diff --git a/workflow/conf/biohpc.config b/workflow/conf/biohpc.config
index 20da91a7f7a241e610708d7186d299d397958c41..11d83b277b06efc7920c5afbe2f09240492f27e6 100755
--- a/workflow/conf/biohpc.config
+++ b/workflow/conf/biohpc.config
@@ -5,14 +5,14 @@ process {
withName:getBag {
executor = 'local'
- container = 'docker://bicf/gudmaprbkfilexfer:1.3'
}
withName:getData {
executor = 'local'
- container = 'docker://bicf/gudmaprbkfilexfer:1.3'
}
withName:trimData {
- container = 'docker://bicf/trimgalore:1.1'
+ queue = '256GB,256GBv1,384GB'
+ }
+ withName:alignReads {
queue = '256GB,256GBv1,384GB'
}
}
@@ -26,4 +26,4 @@ env {
http_proxy = 'http://proxy.swmed.edu:3128'
https_proxy = 'http://proxy.swmed.edu:3128'
all_proxy = 'http://proxy.swmed.edu:3128'
-}
\ No newline at end of file
+}
diff --git a/workflow/nextflow.config b/workflow/nextflow.config
index 37584999cf8152c9776b676d0b013f8aeb5e8709..4f4f34d051a54568aabee33754ba02fd8f30a67d 100644
--- a/workflow/nextflow.config
+++ b/workflow/nextflow.config
@@ -10,6 +10,21 @@ profiles {
}
}
+process {
+ withName:getBag {
+ container = 'bicf/gudmaprbkfilexfer:1.3'
+ }
+ withName:getData {
+ container = 'bicf/gudmaprbkfilexfer:1.3'
+ }
+ withName:trimData {
+ container = 'bicf/trimgalore:1.1'
+ }
+ withName: alignReads {
+ container = 'bicf/gudmaprbkaligner:2.0.0'
+ }
+}
+
trace {
enabled = true
file = 'pipeline_trace.txt'
@@ -37,4 +52,4 @@ manifest {
mainScript = 'rna-seq.nf'
version = 'v0.0.1_indev'
nextflowVersion = '>=19.09.0'
-}
\ No newline at end of file
+}
diff --git a/workflow/rna-seq.nf b/workflow/rna-seq.nf
index c1d72efac168ca372471a35e4baa9ccfd96aae18..8b7823a0565d860725c60c6a134f179a6561ffe9 100755
--- a/workflow/rna-seq.nf
+++ b/workflow/rna-seq.nf
@@ -5,6 +5,8 @@ params.deriva = "${baseDir}/../test_data/credential.json"
params.bdbag = "${baseDir}/../test_data/cookies.txt"
//params.repRID = "16-1ZX4"
params.repRID = "Q-Y5JA"
+params.spikein = "false"
+params.reference = "mouse"
params.outDir = "${baseDir}/../output"
@@ -16,15 +18,25 @@ bdbag = Channel
.fromPath(params.bdbag)
.ifEmpty { exit 1, "deriva cookie file for bdbag not found: ${params.bdbag}" }
-Channel.from(params.repRID)
- .into {
- repRID_getBag
- repRID_getData
- repRID_trimData
- }
-
+repRID = params.repRID
outDir = params.outDir
logsDir = "${outDir}/Logs"
+reference = params.reference
+
+if (params.spikein) {
+ if (params.reference == "human") {
+ reference = file ("/project/BICF/BICF_Core/s181706/github/gudmap/rna-seq/References/GRCh38.p12-S/hisat2")
+ } else if (params.reference == "mouse") {
+ reference = file ("/project/BICF/BICF_Core/s181706/github/gudmap/rna-seq/References/GRCm38.P6-S/hisat2")
+ }
+} else if (params.reference == "mouse") {
+ reference = file ("/project/BICF/BICF_Core/s181706/github/gudmap/rna-seq/References/GRCm38.P6/hisat2")
+} else if (params.reference == "human") {
+ reference = file ("/project/BICF/BICF_Core/s181706/github/gudmap/rna-seq/References/GRCh38.p12/hisat2")
+} else {
+ print ("Warning: Reference genome not specified, defaulting to GRCm38.P6 with NO spike-in")
