Merge pull request #1689 from drpatelh/umitools

Update umitools/dedup module to make --output-stats optional

modules/umitools/dedup/main.nf

@@ -9,12 +9,13 @@ process UMITOOLS_DEDUP {
 
     input:
     tuple val(meta), path(bam), path(bai)
+    val get_output_stats
 
     output:
     tuple val(meta), path("*.bam")             , emit: bam
-    tuple val(meta), path("*edit_distance.tsv"), emit: tsv_edit_distance
-    tuple val(meta), path("*per_umi.tsv")      , emit: tsv_per_umi
-    tuple val(meta), path("*per_position.tsv") , emit: tsv_umi_per_position
+    tuple val(meta), path("*edit_distance.tsv"), optional:true, emit: tsv_edit_distance
+    tuple val(meta), path("*per_umi.tsv")      , optional:true, emit: tsv_per_umi
+    tuple val(meta), path("*per_position.tsv") , optional:true, emit: tsv_umi_per_position
     path  "versions.yml"                       , emit: versions
 
     when:
@@ -24,12 +25,13 @@ process UMITOOLS_DEDUP {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def paired = meta.single_end ? "" : "--paired"
+    def stats = get_output_stats ? "--output-stats $prefix" : ""
     """
     umi_tools \\
         dedup \\
         -I $bam \\
         -S ${prefix}.bam \\
-        --output-stats $prefix \\
+        $stats \\
         $paired \\
         $args
 

modules/umitools/dedup/meta.yml

@@ -26,6 +26,10 @@ input:
     description: |
       BAM index files corresponding to the input BAM file.
     pattern: "*.{bai}"
+  - get_output_stats:
+    type: boolean
+    description: |
+      Whether or not to generate output stats.
 output:
   - meta:
     type: map

tests/modules/umitools/dedup/main.nf

@@ -12,45 +12,72 @@ include { UMITOOLS_DEDUP } from '../../../../modules/umitools/dedup/main.nf'
 // Test with no UMI
 //
 workflow test_umitools_dedup_no_umi {
-    input = [ [ id:'test'], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ],
-              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+    input = [ 
+        [ id:'test'], // meta map
+        file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+        file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)
     ]
+    get_output_stats = false
 
-    UMITOOLS_DEDUP ( input )
+    UMITOOLS_DEDUP ( input, get_output_stats )
 }
 
 //
-// Test with single-end data
+// Test with single-end data without --output-stats
 //
-workflow test_umitools_dedup_single_end {
-    input = [ [ id:'test', single_end:true ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true) ]
+workflow test_umitools_dedup_single_end_no_stats {
+    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)
+    get_output_stats = false
 
     UMITOOLS_EXTRACT ( input )
     BWA_INDEX ( fasta )
     BWA_MEM ( UMITOOLS_EXTRACT.out.reads, BWA_INDEX.out.index, true )
-    SAMTOOLS_INDEX (BWA_MEM.out.bam)
-    UMITOOLS_DEDUP(BWA_MEM.out.bam.join(SAMTOOLS_INDEX.out.bai, by: [0]))
+    SAMTOOLS_INDEX ( BWA_MEM.out.bam )
+    UMITOOLS_DEDUP ( BWA_MEM.out.bam.join(SAMTOOLS_INDEX.out.bai, by: [0]), get_output_stats )
 }
 
 //
-// Test with paired-end data
+// Test with paired-end data without --output-stats
 //
-workflow test_umitools_dedup_paired_end {
-    input = [ [ id:'test', single_end:false ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
-                file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) ]
+workflow test_umitools_dedup_paired_end_no_stats {
+    input = [ 
+        [ id:'test', single_end:false ], // meta map
+        [ 
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) 
+        ]
     ]
-
     fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    get_output_stats = false
 
