From 3d73714a01f76ea8f11914ea791e8a7e622f5e97 Mon Sep 17 00:00:00 2001
From: Venkat Malladi <venkat.malladi@utsouthwestern.edu>
Date: Sat, 30 Sep 2017 22:03:51 -0500
Subject: [PATCH] Add in fastqc step.
---
workflow/main.nf | 36 +++++++++++++++++++++++++++++++++---
workflow/scripts/qc_fastq.py | 20 ++++++++++----------
2 files changed, 43 insertions(+), 13 deletions(-)
diff --git a/workflow/main.nf b/workflow/main.nf
index c025d3b..ee90c3a 100644
--- a/workflow/main.nf
+++ b/workflow/main.nf
@@ -9,12 +9,11 @@ params.pairedEnd = false
params.designFile = "$baseDir/../test_data/design_ENCSR238SGC_SE.txt"
// Define List of Files
-Channel
+readsList = Channel
.fromPath( params.reads )
.flatten()
.map { file -> [ file.getFileName().toString(), file.toString() ].join("\t")}
.collectFile( name: 'fileList.tsv', newLine: true )
- .set { readsList }
// Define regular variables
pairedEnd = params.pairedEnd
@@ -37,7 +36,6 @@ process checkDesignFile {
if (pairedEnd) {
"""
- echo $designFile
python $baseDir/scripts/check_design.py -d $designFile -f $readsList -p
"""
}
@@ -48,3 +46,35 @@ process checkDesignFile {
}
}
+
+// Define channel for raw reads
+if (pairedEnd) {
+ rawReads = designFilePaths
+ .splitCsv(sep: '\t', header: true)
+ .map { row -> [ row.sample_id, [row.fastq_read1, row.fastq_read2], row.biosample, row.factor, row.treatment, row.replicate, row.control_id ] }
+} else {
+rawReads = designFilePaths
+ .splitCsv(sep: '\t', header: true)
+ .map { row -> [ row.sample_id, [row.fastq_read1, row.fastq_read1], row.biosample, row.factor, row.treatment, row.replicate, row.control_id ] }
+}
+
+process fastQc {
+
+ tag "$sampleId-$replicate"
+ publishDir "$baseDir/output/", mode: 'copy',
+ saveAs: {filename -> filename.indexOf(".zip") > 0 ? "zips/$filename" : "$filename"}
+
+ input:
+
+ set sampleId, reads, biosample, factor, treatment, replicate, controlId from rawReads
+
+ output:
+
+ file '*_fastqc.{zip,html}' into fastqc_results
+
+ script:
+
+ """
+ python $baseDir/scripts/qc_fastq.py -f $reads
+ """
+}
diff --git a/workflow/scripts/qc_fastq.py b/workflow/scripts/qc_fastq.py
index 991d927..95d8172 100755
--- a/workflow/scripts/qc_fastq.py
+++ b/workflow/scripts/qc_fastq.py
@@ -17,10 +17,10 @@ For more details:
## SETTINGS
-LOGGER = logging.getLogger(__name__)
-LOGGER.addHandler(logging.NullHandler())
-LOGGER.propagate = False
-LOGGER.setLevel(logging.INFO)
+logger = logging.getLogger(__name__)
+logger.addHandler(logging.NullHandler())
+logger.propagate = False
+logger.setLevel(logging.INFO)
def get_args():
@@ -41,24 +41,24 @@ def get_args():
def check_tools():
'''Checks for required componenets on user system'''
- LOGGER.info('Checking for required libraries and components on this system')
+ logger.info('Checking for required libraries and components on this system')
fastqc_path = shutil.which("fastqc")
if fastqc_path:
- LOGGER.info('Found fastqc: %s', fastqc_path)
+ logger.info('Found fastqc: %s', fastqc_path)
else:
- print("Please install 'fastqc' before using the tool")
- sys.exit()
+ logger.error('Missing fastqc')
+ raise Exception('Missing fastqc')
def check_qual_fastq(fastq):
'''Run fastqc on 1 or 2 files.'''
qc_command = "fastqc -t -f fastq " + " ".join(fastq)
- LOGGER.info("Running fastqc with %s", qc_command)
+ logger.info("Running fastqc with %s", qc_command)
qual_fastq = subprocess.Popen(qc_command, shell=True)
- qual_fastq .communicate()
+ out, err = qual_fastq.communicate()
def main():
--
GitLab