+ reference = file ("/project/BICF/BICF_Core/s181706/github/gudmap/rna-seq/References/GRCm38.P6/hisat2")
+}
// Define fixed files
derivaConfig = Channel.fromPath("${baseDir}/conf/replicate_export_config.json")
@@ -34,26 +46,25 @@ derivaConfig = Channel.fromPath("${baseDir}/conf/replicate_export_config.json")
*/
process getBag {
executor 'local'
- tag "${repRID_getBag}"
- publishDir "${logsDir}/getBag", mode: 'symlink', pattern: "${repRID_getBag}.getBag.err"
+ tag "${repRID}"
+ publishDir "${logsDir}/getBag", mode: 'copy', pattern: "${repRID}.getBag.err"
input:
- val repRID_getBag
path credential, stageAs: 'credential.json' from deriva
path derivaConfig
output:
path ("Replicate_*.zip") into bagit
- file ("${repRID_getBag}.getBag.err")
+ file ("${repRID}.getBag.err")
script:
"""
- hostname >>${repRID_getBag}.getBag.err
- ulimit -a >>${repRID_getBag}.getBag.err
+ hostname >>${repRID}.getBag.err
+ ulimit -a >>${repRID}.getBag.err
export https_proxy=\${http_proxy}
- ln -sf `readlink -e credential.json` ~/.deriva/credential.json 2>>${repRID_getBag}.getBag.err
- echo "LOG: deriva credentials linked" >>${repRID_getBag}.getBag.err
- deriva-download-cli dev.gudmap.org --catalog 2 ${derivaConfig} . rid=${repRID_getBag} 2>>${repRID_getBag}.getBag.err
+ ln -sf `readlink -e credential.json` ~/.deriva/credential.json 2>>${repRID}.getBag.err
+ echo "LOG: deriva credentials linked" >>${repRID}.getBag.err
+ deriva-download-cli dev.gudmap.org --catalog 2 ${derivaConfig} . rid=${repRID} 2>>${repRID}.getBag.err
"""
}
@@ -61,11 +72,10 @@ process getBag {
* getData: fetch study files from consortium with downloaded bdbag.zip
*/
process getData {
- tag "${repRID_getData}"
- publishDir "${logsDir}/getData", mode: 'symlink', pattern: "${repRID_getData}.getData.err"
+ tag "${repRID}"
+ publishDir "${logsDir}/getData", mode: 'copy', pattern: "${repRID}.getData.err"
input:
- val repRID_getData
executor 'local'
path cookies, stageAs: 'deriva-cookies.txt' from bdbag
path bagit
@@ -75,22 +85,22 @@ process getData {
file("**/File.csv") into fileMeta
file("**/Experiment Settings.csv") into experimentSettingsMeta
file("**/Experiment.csv") into experimentMeta
- file ("${repRID_getData}.getData.err")
+ file ("${repRID}.getData.err")
script:
"""
- hostname >>${repRID_getData}.getData.err
- ulimit -a >>${repRID_getData}.getData.err
+ hostname >>${repRID}.getData.err
+ ulimit -a >>${repRID}.getData.err
export https_proxy=\${http_proxy}
- ln -sf `readlink -e deriva-cookies.txt` ~/.bdbag/deriva-cookies.txt >>${repRID_getData}.getData.err
- echo "LOG: deriva cookie linked" >>${repRID_getData}.getData.err
+ ln -sf `readlink -e deriva-cookies.txt` ~/.bdbag/deriva-cookies.txt >>${repRID}.getData.err
+ echo "LOG: deriva cookie linked" >>${repRID}.getData.err
replicate=\$(basename "${bagit}" | cut -d '.' -f1)
- echo "LOG: \${replicate}" >>${repRID_getData}.getData.err
- unzip ${bagit} 2>>${repRID_getData}.getData.err
- echo "LOG: replicate bdbag unzipped" >>${repRID_getData}.getData.err
