFile ${genotypedChrVcfBeagleGenotypeProbabilities}.vcf.gz
+
+${stage} ngs-utils/${ngsutilsVersion}
+${stage} GLib/${GLibVersion}
+${stage} zlib/${zlibVersion}
+${stage} bzip2/${bzip2Version}
+${stage} GCC/4.9.3-binutils-2.25
+${checkStage}
+
+#Run conversion script beagle vcf to shapeit format
+if $EBROOTNGSMINUTILS/prepareGenFromBeagle4_modified20160601/bin/prepareGenFromBeagle4 \
+ --likelihoods ${genotypedChrVcfGL} \
+ --posteriors ${genotypedChrVcfBeagleGenotypeProbabilities}.vcf.gz \
+ --threshold 0.995 \
+ --output ${genotypedChrVcfShapeitInputPrefix}
+then
+ echo "returncode: $?";
+ putFile ${genotypedChrVcfShapeitInputPrefix}.gen.gz
+ putFile ${genotypedChrVcfShapeitInputPrefix}.gen.sample
+ putFile ${genotypedChrVcfShapeitInputPrefix}.hap.gz
+ putFile ${genotypedChrVcfShapeitInputPrefix}.hap.sample
+ cd ${beagleDir}
+ bname=$(basename ${genotypedChrVcfShapeitInputPrefix}.gen.gz)
+ md5sum ${bname} > ${bname}.md5
+ bname=$(basename ${genotypedChrVcfShapeitInputPrefix}.gen.sample)
+ md5sum ${bname} > ${bname}.md5
+ bname=$(basename ${genotypedChrVcfShapeitInputPrefix}.hap.gz)
+ md5sum ${bname} > ${bname}.md5
+ bname=$(basename ${genotypedChrVcfShapeitInputPrefix}.hap.sample)
+ md5sum ${bname} > ${bname}.md5
+ cd -
+ echo "succes moving files";
+else
+ echo "returncode: $?";
+ echo "fail";
+fi
+
+echo "## "$(date)" ##  $0 Done "
+

From 4305fcb50764bef689b197baa41fbe9ba98657c5 Mon Sep 17 00:00:00 2001
From: npklein <niekdeklein@gamail.com>
Date: Tue, 11 Oct 2016 16:40:30 +0200
Subject: [PATCH 3/4] add comments

---
 .../compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh  | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
index fe003a22..e2502be8 100755
--- a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
+++ b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
@@ -27,7 +27,8 @@ ${stage} ngs-utils/${ngsutilsVersion}
 ${stage} GLib/${GLibVersion}
 ${stage} zlib/${zlibVersion}
 ${stage} bzip2/${bzip2Version}
-${stage} GCC/4.9.3-binutils-2.25
+# THIS NEEDS TO BE LOADED AFTER NGS-UTILS TO PREVENT GCCXX ERROR
+${stage} GCC/${GCCversion}
 ${checkStage}
 
 #Run conversion script beagle vcf to shapeit format

From adf14f5d72c57077548407007257698201e27485 Mon Sep 17 00:00:00 2001
From: npklein <niekdeklein@gamail.com>
Date: Tue, 11 Oct 2016 16:54:11 +0200
Subject: [PATCH 4/4] declare GCC version string

---
 .../compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh   | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
index e2502be8..ece87e39 100755
--- a/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
+++ b/molgenis-pipelines/compute5/BIOS_phasing/protocols/ConvertBeagleToShapeit.sh
@@ -16,7 +16,7 @@
 #string ngsutilsVersion
 #string zlibVersion
 #string bzip2Version
-
+#string GCCversion
 
 echo "## "$(date)" Start $0"