diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 128756e1c26f5b1b7d4d9609675e67ca593c31f5..3e6393ed8d83c8b402f9b37d1713fe2244c49433 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -1,7 +1,7 @@
before_script:
- module add python/3.6.1-2-anaconda
- pip install --user pytest-pythonpath==0.7.1 pytest-cov==2.5.1
- - module load nextflow/0.31.0
+ - module load nextflow/0.31.0
- ln -s /project/shared/bicf_workflow_ref/workflow_testdata/chipseq/*fastq.gz test_data/
stages:
@@ -76,7 +76,7 @@ single_end_skip:
only:
- master
script:
- - nextflow run workflow/main.nf --designFile "$CI_PROJECT_DIR/test_data/design_diff_SE.txt" --genome 'GRCm38' --skipDiff true --skipMotif true --astrocyte false -resume
+ - nextflow run workflow/main.nf --designFile "$CI_PROJECT_DIR/test_data/design_diff_SE.txt" --genome 'GRCm38' --skipDiff true --skipMotif true --skipPlotProfile true --astrocyte false -resume
- pytest -m singleskip_true
artifacts:
expire_in: 2 days
diff --git a/astrocyte_pkg.yml b/astrocyte_pkg.yml
index ff68685e488ac30e45c0d2a863cdf131b71c41cc..704bc8fa415a0e624163971f225f034115f4c9f3 100644
--- a/astrocyte_pkg.yml
+++ b/astrocyte_pkg.yml
@@ -142,6 +142,15 @@ workflow_parameters:
description: |
Skip motif calling
+ - id: skipPlotProfile
+ type: select
+ required: true
+ choices:
+ - [ 'true', 'true']
+ - [ 'false', 'false']
+ description: |
+ Skip Plot Profile Analysis
+
- id: astrocyte
type: select
choices:
diff --git a/workflow/conf/biohpc.config b/workflow/conf/biohpc.config
index 6e2e885661a011956aaf1bac1e4fb0da3969d686..c514d076c9628538f1e9ed0af484a42e9a56c5e4 100644
--- a/workflow/conf/biohpc.config
+++ b/workflow/conf/biohpc.config
@@ -44,6 +44,10 @@ process {
module = ['python/3.6.1-2-anaconda', 'macs/2.1.0-20151222', 'UCSC_userApps/v317', 'bedtools/2.26.0', 'phantompeakqualtools/1.2']
queue = '128GB,256GB,256GBv1'
}
+ withName: plotProfile {
+ module = ['deeptools/2.5.0.1']
+ cpus = 32
+ }
withName: consensusPeaks {
module = ['python/3.6.1-2-anaconda', 'bedtools/2.26.0']
executor = 'local'
@@ -74,18 +78,21 @@ params {
genomesize = 'hs'
chromsizes = '/project/shared/bicf_workflow_ref/GRCh38/genomefile.txt'
fasta = '/project/shared/bicf_workflow_ref/GRCh38/genome.fa'
+ gta = '/project/shared/bicf_workflow_ref/GRCh38/gencode.gtf'
}
'GRCh37' {
bwa = '/project/shared/bicf_workflow_ref/GRCh37'
genomesize = 'hs'
chromsizes = '/project/shared/bicf_workflow_ref/GRCh37/genomefile.txt'
fasta = '/project/shared/bicf_workflow_ref/GRCh37/genome.fa'
+ gta = '/project/shared/bicf_workflow_ref/GRCh37/gencode.gtf'
}
'GRCm38' {
bwa = '/project/shared/bicf_workflow_ref/GRCm38'
genomesize = 'mm'
chromsizes = '/project/shared/bicf_workflow_ref/GRCm38/genomefile.txt'
fasta = '/project/shared/bicf_workflow_ref/GRCm38/genome.fa'
+ gta = '/project/shared/bicf_workflow_ref/GRCm38/gencode.gtf'
}
}
}
diff --git a/workflow/main.nf b/workflow/main.nf
index 9b45adb8235ef8964ac0d00fc86fd47b12d5cf0a..f392faa9114e2e2850b26230d2e1f38a3bfabecb 100644
--- a/workflow/main.nf
+++ b/workflow/main.nf
@@ -25,8 +25,10 @@ params.topPeakCount = 600
params.astrocyte = false
params.skipDiff = false
params.skipMotif = false
+params.skipPlotProfile = false
params.references = "$baseDir/../docs/references.md"
params.multiqc = "$baseDir/conf/multiqc_config.yaml"
+params.gtf = "/project/shared/bicf_workflow_ref/$params.genome/gencode.gtf"
// Assign variables if astrocyte
if (params.astrocyte) {
@@ -78,8 +80,10 @@ extendReadsLen = params.extendReadsLen
topPeakCount = params.topPeakCount
skipDiff = params.skipDiff
skipMotif = params.skipMotif
+skipPlotProfile = params.skipPlotProfile
references = params.references
multiqc = params.multiqc
+gtfFile = Channel.fromPath(params.gtf)
// Check design file for errors
process checkDesignFile {
