From cbc47767f7bb4751a9d178395bd66f051401a0c1 Mon Sep 17 00:00:00 2001
From: Matthias De Smet <11850640+matthdsm@users.noreply.github.com>
Date: Fri, 13 May 2022 15:26:41 +0200
Subject: [PATCH] add sort option to bowtie2

---
 modules/bowtie2/align/main.nf        |  6 +++---
 modules/bowtie2/align/meta.yml       |  9 +++++++++
 tests/modules/bowtie2/align/main.nf  | 29 +++++++++++++++++++++++-----
 tests/modules/bowtie2/align/test.yml | 10 ++++++++++
 4 files changed, 46 insertions(+), 8 deletions(-)

diff --git a/modules/bowtie2/align/main.nf b/modules/bowtie2/align/main.nf
index e4bb4327..c74e376f 100644
--- a/modules/bowtie2/align/main.nf
+++ b/modules/bowtie2/align/main.nf
@@ -11,6 +11,7 @@ process BOWTIE2_ALIGN {
     tuple val(meta), path(reads)
     path  index
     val   save_unaligned
+    val   sort_bam
 
     output:
     tuple val(meta), path("*.bam")    , emit: bam
@@ -36,8 +37,7 @@ process BOWTIE2_ALIGN {
         reads_args = "-1 ${reads[0]} -2 ${reads[1]}"
     }
 
-    def samtools_command = "samtools view -@ $task.cpus --bam --with-header ${args2} > ${prefix}.bam"
-
+    def samtools_command = sort_bam ? 'sort' : 'view'
 
     """
     INDEX=`find -L ./ -name "*.rev.1.bt2" | sed "s/.rev.1.bt2//"`
@@ -51,7 +51,7 @@ process BOWTIE2_ALIGN {
         $unaligned \\
         $args \\
         2> ${prefix}.bowtie2.log \\
-        | $samtools_command
+        | samtools $samtools_command $args2 --threads $task.cpus -o ${prefix}.bam -
 
     if [ -f ${prefix}.unmapped.fastq.1.gz ]; then
         mv ${prefix}.unmapped.fastq.1.gz ${prefix}.unmapped_1.fastq.gz
diff --git a/modules/bowtie2/align/meta.yml b/modules/bowtie2/align/meta.yml
index f80421ec..c99fa4e3 100644
--- a/modules/bowtie2/align/meta.yml
+++ b/modules/bowtie2/align/meta.yml
@@ -29,6 +29,15 @@ input:
       type: file
       description: Bowtie2 genome index files
       pattern: "*.ebwt"
+  - save_unaligned:
+      type: boolean
+      description: |
+        Save reads that do not map to the reference (true) or discard them (false)
+        (default: false)
+  - sort_bam:
+      type: boolean
+      description: use samtools sort (true) or samtools view (false)
+      pattern: "true or false"
 output:
   - bam:
       type: file
diff --git a/tests/modules/bowtie2/align/main.nf b/tests/modules/bowtie2/align/main.nf
index 42e2306a..75ab33ac 100644
--- a/tests/modules/bowtie2/align/main.nf
+++ b/tests/modules/bowtie2/align/main.nf
@@ -14,9 +14,25 @@ workflow test_bowtie2_align_single_end {
     ]
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
     save_unaligned = false
+    sort = false
 
     BOWTIE2_BUILD ( fasta )
-    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned, sort )
+}
+
+workflow test_bowtie2_align_single_end_sorted {
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    save_unaligned = false
+    sort = true
+
+    BOWTIE2_BUILD ( fasta )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned, sort )
 }
 
 workflow test_bowtie2_align_paired_end {
@@ -29,9 +45,10 @@ workflow test_bowtie2_align_paired_end {
     ]
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
     save_unaligned = false
+    sort = false
 
     BOWTIE2_BUILD ( fasta )
-    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned, sort )
 }
 
 workflow test_bowtie2_align_single_end_large_index {
@@ -43,9 +60,10 @@ workflow test_bowtie2_align_single_end_large_index {
     ]
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
     save_unaligned = false
+    sort = false
 
     BOWTIE2_BUILD ( fasta )
-    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned, sort )
 }
 
 workflow test_bowtie2_align_paired_end_large_index {
@@ -58,7 +76,8 @@ workflow test_bowtie2_align_paired_end_large_index {
     ]
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
     save_unaligned = false
+    sort = false
 
     BOWTIE2_BUILD ( fasta )
-    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
-}
\ No newline at end of file
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned, sort )
+}
diff --git a/tests/modules/bowtie2/align/test.yml b/tests/modules/bowtie2/align/test.yml
index 103d65e9..4f8e4ff1 100644
--- a/tests/modules/bowtie2/align/test.yml
+++ b/tests/modules/bowtie2/align/test.yml
@@ -8,6 +8,16 @@
     - path: ./output/bowtie2/test.bowtie2.log
     - path: ./output/bowtie2/versions.yml
 
+- name: bowtie2 align test_bowtie2_align_single_end
+  command: nextflow run ./tests/modules/bowtie2/align -entry test_bowtie2_align_single_end_sorted -c ./tests/config/nextflow.config -c ./tests/modules/bowtie2/align/nextflow.config
+  tags:
+    - bowtie2
+    - bowtie2/align
+  files:
+    - path: ./output/bowtie2/test.bam
+    - path: ./output/bowtie2/test.bowtie2.log
+    - path: ./output/bowtie2/versions.yml
+
 - name: bowtie2 align test_bowtie2_align_paired_end
   command: nextflow run tests/modules/bowtie2/align -entry test_bowtie2_align_paired_end -c tests/config/nextflow.config -c tests/modules/bowtie2/align/nextflow.config
   tags: