Fix bcftools/stats module (#1953)

* test verified commit bcftools

* adjust tests

* add suggestions

* test for actual file content

modules/bcftools/stats/main.nf

@@ -8,8 +8,10 @@ process BCFTOOLS_STATS {
         'quay.io/biocontainers/bcftools:1.15.1--h0ea216a_0' }"
 
     input:
-    tuple val(meta), path(vcf)
-    path(target_bed)
+    tuple val(meta), path(vcf), path(tbi)
+    path regions
+    path targets
+    path samples
 
     output:
     tuple val(meta), path("*stats.txt"), emit: stats
@@ -21,11 +23,15 @@ process BCFTOOLS_STATS {
     script:
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
-    def target = target_bed ? "--regions-file ${target_bed}" : ""
+    def regions_file = regions ? "--regions-file ${regions}" : ""
+    def targets_file = targets ? "--targets-file ${targets}" : ""
+    def samples_file =  samples ? "--samples-file ${samples}" : ""
     """
     bcftools stats \\
         $args \\
-        $target \\
+        $regions_file \\
+        $targets_file \\
+        $samples_file \\
         $vcf > ${prefix}.bcftools_stats.txt
 
     cat <<-END_VERSIONS > versions.yml

modules/bcftools/stats/meta.yml

@@ -19,14 +19,28 @@ input:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - target_bed:
-      type: file
-      description: target bed file
-      pattern: "*.{bed}"
   - vcf:
       type: file
       description: VCF input file
       pattern: "*.{vcf}"
+  - tbi:
+      type: file
+      description: |
+        The tab index for the VCF file to be inspected. Optional: only required when parameter regions is chosen.
+      pattern: "*.tbi"
+  - regions:
+      type: file
+      description: |
+        Optionally, restrict the operation to regions listed in this file. (VCF, BED or tab-delimited)
+  - targets:
+      type: file
+      description: |
+        Optionally, restrict the operation to regions listed in this file (doesn't rely upon tbi index files)
+  - samples:
+      type: file
+      description: |
+        Optional, file of sample names to be included or excluded.
+        e.g. 'file.tsv'
 output:
   - meta:
       type: map

tests/modules/bcftools/stats/main.nf

@@ -6,8 +6,33 @@ include { BCFTOOLS_STATS } from '../../../../modules/bcftools/stats/main.nf'
 
 workflow test_bcftools_stats {
     input = [ [ id:'test' ], // meta map
-              [ file(params.test_data['sarscov2']['illumina']['test_vcf'], checkIfExists: true) ],
-            ]
+            file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+            []]
+    regions = []
+    targets = []
+    samples = []
 
-    BCFTOOLS_STATS ( input, [] )
+    BCFTOOLS_STATS ( input, regions, targets, samples )
+}
+
+workflow test_bcftools_stats_regions {
+    input = [ [ id:'test' ], // meta map
+            file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_vcf_gz_tbi'], checkIfExists: true)]
+    regions = file(params.test_data['sarscov2']['illumina']['test3_vcf_gz'], checkIfExists: true)
+    targets = []
+    samples = []
+
+    BCFTOOLS_STATS ( input, regions, targets, samples )
+}
+
+workflow test_bcftools_stats_targets {
+    input = [ [ id:'test' ], // meta map
+            file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+            []]
+    regions = []
+    targets = file(params.test_data['sarscov2']['illumina']['test2_vcf_targets_tsv_gz'], checkIfExists: true)
+    samples = []
+
+    BCFTOOLS_STATS ( input, regions, targets, samples )
 }

tests/modules/bcftools/stats/test.yml

@@ -1,8 +1,26 @@
 - name: bcftools stats test_bcftools_stats
   command: nextflow run ./tests/modules/bcftools/stats -entry test_bcftools_stats -c ./tests/config/nextflow.config  -c ./tests/modules/bcftools/stats/nextflow.config
   tags:
-    - bcftools
     - bcftools/stats
+    - bcftools
   files:
     - path: output/bcftools/test.bcftools_stats.txt
-      md5sum: b7b9abecf0f6a77c1472a380feca708b
+      contains: ["TSTV	0	6	2	3.00	6	2	3.00", "DP	0	1	0	0.000000	5	55.555556"]
+
+- name: bcftools stats test_bcftools_stats_regions
+  command: nextflow run ./tests/modules/bcftools/stats -entry test_bcftools_stats_regions -c ./tests/config/nextflow.config  -c ./tests/modules/bcftools/stats/nextflow.config
+  tags:
+    - bcftools/stats
+    - bcftools
+  files:
+    - path: output/bcftools/test.bcftools_stats.txt
+      contains: ["TSTV	0	5	1	5.00	5	1	5.00", "DP	0	1	0	0.000000	3	42.857143"]
+
+- name: bcftools stats test_bcftools_stats_targets
+  command: nextflow run ./tests/modules/bcftools/stats -entry test_bcftools_stats_targets -c ./tests/config/nextflow.config  -c ./tests/modules/bcftools/stats/nextflow.config
+  tags:
+    - bcftools/stats
+    - bcftools
+  files:
+    - path: output/bcftools/test.bcftools_stats.txt
+      contains: ["TSTV	0	6	2	3.00	6	2	3.00", "DP	0	1	0	0.000000	5	55.555556"]