Update gatk3 modules (nf-core#3660)

* update mapdage2 module * Add meta for all files in gatk modules
limrp · Jul 28, 2023 · 664e3c4 · 664e3c4
1 parent 35a1d9d
commit 664e3c4
Show file tree

Hide file tree

Showing 9 changed files with 161 additions and 56 deletions.
diff --git a/modules/nf-core/gatk/indelrealigner/main.nf b/modules/nf-core/gatk/indelrealigner/main.nf
@@ -9,10 +9,10 @@ process GATK_INDELREALIGNER {
 
     input:
     tuple val(meta), path(bam), path(bai), path(intervals)
-    path(fasta)
-    path(fai)
-    path(dict)
-    path(known_vcf)
+    tuple val(meta2), path(fasta)
+    tuple val(meta3), path(fai)
+    tuple val(meta4), path(dict)
+    tuple val(meta5), path(known_vcf)
 
     output:
     tuple val(meta), path("*.bam"), path("*.bai"), emit: bam

diff --git a/modules/nf-core/gatk/indelrealigner/meta.yml b/modules/nf-core/gatk/indelrealigner/meta.yml
@@ -31,18 +31,38 @@ input:
       type: file
       description: Intervals file created by gatk3 RealignerTargetCreator
       pattern: "*.{intervals,list}"
+  - meta2:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fasta:
       type: file
       description: Reference file used to generate BAM file
       pattern: ".{fasta,fa,fna}"
+  - meta3:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fai:
       type: file
       description: Index of reference file used to generate BAM file
       pattern: ".fai"
+  - meta4:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - dict:
       type: file
       description: GATK dict file for reference
       pattern: ".dict"
+  - meta5:
+      type: map
+      description: |
+        Groovy Map containing file meta-information for known_vcf.
+        e.g. [ id:'test', single_end:false ]
   - known_vcf:
       type: file
       description: Optional input VCF file(s) with known indels

diff --git a/modules/nf-core/gatk/realignertargetcreator/main.nf b/modules/nf-core/gatk/realignertargetcreator/main.nf
@@ -8,11 +8,11 @@ process GATK_REALIGNERTARGETCREATOR {
         'biocontainers/gatk:3.5--hdfd78af_11' }"
 
     input:
-    tuple val(meta), path(input), path(index)
-    path fasta
-    path fai
-    path dict
-    path known_vcf
+    tuple val(meta), path(bam), path(bai)
+    tuple val(meta2), path(fasta)
+    tuple val(meta3), path(fai)
+    tuple val(meta4), path(dict)
+    tuple val(meta5), path(known_vcf)
 
     output:
     tuple val(meta), path("*.intervals"), emit: intervals
@@ -25,7 +25,7 @@ process GATK_REALIGNERTARGETCREATOR {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def known = known_vcf ? "-known ${known_vcf}" : ""
-    if ("$input" == "${prefix}.bam") error "Input and output names are the same, set prefix in module configuration to disambiguate!"
+    if ("$bam" == "${prefix}.bam") error "Input and output names are the same, set prefix in module configuration to disambiguate!"
 
     def avail_mem = 3072
     if (!task.memory) {
@@ -39,7 +39,7 @@ process GATK_REALIGNERTARGETCREATOR {
         -Xmx${avail_mem}M \\
         -T RealignerTargetCreator \\
         -nt ${task.cpus} \\
-        -I ${input} \\
+        -I ${bam} \\
         -R ${fasta} \\
         -o ${prefix}.intervals \\
         ${known} \\

diff --git a/modules/nf-core/gatk/realignertargetcreator/meta.yml b/modules/nf-core/gatk/realignertargetcreator/meta.yml
@@ -20,26 +20,46 @@ input:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - input:
+  - bam:
       type: file
       description: Sorted and indexed BAM/CRAM/SAM file
       pattern: "*.bam"
-  - index:
+  - bai:
       type: file
       description: BAM index file
       pattern: "*.bai"
+  - meta2:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fasta:
       type: file
       description: Reference file used to generate BAM file
       pattern: ".{fasta,fa,fna}"
+  - meta3:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fai:
       type: file
       description: Index of reference file used to generate BAM file
       pattern: ".fai"
+  - meta4:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - dict:
       type: file
       description: GATK dict file for reference
       pattern: ".dict"
+  - meta5:
+      type: map
+      description: |
+        Groovy Map containing file meta-information for known_vcf.
+        e.g. [ id:'test', single_end:false ]
   - known_vcf:
       type: file
       description: Optional input VCF file(s) with known indels

diff --git a/modules/nf-core/gatk/unifiedgenotyper/main.nf b/modules/nf-core/gatk/unifiedgenotyper/main.nf
@@ -8,14 +8,14 @@ process GATK_UNIFIEDGENOTYPER {
         'biocontainers/gatk:3.5--hdfd78af_11' }"
 
     input:
-    tuple val(meta), path(input), path(index)
-    path  fasta
-    path  fai
-    path  dict
-    path  intervals
-    path  contamination
-    path  dbsnp
-    path  comp
+    tuple val(meta), path(bam), path(bai)
+    tuple val(meta2), path(fasta)
+    tuple val(meta3), path(fai)
+    tuple val(meta4), path(dict)
+    tuple val(meta5), path(intervals)
+    tuple val(meta6), path(contamination)
+    tuple val(meta7), path(dbsnp)
+    tuple val(meta8), path(comp)
 
