From 854a2a43a3a67b467f6d5d81ed10c4bf8fc6d1bc Mon Sep 17 00:00:00 2001
From: Laurent Modolo <laurent.modolo@ens-lyon.fr>
Date: Wed, 22 Aug 2018 17:16:50 +0200
Subject: [PATCH] SRAtoolkit: update nf strcuture and improve script

---
 .../{sratoolkit.config => fastqdump.config}   |  0
 .../SRAtoolkit/{tests => }/fastqdump.nf       | 23 +++++-----
 .../SRAtoolkit/{tests => }/list-srr.txt       |  1 -
 src/nf_modules/SRAtoolkit/sratoolkit.nf       | 43 -------------------
 src/nf_modules/SRAtoolkit/tests.sh            |  4 ++
 src/nf_modules/SRAtoolkit/tests/tests.sh      |  4 --
 6 files changed, 17 insertions(+), 58 deletions(-)
 rename src/nf_modules/SRAtoolkit/{sratoolkit.config => fastqdump.config} (100%)
 rename src/nf_modules/SRAtoolkit/{tests => }/fastqdump.nf (61%)
 rename src/nf_modules/SRAtoolkit/{tests => }/list-srr.txt (93%)
 delete mode 100644 src/nf_modules/SRAtoolkit/sratoolkit.nf
 create mode 100755 src/nf_modules/SRAtoolkit/tests.sh
 delete mode 100755 src/nf_modules/SRAtoolkit/tests/tests.sh

diff --git a/src/nf_modules/SRAtoolkit/sratoolkit.config b/src/nf_modules/SRAtoolkit/fastqdump.config
similarity index 100%
rename from src/nf_modules/SRAtoolkit/sratoolkit.config
rename to src/nf_modules/SRAtoolkit/fastqdump.config
diff --git a/src/nf_modules/SRAtoolkit/tests/fastqdump.nf b/src/nf_modules/SRAtoolkit/fastqdump.nf
similarity index 61%
rename from src/nf_modules/SRAtoolkit/tests/fastqdump.nf
rename to src/nf_modules/SRAtoolkit/fastqdump.nf
index 32579e7..3dde596 100644
--- a/src/nf_modules/SRAtoolkit/tests/fastqdump.nf
+++ b/src/nf_modules/SRAtoolkit/fastqdump.nf
@@ -15,30 +15,33 @@ log.info "downloading list srr : ${params.list_srr}"
 Channel
   .fromPath( params.list_srr )
   .ifEmpty { error "Cannot find any bam files matching: ${params.list_srr}" }
-  .splitCsv(header: true)
+  .splitCsv()
+  .map { it -> it[0]}
   .set { SRR }
 
 //run is the column name containing SRR ids
 
 process fastq_dump {
-  tag {"${x.run}"}
-  publishDir "results/download/fastq/${x.run}/", mode: 'copy'
+  tag "$file_id"
+  publishDir "results/download/fastq/${file_id}/", mode: 'copy'
 
   input:
-    val x  from SRR
+    val file_id from SRR
 
   output:
-    file("*") into fastq
+    set file_id, "*.fastq" into fastq
 
   script:
 """
 #for test only 10000  reads are downloading with the option -N 10000 -X 20000
-fastq-dump --split-files --defline-seq '@\$ac_\$si/\$ri' --defline-qual "+" -N 10000 -X 20000 ${x.run}
-if [ -f ${x.run}_1.fastq ]
+fastq-dump --split-files --defline-seq '@\$ac_\$si/\$ri' --defline-qual "+" -N 10000 -X 20000 ${file_id}
+if [ -f ${file_id}_1.fastq ]
 then
-  true
-else
-  touch ${x.run}.fastq
+  mv ${file_id}_1.fastq ${file_id}_R1.fastq
+fi
+if [ -f ${file_id}_2.fastq ]
+then
+  mv ${file_id}_2.fastq ${file_id}_R2.fastq
 fi
 """
 }
diff --git a/src/nf_modules/SRAtoolkit/tests/list-srr.txt b/src/nf_modules/SRAtoolkit/list-srr.txt
similarity index 93%
rename from src/nf_modules/SRAtoolkit/tests/list-srr.txt
rename to src/nf_modules/SRAtoolkit/list-srr.txt
index a9cd1d0..a58fc10 100644
--- a/src/nf_modules/SRAtoolkit/tests/list-srr.txt
+++ b/src/nf_modules/SRAtoolkit/list-srr.txt
@@ -1,4 +1,3 @@
-run
 ERR572281
 ERR572146
 ERR572201
diff --git a/src/nf_modules/SRAtoolkit/sratoolkit.nf b/src/nf_modules/SRAtoolkit/sratoolkit.nf
deleted file mode 100644
index 9fce5e4..0000000
--- a/src/nf_modules/SRAtoolkit/sratoolkit.nf
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
-* sra-tools :
-
-*/
-
-/*                      fastq-dump
-* Imputs : srr list
-* Outputs : fastq files
-*/
-
-params.list_srr = "$baseDir/data/SRR/*.txt"
-
-log.info "downloading list srr : ${params.list_srr}"
-
-Channel
-  .fromPath( params.list_srr )
-  .ifEmpty { error "Cannot find any bam files matching: ${params.list_srr}" }
-  .splitCsv(header: true)
-  .set { SRR }
-
-//run is the column name containing SRR ids
-
-process fastq_dump {
-  tag {"${x.run}"}
-  publishDir "results/download/fastq/${x.run}/", mode: 'copy'
-
-  input:
-    val x  from SRR
-
-  output:
-    file("*") into fastq
-
-  script:
-"""
-fastq-dump --split-files --defline-seq '@\$ac_\$si/\$ri' --defline-qual "+"  ${x.run}
-if [ -f ${x.run}_1.fastq ]
-then
-  true
-else
-  touch ${x.run}.fastq
-fi
-"""
-}
diff --git a/src/nf_modules/SRAtoolkit/tests.sh b/src/nf_modules/SRAtoolkit/tests.sh
new file mode 100755
index 0000000..526cc7c
--- /dev/null
+++ b/src/nf_modules/SRAtoolkit/tests.sh
@@ -0,0 +1,4 @@
+nextflow src/nf_modules/SRAtoolkit/fastqdump.nf \
+  -c src/nf_modules/SRAtoolkit/fastqdump.config \
+  -profile docker \
+  --list_srr "src/nf_modules/SRAtoolkit/list-srr.txt"
diff --git a/src/nf_modules/SRAtoolkit/tests/tests.sh b/src/nf_modules/SRAtoolkit/tests/tests.sh
deleted file mode 100755
index c5efbcc..0000000
--- a/src/nf_modules/SRAtoolkit/tests/tests.sh
+++ /dev/null
@@ -1,4 +0,0 @@
-nextflow src/nf_modules/SRAtoolkit/tests/fastqdump.nf \
-  -c src/nf_modules/SRAtoolkit/sratoolkit.config \
-  -profile docker \
-  --list_srr "src/nf_modules/SRAtoolkit/tests/list-srr.txt"
-- 
GitLab