- sh ${baseDir}/scripts/bdbagFetch.sh \${replicate} ${repRID_getData} 2>>${repRID_getData}.getData.err
- echo "LOG: replicate bdbag fetched" >>${repRID_getData}.getData.err
+ echo "LOG: \${replicate}" >>${repRID}.getData.err
+ unzip ${bagit} 2>>${repRID}.getData.err
+ echo "LOG: replicate bdbag unzipped" >>${repRID}.getData.err
+ sh ${baseDir}/scripts/bdbagFetch.sh \${replicate} ${repRID} 2>>${repRID}.getData.err
+ echo "LOG: replicate bdbag fetched" >>${repRID}.getData.err
"""
}
@@ -98,34 +108,54 @@ process getData {
* trimData: trims any adapter or non-host sequences from the data
*/
process trimData {
- tag "${repRID_trimData}"
- publishDir "${logsDir}/trimData", mode: 'symlink', pattern: "\${repRID_trimData}.trimData.*"
+ tag "${repRID}"
+ publishDir "${logsDir}/trimData", mode: 'copy', pattern: "\${repRID}.trimData.*"
input:
- val repRID_trimData
file(fastq) from fastqs
output:
- path ("*.fq.gz") into fastqs_trimmed
- val ends
- file ("${repRID_trimData}.trimData.log")
- file ("${repRID_trimData}.trimData.err")
+ tuple file ("params.csv"), file ("*.fq.gz") into fastqs_trimmed
+ file ("${repRID}.trimData.log")
+ file ("${repRID}.trimData.err")
script:
"""
- if [ `nproc` -gt 8 ]
- then
- ncore=8
- else
- ncore=`nproc`
- fi
if [ '${fastq[1]}' == 'null' ]
then
ends='se'
- trim_galore --gzip -q 25 --illumina --length 35 --basename ${repRID_trimData} -j \${ncore} ${fastq[0]} 1>>${repRID_trimData}.trimData.log 2>>${repRID_trimData}.trimData.err;
+ trim_galore --gzip -q 25 --illumina --length 35 --basename ${repRID} -j `nproc` ${fastq[0]} 1>>${repRID}.trimData.log 2>>${repRID}.trimData.err;
else
ends='pe'
- trim_galore --gzip -q 25 --illumina --length 35 --paired --basename ${repRID_trimData} -j \${ncore} ${fastq[0]} ${fastq[1]} 1>>${repRID_trimData}.trimData.log 2>>${repRID_trimData}.trimData.err;
+ trim_galore --gzip -q 25 --illumina --length 35 --paired --basename ${repRID} -j `nproc` ${fastq[0]} ${fastq[1]} 1>>${repRID}.trimData.log 2>>${repRID}.trimData.err;
fi
+ echo \${ends} > params.csv;
"""
-}
\ No newline at end of file
+}
+
+/*
+ * alignReads: aligns the reads to a reference database
+*/
+process alignReads {
+ tag "align-${repRID}"
+ publishDir "${outDir}/tempOut/aligned", mode: "copy"
+
+ input:
+ set ends, fqs from fastqs_trimmed
+ file reference
+
+ output:
+ set repRID, file ("${repRID}.unal.gz"), file ("${repRID}.bam"), file ("${repRID}.bai")
+
+ script:
+ """
+ ends=`cat ${ends}`;
+ if [ "\${ends}" == 'pe' ]; then
+ hisat2 -p `nproc` --add-chrname --un-gz ${repRID}.unal.gz -S ${repRID}.sam -x ${reference}/genome -1 ${fqs[0]} -2 ${fqs[1]} 1>${repRID}.align.out 2> ${repRID}.align.err;
+ else hisat2 -p `nproc` --add-chrname --un-gz ${repRID}.unal.gz -S ${repRID}.sam -x ${reference}/genome -U ${fqs[0]} 1>${repRID}.align.out 2> ${repRID}.align.err;
+ fi;
+ samtools view -1 --threads `nproc` -o ${repRID}.bam ${repRID}.sam 1>${repRID}.align.out 2> ${repRID}.align.err;
+ samtools sort -@ `nproc` -O BAM ${repRID}.bam 1>${repRID}.align.out 2> ${repRID}.align.err;
+ """
+}
+