     output:
     tuple val(meta), path("*.vcf.gz"), emit: vcf
@@ -44,7 +44,7 @@ process GATK_UNIFIEDGENOTYPER {
         -Xmx${avail_mem}M \\
         -nt ${task.cpus} \\
         -T UnifiedGenotyper \\
-        -I ${input} \\
+        -I ${bam} \\
         -R ${fasta} \\
         ${contamination_file} \\
         ${dbsnp_file} \\

diff --git a/modules/nf-core/gatk/unifiedgenotyper/meta.yml b/modules/nf-core/gatk/unifiedgenotyper/meta.yml
@@ -17,38 +17,73 @@ input:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - input:
+  - bam:
       type: file
       description: Sorted and indexed BAM/CRAM/SAM file
       pattern: "*.bam"
-  - index:
+  - bai:
       type: file
       description: BAM index file
       pattern: "*.bai"
+  - meta2:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fasta:
       type: file
       description: Reference file used to generate BAM file
       pattern: ".{fasta,fa,fna}"
+  - meta3:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - fai:
       type: file
       description: Index of reference file used to generate BAM file
       pattern: ".fai"
+  - meta4:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - dict:
       type: file
       description: GATK dict file for reference
       pattern: ".dict"
+  - meta5:
+      type: map
+      description: |
+        Groovy Map containing reference information.
+        e.g. [ id:'test', single_end:false ]
   - intervals:
       type: file
       description: Bed file with the genomic regions included in the library (optional)
       pattern: "*.intervals"
+  - meta6:
+      type: map
+      description: |
+        Groovy Map containing file meta-information for the contamination file.
+        e.g. [ id:'test', single_end:false ]
   - contamination:
       type: file
       description: Tab-separated file containing fraction of contamination in sequencing data (per sample) to aggressively remove
       pattern: "*"
+  - meta7:
+      type: map
+      description: |
+        Groovy Map containing file meta-information for the dbsnps file.
+        e.g. [ id:'test', single_end:false ]
   - dbsnps:
       type: file
       description: VCF file containing known sites (optional)
       pattern: "*"
+  - meta8:
+      type: map
+      description: |
+        Groovy Map containing file meta-information for the VCF comparison file.
+        e.g. [ id:'test', single_end:false ]
   - comp:
       type: file
       description: Comparison VCF file (optional)

diff --git a/tests/modules/nf-core/gatk/indelrealigner/main.nf b/tests/modules/nf-core/gatk/indelrealigner/main.nf
@@ -8,26 +8,36 @@ include { GATK_INDELREALIGNER } from '../../../../../modules/nf-core/gatk/indelr
 
 workflow test_gatk_indelrealigner {
 
-
-    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
-    dict  = file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true)
-
-    input_realignertargetcreator     = [ [ 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),
-                ]
-
-    GATK_REALIGNERTARGETCREATOR ( input_realignertargetcreator, fasta, fai, dict, [] )
+    fasta = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+    fai = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    ]
+    dict = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true)
+    ]
+
+    input_realignertargetcreator = [
+        [ 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),
+    ]
+
+    GATK_REALIGNERTARGETCREATOR ( input_realignertargetcreator, fasta, fai, dict, [[],[]] )
 
     ch_intervals = GATK_REALIGNERTARGETCREATOR.out.intervals
 
-    ch_bams_indelrealigner = Channel.of([ [ 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)
-                ])
+    ch_bams_indelrealigner = Channel.of([
+        [ 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)
+    ])
 
     ch_input_indelrealigner = ch_bams_indelrealigner.mix(ch_intervals).groupTuple(by: 0).map{ [it[0], it[1][0], it[2], it[1][1] ] }
 
-    GATK_INDELREALIGNER ( ch_input_indelrealigner, fasta, fai, dict, [] )
+    GATK_INDELREALIGNER ( ch_input_indelrealigner, fasta, fai, dict, [[],[]] )
 }
diff --git a/tests/modules/nf-core/gatk/realignertargetcreator/main.nf b/tests/modules/nf-core/gatk/realignertargetcreator/main.nf
@@ -6,13 +6,23 @@ include { GATK_REALIGNERTARGETCREATOR } from '../../../../../modules/nf-core/gat
 
 workflow test_gatk_realignertargetcreator {
 
-    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),
-                ]
-    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
-    dict  = file(params.test_data['sarscov2']['genome']['genome_dict'], 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),
+    ]
+    fasta = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+    fai = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    ]
+    dict = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true)
+    ]
 
-    GATK_REALIGNERTARGETCREATOR ( input, fasta, fai, dict, [] )
+    GATK_REALIGNERTARGETCREATOR ( input, fasta, fai, dict, [[],[]] )
 }
diff --git a/tests/modules/nf-core/gatk/unifiedgenotyper/main.nf b/tests/modules/nf-core/gatk/unifiedgenotyper/main.nf
@@ -6,13 +6,23 @@ include { GATK_UNIFIEDGENOTYPER } from '../../../../../modules/nf-core/gatk/unif
 
 workflow test_gatk_unifiedgenotyper {
 
-    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),
-                ]
-    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    fai = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
-    dict = file(params.test_data['sarscov2']['genome']['genome_dict'], 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),
+    ]
+    fasta = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+    fai = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    ]
+    dict = [
+        [id: 'test'],
+        file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true)
+    ]
 
-    GATK_UNIFIEDGENOTYPER ( input, fasta, fai, dict, [], [], [], [])
+    GATK_UNIFIEDGENOTYPER ( input, fasta, fai, dict, [[],[]], [[],[]], [[],[]], [[],[]])
 }