Updated fastqc for better multithreading, updated module lists.

astrocyte_pkg.yml

@@ -9,7 +9,7 @@
 # A unique identifier for the workflow package, text/underscores only
 name: 'cellranger_mkfastq'
 # Who wrote this?
-author: 'Gervaise Henry and Venkat Malladi'
+author: 'Gervaise Henry,Venkat Malladi, and Jon Gesell'
 # A contact email address for questions
 email: 'bicf@utsouthwestern.edu'
 # A more informative title for the workflow package

workflow/main.nf

@@ -22,6 +22,7 @@ outDir = params.outDir
 process checkDesignFile {
   tag "$name"
   publishDir "$outDir/misc/${task.process}/$name", mode: 'copy'
+  module 'python/3.6.1-2-anaconda'
 
   input:
 
@@ -36,7 +37,6 @@ process checkDesignFile {
   """
   hostname
   ulimit -a
-  module load python/3.6.1-2-anaconda
   python3 $baseDir/scripts/check_design.py -d $designLocation
   """
 }
@@ -45,6 +45,7 @@ process checkDesignFile {
 process untarBCL {
   tag "$tar"
   publishDir "$outDir/${task.process}", mode: 'copy'
+  module 'pigz/2.4'
 
   input:
 
@@ -61,8 +62,8 @@ process untarBCL {
   ulimit -a
   name=`echo ${tar} | rev | cut -f1 -d '.' | rev`;
   if [ "\${name}" == "gz" ];
-  then   module load pigz/2.4;
-  tar -xvf $tar -I pigz;
+  then tar -xvf ${tar} -I pigz;
+  rm \${name};
   else tar -xvf ${tar};
   fi;
   """
@@ -71,7 +72,8 @@ process untarBCL {
 process mkfastq {
   tag "${bcl.baseName}"
   queue '128GB,256GB,256GBv1,384GB'
-  publishDir "$outDir/${task.process}", mode: 'copy'
+  publishDir "$outDir/${task.process}", mode: 'copy', pattern: "{*.fastq.gz,Stats.json}"
+  module 'cellranger/3.0.2:bcl2fastq/2.19.1'
 
   input:
 
@@ -80,7 +82,6 @@ process mkfastq {
 
   output:
 
-  file("**/outs/fastq_path/**/*") into mkfastqPaths
   file("**/outs/**/*.fastq.gz") into fastqPaths
   file("**/outs/fastq_path/Stats/Stats.json") into bqcPaths
   file("version*.txt") into versionPaths_mkfastq
@@ -90,8 +91,6 @@ process mkfastq {
   """
   hostname
   ulimit -a
-  module load cellranger/3.0.2
-  module load bcl2fastq/2.19.1
   sh $baseDir/scripts/versions_mkfastq.sh
   cellranger mkfastq --id="${bcl.baseName}" --run=$bcl --csv=$designPaths -r \$SLURM_CPUS_ON_NODE  -p \$SLURM_CPUS_ON_NODE  -w \$SLURM_CPUS_ON_NODE 
   """
@@ -102,13 +101,14 @@ process fastqc {
   tag "$name"
   queue 'super'
   publishDir "$outDir/misc/${task.process}/$name", mode: 'copy'
+  module 'fastqc/0.11.5'
 
   input:
   file fastqPaths
 
   output:
 
-  file("*fastqc.*") into fqcPaths
+  file("*fastqc.zip") into fqcPaths
   file("version*.txt") into versionPaths_fastqc
 
   script:
@@ -116,9 +116,7 @@ process fastqc {
   """
   hostname
   ulimit -a
-  module load fastqc/0.11.5
-  module load parallel
-  sh $baseDir/scripts/fastqc.sh
+  bash $baseDir/scripts/fastqc.sh
   """
 }
 
@@ -126,6 +124,7 @@ process fastqc {
 process versions {
   tag "$name"
   publishDir "$outDir/misc/${task.process}/$name", mode: 'copy'
+  module 'python/3.6.1-2-anaconda'
 
   input:
 
@@ -141,7 +140,6 @@ process versions {
   """
   hostname
   ulimit -a
-  module load python/3.6.1-2-anaconda
   echo $workflow.nextflow.version > version_nextflow.txt
   python3 $baseDir/scripts/generate_versions.py -f version_*.txt -o versions
   """
@@ -152,6 +150,7 @@ process multiqc {
   tag "$name"
   queue 'super'
   publishDir "$outDir/${task.process}/$name", mode: 'copy'
+  module 'multiqc/1.7'
 
   input:
 
@@ -168,7 +167,6 @@ process multiqc {
   """
   hostname
   ulimit -a
-  module load multiqc/1.7
-  multiqc . -c $baseDir/scripts/.multiqc_config.yaml 
+  multiqc *fastqc.zip -c $baseDir/scripts/.multiqc_config.yaml 
   """
 }

workflow/scripts/fastqc.sh

 #!/bin/bash
 
-find . -name '*.fastq.gz' | awk '{printf("fastqc \"%s\"\n", $0)}' | parallel -j 25 --verbose
-#find . -name '*fastqc.*' | xargs -I '{}' mv '{}' ./
-
-fastqc --version |& grep 'FastQC v' | sed -n -e 's/^FastQC v//p' > version_fastqc.txt  
+find . -name '*.fastq.gz' | awk '{printf("fastqc \"%s\"\n", $0)}' | parallel -j `grep -c ^processor /proc/cpuinfo` --verbose
+#find . -name '*fastqc.*' | xargs -I '{}' mv '{}' ./ 
+#for i in `ls *.fastq.gz`;
+#do echo "fastqc ${i}";
+#done | parallel -j `grep -c ^processor /proc/cpuinfo` --verbose;
+fastqc --version | grep 'FastQC v' | sed -n -e 's/^FastQC v//p' > version_fastqc.txt