diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 30b1986eb9a107283e09326d2d0dca51cbad575e..c911d0b3550384fb297634663d0b5bfc7958799d 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -17,6 +17,13 @@ user_configuration:
script:
- pytest -m unit --cov=./workflow/scripts
+bash_tests:
+ stage: unit
+ script:
+ - module load singularity/3.0.2
+ - module load deeptools/2.5.0.1
+ - singularity run docker://bats/bats:v1.1.0 --tap workflow/tests/plot_profile.bats
+
astrocyte:
stage: astrocyte
script:
diff --git a/README.md b/README.md
index 5fa408e50bde9f859914207705790addbba62a28..2420017468b5a170ceb596852b717fa6e180661a 100644
--- a/README.md
+++ b/README.md
@@ -75,6 +75,7 @@ $ git clone git@git.biohpc.swmed.edu:BICF/Astrocyte/chipseq_analysis.git
9. Annotate all peaks using ChipSeeker
10. Calculate Differential Binding Activity with DiffBind (If more than 1 rep in more than 1 experiment)
11. Use MEME-ChIP to find motifs in original peaks
+ 12. Plot enrichment of signal around TSS
See [FLOWCHART](docs/flowchart.pdf)
diff --git a/docs/index.md b/docs/index.md
index 3b4a685cfc112291a221c04d7efa4fb85db9545a..e4e3e018513597745e2b2944e42d2ef05ac02277 100644
--- a/docs/index.md
+++ b/docs/index.md
@@ -20,6 +20,7 @@ Report issues to the Bioinformatic Core Facility [BICF](mailto:BICF@UTSouthweste
9. Annotate all peaks using ChipSeeker
10. Calculate Differential Binding Activity with DiffBind (If more than 1 rep in more than 1 experiment)
11. Use MEME-ChIP to find motifs in original peaks
+ 12. Plot enrichment of signal around TSS
## Workflow Parameters
diff --git a/workflow/main.nf b/workflow/main.nf
index 765ba3b1d1cd5b1d37d8dc339464d694c46ab8e2..ddebc750796a51060c1a5be040f61efd91b648d5 100644
--- a/workflow/main.nf
+++ b/workflow/main.nf
@@ -98,9 +98,8 @@ skipMotif = params.skipMotif
skipPlotProfile = params.skipPlotProfile
references = params.references
multiqc = params.multiqc
-gtfFile_plotProfile = Channel.fromPath(params.gtf)
-gtfFile_annotPeaks = Channel.fromPath(params.gtf)
-geneNames = Channel.fromPath(params.geneNames)
+gtfFile = params.gtf
+geneNames = params.geneNames
// Check design file for errors
process checkDesignFile {
@@ -484,8 +483,7 @@ process plotProfile {
input:
- file ("*.pooled.fc_signal.bw") from bigwigs.collect()
- file gtf from gtfFile_plotProfile
+ file bigWigList from bigwigs.collect()
output:
@@ -498,7 +496,7 @@ process plotProfile {
script:
"""
module load deeptools/2.5.0.1
- bash $baseDir/scripts/plotProfile.sh
+ bash $baseDir/scripts/plot_profile.sh -g $gtfFile
"""
}
@@ -540,8 +538,6 @@ process peakAnnotation {
input:
file designAnnotatePeaks
- file gtf from gtfFile_annotPeaks
- file geneNames
output:
@@ -552,7 +548,7 @@ process peakAnnotation {
"""
module load R/3.3.2-gccmkl
- Rscript $baseDir/scripts/annotate_peaks.R $designAnnotatePeaks $gtf $geneNames
+ Rscript $baseDir/scripts/annotate_peaks.R $designAnnotatePeaks $gtfFile $geneNames
"""
}
diff --git a/workflow/scripts/plotProfile.sh b/workflow/scripts/plotProfile.sh
deleted file mode 100644
index 0f50501c410d2ab0a7549190ba830aa0a90d7602..0000000000000000000000000000000000000000
--- a/workflow/scripts/plotProfile.sh
+++ /dev/null
@@ -1,16 +0,0 @@
-#!/bin/bash
-#plotProfile.sh
-
-bws=$(ls *.bw)
-gtf=$(ls *.gtf *.bed)
-
-computeMatrix reference-point \
- --referencePoint TSS \
- -S $bws \
- -R $gtf \
- --skipZeros \
- -o computeMatrix.gz
- -p max/2
-
-plotProfile -m computeMatrix.gz \
- -out plotProfile.png \
diff --git a/workflow/scripts/plot_profile.sh b/workflow/scripts/plot_profile.sh