     UMITOOLS_EXTRACT ( input )
     BWA_INDEX ( fasta )
     BWA_MEM ( UMITOOLS_EXTRACT.out.reads, BWA_INDEX.out.index, true )
-    SAMTOOLS_INDEX (BWA_MEM.out.bam)
-    UMITOOLS_DEDUP(BWA_MEM.out.bam.join(SAMTOOLS_INDEX.out.bai, by: [0]))
+    SAMTOOLS_INDEX ( BWA_MEM.out.bam )
+    UMITOOLS_DEDUP ( BWA_MEM.out.bam.join(SAMTOOLS_INDEX.out.bai, by: [0]), get_output_stats )
+}
+
+//
+// Test with paired-end data with --output-stats
+//
+workflow test_umitools_dedup_paired_end_stats {
+    input = [ 
+        [ id:'test', single_end:false ], // meta map
+        [ 
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) 
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    get_output_stats = true
+
+    UMITOOLS_EXTRACT ( input )
+    BWA_INDEX ( fasta )
+    BWA_MEM ( UMITOOLS_EXTRACT.out.reads, BWA_INDEX.out.index, true )
+    SAMTOOLS_INDEX ( BWA_MEM.out.bam )
+    UMITOOLS_DEDUP ( BWA_MEM.out.bam.join(SAMTOOLS_INDEX.out.bai, by: [0]), get_output_stats )
 }

tests/modules/umitools/dedup/nextflow.config

@@ -7,11 +7,7 @@ process {
     }
 
     withName: UMITOOLS_DEDUP {
-        ext.args = ''
-        ext.prefix = 'dedup'
+        ext.prefix = { "${meta.id}.dedup" }
     }
 
-    withName: BWA_MEM {
-        ext.args2 = ''
-    }
 }

tests/modules/umitools/dedup/test.yml

@@ -1,54 +1,87 @@
 - name: umitools dedup test_umitools_dedup_no_umi
-  command: nextflow run tests/modules/umitools/dedup -entry test_umitools_dedup_no_umi -c tests/config/nextflow.config
+  command: nextflow run ./tests/modules/umitools/dedup -entry test_umitools_dedup_no_umi -c ./tests/config/nextflow.config  -c ./tests/modules/umitools/dedup/nextflow.config
   tags:
     - umitools/dedup
     - umitools
   files:
-    - path: output/umitools/dedup.bam
-      md5sum: 53b4edc399db81b87d2343e78af73cf0
-    - path: output/umitools/dedup_edit_distance.tsv
-      md5sum: 65186b0964e2f8d970cc04d736d8b119
-    - path: output/umitools/dedup_per_umi.tsv
-      md5sum: 8e6783a4a79437b095f095f2aefe7c01
-    - path: output/umitools/dedup_per_umi_per_position.tsv
-      md5sum: 9386db4a104b8e4e32f3ca4a84efa4ac
-    - path: output/umitools/versions.yml
-      md5sum: 4aaaa33565bcd9a984255139933d6446
+    - path: output/umitools/test.dedup.bam
 
-- name: umitools dedup test_umitools_dedup_single_end
-  command: nextflow run tests/modules/umitools/dedup -entry test_umitools_dedup_single_end -c tests/config/nextflow.config
+- name: umitools dedup test_umitools_dedup_single_end_no_stats
+  command: nextflow run ./tests/modules/umitools/dedup -entry test_umitools_dedup_single_end_no_stats -c ./tests/config/nextflow.config  -c ./tests/modules/umitools/dedup/nextflow.config
   tags:
-    - umitools
     - umitools/dedup
+    - umitools
   files:
+    - path: output/bwa/bwa/genome.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: output/bwa/bwa/genome.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+    - path: output/bwa/bwa/genome.bwt
+      md5sum: 0469c30a1e239dd08f68afe66fde99da
+    - path: output/bwa/bwa/genome.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: output/bwa/bwa/genome.sa
+      md5sum: ab3952cabf026b48cd3eb5bccbb636d1
     - path: output/bwa/test.bam
-      md5sum: ea41a3cdca1856b22845e1067fd31f37
-    - path: output/bwa/versions.yml
-      md5sum: ce4d987f2c53f4c01b31d210c357b24a
+      md5sum: 3ecbe569cadb9b6c881917ce60779f75
     - path: output/samtools/test.bam.bai
       md5sum: 095af0ad3921212597ffd7c342ecd5a0
-    - path: output/samtools/versions.yml
-      md5sum: 69b7cde627c9b4e8403dfc125db71cc7
-    - path: output/umitools/dedup.bam
-      md5sum: d95df177063432748ff33f473910cb1e
-    - path: output/umitools/versions.yml
-      md5sum: 730e768dd199d2f5bfb6fd0850446344
+    - path: output/umitools/test.dedup.bam
+    - path: output/umitools/test.umi_extract.fastq.gz
+    - path: output/umitools/test.umi_extract.log
 