@@ -424,6 +428,7 @@ process callPeaksMACS {
set sampleId, file('*.narrowPeak'), file('*.fc_signal.bw'), file('*.pvalue_signal.bw'), experimentId, biosample, factor, treatment, replicate, controlId into experimentPeaks
file '*.xls' into callPeaksMACSsummit
file('version_*.txt') into callPeaksMACSVersions
+ file("*.fc_signal.bw") into bigwigs
script:
@@ -456,6 +461,26 @@ peaksDesign = experimentPeaks
"$sampleId\t$peak\t$fcSignal\t$pvalueSignal\t$experimentId\t$biosample\t$factor\t$treatment\t$replicate\t$controlId\n"}
.collectFile(name:'design_peak.tsv', seed:"sample_id\tpeaks\tfc_signal\tpvalue_signal\texperiment_id\tbiosample\tfactor\ttreatment\treplicate\tcontrol_id\n", storeDir:"$outDir/design")
+//plotProfile
+process plotProfile {
+ publishDir "$outDir/${task.process}", mode: 'copy'
+
+ input:
+
+ file ("*.pooled.fc_signal.bw") from bigwigs.collect()
+ file gtf from gtfFile
+
+ when:
+
+ !skipPlotProfile
+
+ script:
+ """
+ module load deeptools/2.5.0.1
+ bash $baseDir/scripts/plotProfile.sh
+ """
+}
+
// Calculate Consensus Peaks
process consensusPeaks {
@@ -575,7 +600,6 @@ process diffPeaks {
// Generate Multiqc Report, gerernate Software Versions and references
process multiqcReport {
-
publishDir "$outDir/${task.process}", mode: 'copy'
input:
@@ -611,6 +635,7 @@ process multiqcReport {
module load multiqc/1.7
echo $workflow.nextflow.version > version_nextflow.txt
multiqc --version > version_multiqc.txt
+ python --version &> version_python.txt
python3 $baseDir/scripts/generate_references.py -r $references -o software_references
python3 $baseDir/scripts/generate_versions.py -o software_versions
multiqc -c $multiqc .
diff --git a/workflow/nextflow.config b/workflow/nextflow.config
index f9fbe846df7cbce559c0e8ec87097921792640a2..e1dd50f4e24db974e055d648003c901258f2c8c2 100644
--- a/workflow/nextflow.config
+++ b/workflow/nextflow.config
@@ -7,8 +7,8 @@ profiles {
manifest {
name = 'chipseq_analysis'
description = 'BICF ChIP-seq Analysis Workflow.'
- homePage = 'https://github.com/nf-core/rnaseq'
- version = '1.0.0'
+ homePage = 'https://git.biohpc.swmed.edu/BICF/Astrocyte/chipseq_analysis'
+ version = '1.0.6'
mainScript = 'main.nf'
nextflowVersion = '>=0.31.0'
}
diff --git a/workflow/scripts/generate_versions.py b/workflow/scripts/generate_versions.py
index 98c319331b7ff3574cb89bda39e0cc41397e7ded..4f0d8b143b36efbe207a303b2eb164d0de1a0e8a 100644
--- a/workflow/scripts/generate_versions.py
+++ b/workflow/scripts/generate_versions.py
@@ -46,6 +46,7 @@ SOFTWARE_REGEX = {
'MEME-ChIP': ['motifSearch_vf/version_memechip.txt', r"Version (\S+)"],
'DiffBind': ['diffPeaks_vf/version_DiffBind.txt', r"Version (\S+)\""],
'deepTools': ['experimentQC_vf/version_deeptools.txt', r"deeptools (\S+)"],
+ 'Python': ['version_python.txt', r"python, version (\S+)"],
'MultiQC': ['version_multiqc.txt', r"multiqc, version (\S+)"],
}
@@ -108,6 +109,7 @@ def main():
results['DiffBind'] = '<span style="color:#999999;\">Not Run</span>'
results['deepTools'] = '<span style="color:#999999;\">Not Run</span>'
results['MultiQC'] = '<span style="color:#999999;\">Not Run</span>'
+ results['Python'] = '<span style="color:#999999;\">Not Run</span>'
# list all files
files = glob.glob('**/*.txt', recursive=True)
diff --git a/workflow/scripts/plotProfile.sh b/workflow/scripts/plotProfile.sh
new file mode 100644
index 0000000000000000000000000000000000000000..3ab4d901ee4a3c9da7cc1aa55c8215f831805f02
--- /dev/null
+++ b/workflow/scripts/plotProfile.sh
@@ -0,0 +1,15 @@
+#!/bin/bash
+#plotProfile.sh
+
+bws=`ls *.bw`
+gtf=`ls *.gtf *.bed`
+
+computeMatrix reference-point \
+ --referencePoint TSS \
+ -S $bws \
+ -R $gtf \
+ --skipZeros \
+ -o computeMatrix.gz
+
+plotProfile -m computeMatrix.gz \
+ -out plotProfile.png \