new file mode 100755
index 0000000000000000000000000000000000000000..2481c8965856b507a19798f1c35302dc88724bec
--- /dev/null
+++ b/workflow/scripts/plot_profile.sh
@@ -0,0 +1,105 @@
+#!/bin/bash
+#plot_profile.sh
+
+script_name="plot_profile.sh"
+
+#Help function
+usage() {
+ echo "-h --Help documentation for $script_name"
+ echo "-g --File path to gtf/bed files"
+ echo "Example: $script_name -g 'genome.gtf'"
+ exit 1
+}
+
+
+raise()
+{
+ echo "${1}" >&2
+}
+
+check_tools() {
+ raise "
+ Checking for required libraries and components on this system
+ "
+ deeptools --version &> version_deeptools.txt
+ if [ $? -gt 0 ]
+ then
+ raise "Missing deeptools"
+ return 1
+ fi
+}
+
+compute_matrix() {
+ raise "
+ Computing matrix on ${1} using ${2}
+ "
+
+ computeMatrix reference-point \
+ --referencePoint TSS \
+ -S ${1} \
+ -R ${2} \
+ --skipZeros \
+ -o computeMatrix.gz \
+ -p max/2
+
+ if [ $? -gt 0 ]
+ then
+ raise "Problem building matrix"
+ return 1
+ fi
+}
+
+plot_profile() {
+ raise "
+ Plotting profile
+ "
+
+ plotProfile -m computeMatrix.gz \
+ -out plotProfile.png
+
+ if [ $? -gt 0 ]
+ then
+ raise "Problem plotting"
+ return 1
+ fi
+}
+
+
+run_main() {
+
+ # Parsing options
+ OPTIND=1 # Reset OPTIND
+ while getopts :g:h opt
+ do
+ case $opt in
+ g) gtf=$OPTARG;;
+ h) usage;;
+ esac
+ done
+
+ shift $(($OPTIND -1))
+
+ # Check for mandatory options
+ if [[ -z $gtf ]]; then
+ usage
+ fi
+
+ bws=$(ls *pooled.fc_signal.bw)
+
+ check_tools || exit 1
+
+ compute_matrix "${bws}" "${gtf}" || return 1
+
+ plot_profile || return 1
+
+ raise "ALL COMPLETE"
+}
+
+if [[ "${BASH_SOURCE[0]}" == "${0}" ]]
+then
+ run_main "$@"
+ if [ $? -gt 0 ]
+ then
+ exit 1
+ fi
+fi
diff --git a/workflow/tests/plot_profile.bats b/workflow/tests/plot_profile.bats
new file mode 100644
index 0000000000000000000000000000000000000000..d5da39f765f273488fcc6dce5675c49cd6e6ba18
--- /dev/null
+++ b/workflow/tests/plot_profile.bats
@@ -0,0 +1,26 @@
+#!/opt/bats/libexec/bats-core/ bats
+
+profile_script="./workflow/scripts/plot_profile.sh"
+
+@test "Test deeptools present" {
+ source ${profile_script}
+ run check_tools
+}
+
+@test "Test deeptools computeMatrix" {
+ source ${profile_script}
+ run compute_matrix test_data/ENCSR238SGC_pooled.fc_signal.bw /project/shared/bicf_workflow_ref/mouse/GRCm38/gencode.vM20.annotation.gtf
+ FILE=computeMatrix.gz
+ if [[ -s "$FILE" ]]; then
+ echo "$FILE exists and not empty"
+ fi
+}
+
+@test "Test deeptools plotProfile" {
+ source ${profile_script}
+ run plot_profile computeMatrix.gz
+ FILE=plotProfile.png
+ if [[ -s "$FILE" ]]; then
+ echo "$FILE exists and not empty"
+ fi
+}
diff --git a/workflow/tests/test_plot_profile.py b/workflow/tests/test_plot_profile.py
index 6c9605d6d654f66adec865372223e13ddeaf6b19..0fe03a6fc0d044227f719f903a7339d264b3fe07 100644
--- a/workflow/tests/test_plot_profile.py
+++ b/workflow/tests/test_plot_profile.py
@@ -11,8 +11,9 @@ test_output_path = os.path.dirname(os.path.abspath(__file__)) + \
@pytest.mark.singleend
def test_plot_singleend():
assert os.path.exists(os.path.join(test_output_path, 'plotProfile.png'))
-
+ assert os.path.getsize(os.path.join(test_output_path, 'plotProfile.png')) > 0
@pytest.mark.pairedend
def test_plot_pairedend():
- assert os.path.exists(os.path.join(test_output_path, 'computeMatrix.gz'))
+ assert os.path.exists(os.path.join(test_output_path, 'plotProfile.png'))
+ assert os.path.getsize(os.path.join(test_output_path, 'plotProfile.png')) > 0