-- name: umitools dedup test_umitools_dedup_paired_end
-  command: nextflow run tests/modules/umitools/dedup -entry test_umitools_dedup_paired_end -c tests/config/nextflow.config
+- name: umitools dedup test_umitools_dedup_paired_end_no_stats
+  command: nextflow run ./tests/modules/umitools/dedup -entry test_umitools_dedup_paired_end_no_stats -c ./tests/config/nextflow.config  -c ./tests/modules/umitools/dedup/nextflow.config
   tags:
-    - umitools
     - umitools/dedup
+    - umitools
   files:
+    - path: output/bwa/bwa/genome.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: output/bwa/bwa/genome.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+    - path: output/bwa/bwa/genome.bwt
+      md5sum: 0469c30a1e239dd08f68afe66fde99da
+    - path: output/bwa/bwa/genome.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: output/bwa/bwa/genome.sa
+      md5sum: ab3952cabf026b48cd3eb5bccbb636d1
     - path: output/bwa/test.bam
-      md5sum: 1ad786cae0ff2254c655e3a206929617
-    - path: output/bwa/versions.yml
-      md5sum: b524c5ddf61c20f4a0a93ae8fc78b851
+      md5sum: e7dcbac1825bf210409b762dbb4fec8f
     - path: output/samtools/test.bam.bai
-      md5sum: 7496f4056a8e86327ca93e350f282fc2
-    - path: output/samtools/versions.yml
-      md5sum: 72fc2ab934fd4bca0f7f14a705530d34
-    - path: output/umitools/dedup.bam
-      md5sum: e8d1eae2aacef76254948c5568e94555
-    - path: output/umitools/versions.yml
-      md5sum: fd39e05042d354b3d8de49b617d3183d
+      md5sum: f75780d1de7860329b7fb4afeadc4bed
+    - path: output/umitools/test.dedup.bam
+    - path: output/umitools/test.umi_extract.log
+    - path: output/umitools/test.umi_extract_1.fastq.gz
+    - path: output/umitools/test.umi_extract_2.fastq.gz
+
+- name: umitools dedup test_umitools_dedup_paired_end_stats
+  command: nextflow run ./tests/modules/umitools/dedup -entry test_umitools_dedup_paired_end_stats -c ./tests/config/nextflow.config  -c ./tests/modules/umitools/dedup/nextflow.config
+  tags:
+    - umitools/dedup
+    - umitools
+  files:
+    - path: output/bwa/bwa/genome.amb
+      md5sum: 3a68b8b2287e07dd3f5f95f4344ba76e
+    - path: output/bwa/bwa/genome.ann
+      md5sum: c32e11f6c859f166c7525a9c1d583567
+    - path: output/bwa/bwa/genome.bwt
+      md5sum: 0469c30a1e239dd08f68afe66fde99da
+    - path: output/bwa/bwa/genome.pac
+      md5sum: 983e3d2cd6f36e2546e6d25a0da78d66
+    - path: output/bwa/bwa/genome.sa
+      md5sum: ab3952cabf026b48cd3eb5bccbb636d1
+    - path: output/bwa/test.bam
+      md5sum: e7dcbac1825bf210409b762dbb4fec8f
+    - path: output/samtools/test.bam.bai
+      md5sum: f75780d1de7860329b7fb4afeadc4bed
+    - path: output/umitools/test.dedup.bam
+    - path: output/umitools/test.dedup_edit_distance.tsv
+      md5sum: c247a49b58768e6e2e86a6c08483e612
+    - path: output/umitools/test.dedup_per_umi.tsv
+      md5sum: 10e35ca37f2bfb521ac6dd7314951a68
+    - path: output/umitools/test.dedup_per_umi_per_position.tsv
+      md5sum: 2e1a12e6f720510880068deddeefe063
+    - path: output/umitools/test.umi_extract.log
+    - path: output/umitools/test.umi_extract_1.fastq.gz
+    - path: output/umitools/test.umi_extract_2.fastq.gz