Added in test data for comparisions.

.gitlab-ci.yml

@@ -2,7 +2,7 @@ before_script:
   - module add  python/3.6.1-2-anaconda
   - pip install --user pytest-pythonpath pytest-cov
   - module load  nextflow/0.31.0
-  - ln -s /project/shared/bicf_workflow_ref/workflow_testdata/chipseq/*fastq.gz test_data/
+  - ln -s /work/BICF/s163035/chipseq/*fastq.gz test_data/
 
 stages:
   - unit
@@ -25,6 +25,20 @@ single_end_mouse:
 paired_end_human:
   stage: integration
   script:
-  - nextflow run workflow/main.nf --designFile "/work/BICF/s163035/builds/932983dd/0/BICF/Astrocyte/chipseq_analysis/test_data/design_ENCSR729LGA_PE.txt" --genome 'GRCh38' --pairedEnd true -resume
+  - nextflow run workflow/main.nf --designFile "$CI_PROJECT_DIR/test_data/design_ENCSR729LGA_PE.txt" --genome 'GRCh38' --pairedEnd true
+  artifacts:
+    expire_in: 2 days
+
+single_end_diff:
+  stage: integration
+  script:
+  - nextflow run workflow/main.nf --designFile "$CI_PROJECT_DIR/test_data/design_diff_SE.txt" --genome 'GRCm38'
+  artifacts:
+    expire_in: 2 days
+
+paired_end_diff:
+  stage: integration
+  script:
+  - nextflow run workflow/main.nf --designFile "$CI_PROJECT_DIR/test_data/design_diff_PE.txt" --genome 'GRCh38' --pairedEnd true
   artifacts:
     expire_in: 2 days

test_data/design_diff_PE.txt

+sample_id	experiment_id	biosample	factor	treatment	replicate	control_id	fastq_read1	fastq_read2
+ENCLB637LZP	ENCSR729LGA	MCF-7	SP1	None	1	ENCLB678IDC	ENCFF957SQS.fastq.gz	ENCFF582IOZ.fastq.gz
+ENCLB568IYX	ENCSR729LGA	MCF-7	SP1	None	2	ENCLB336TVW	ENCFF330MCZ.fastq.gz	ENCFF293YFE.fastq.gz
+ENCLB678IDC	ENCSR217LRF	MCF-7	Control	None	1	ENCLB678IDC	ENCFF002DTU.fastq.gz	ENCFF002EFI.fastq.gz
+ENCLB336TVW	ENCSR217LRF	MCF-7	Control	None	2	ENCLB336TVW	ENCFF002EFG.fastq.gz	ENCFF002DTS.fastq.gz
+ENCLB552ACZ	ENCSR757EMK	MCF-7	SUZ12	None	1	ENCLB678IDC	ENCFF833EZX.fastq.gz	ENCFF161HBP.fastq.gz
+ENCLB568IYX	ENCSR757EMK	MCF-7	SUZ12	None	2	ENCLB336TVW	ENCFF776KZU.fastq.gz	ENCFF119KHM.fastq.gz

test_data/design_diff_SE.txt

+sample_id	experiment_id	biosample	factor	treatment	replicate	control_id	fastq_read1
+ENCLB144FDT	ENCSR238SGC	limb	H3K4me1	None	1	ENCLB304SBJ	ENCFF833BLU.fastq.gz
+ENCLB831RUI	ENCSR238SGC	limb	H3K4me1	None	2	ENCLB410VVO	ENCFF646LXU.fastq.gz
+ENCLB304SBJ	ENCSR687ALB	limb	Control	None	1	ENCLB304SBJ	ENCFF524CAC.fastq.gz
+ENCLB410VVO	ENCSR687ALB	limb	Control	None	2	ENCLB410VVO	ENCFF163AJI.fastq.gz
+ENCLB140BPV	ENCSR272GNQ	midbrain	H3K4me1	None	1	ENCLB841FLB	ENCFF278VQW.fastq.gz
+ENCLB785CNN	ENCSR272GNQ	midbrain	H3K4me1	None	2	ENCLB735ZEL	ENCFF466CFM.fastq.gz
+ENCLB841FLB	ENCSR842LMA	midbrain	Control	None	1	ENCLB841FLB	ENCFF914QXH.fastq.gz
+ENCLB735ZEL	ENCSR842LMA	midbrain	Control	None	2	ENCLB735ZEL	ENCFF942UQV.fastq.gz

test_data/fetch_test_data.sh

@@ -3,6 +3,11 @@ wget https://www.encodeproject.org/files/ENCFF833BLU/@@download/ENCFF833BLU.fast
 wget https://www.encodeproject.org/files/ENCFF646LXU/@@download/ENCFF646LXU.fastq.gz
 wget https://www.encodeproject.org/files/ENCFF524CAC/@@download/ENCFF524CAC.fastq.gz
 wget https://www.encodeproject.org/files/ENCFF163AJI/@@download/ENCFF163AJI.fastq.gz
+echo "Downloading Single-end data set Mouse ENCSR272GNQ and ENCSR842LMA"
+wget https://www.encodeproject.org/files/ENCFF278VQW/@@download/ENCFF278VQW.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF466CFM/@@download/ENCFF466CFM.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF914QXH/@@download/ENCFF914QXH.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF942UQV/@@download/ENCFF942UQV.fastq.gz
 echo "Done with Single-end"
 
 echo "Downloading Paired-end data set Human ENCSR729LGA and ENCSR217LRF"
@@ -14,4 +19,9 @@ wget https://www.encodeproject.org/files/ENCFF002DTU/@@download/ENCFF002DTU.fast
 wget https://www.encodeproject.org/files/ENCFF002EFI/@@download/ENCFF002EFI.fastq.gz
 wget https://www.encodeproject.org/files/ENCFF002EFG/@@download/ENCFF002EFG.fastq.gz
 wget https://www.encodeproject.org/files/ENCFF002DTS/@@download/ENCFF002DTS.fastq.gz
+echo "Downloading Paired-end data set Human ENCSR757EMK"
+wget https://www.encodeproject.org/files/ENCFF833EZX/@@download/ENCFF833EZX.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF161HBP/@@download/ENCFF161HBP.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF776KZU/@@download/ENCFF776KZU.fastq.gz
+wget https://www.encodeproject.org/files/ENCFF119KHM/@@download/ENCFF119KHM.fastq.gz
 echo "Done with Paired-end"

workflow/main.nf

@@ -384,7 +384,7 @@ process consensusPeaks {
 }
 
 // Define channel to find number of unique experiments
-uniqueExperiments.splitCsv(sep: '\t', header: true).toList().map{ it.size().toInteger() }.set { noUniqueExperiments }
+noUniqueExperiments = uniqueExperiments.splitCsv(sep: '\t', header: true).toList()
 
 // Annotate Peaks
 process peakAnnotation {
@@ -426,7 +426,7 @@ process diffPeaks {
   file 'normcount_peaksets.txt' into normCountPeaks
 
   script:
-  if (noUniqueExperiments <= 1) {
+  if (noUniqueExperiments.size() == 1) {
     """
     touch design_diffpeak_annotatePeaks.tsv
     touch no_diffbind.bed