update test data paths (#373)

* update test data paths

* Update test md5sums

* gatk test fixes & update variantfiltration main

* few extra fixes after review

* fix suspected format error

* Update software/gatk4/variantfiltration/main.nf

* Update software/gatk4/variantfiltration/main.nf

* Apply suggestions from code review

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

software/gatk4/variantfiltration/main.nf

@@ -20,11 +20,13 @@ process GATK4_VARIANTFILTRATION {
 
     input:
     tuple val(meta), path(vcf)
-    tuple path(fasta), path(fai), path(dict)
+    path fasta
+    path fai
+    path dict
 
     output:
-    tuple val(meta), path("*filtered.vcf"), emit: vcf
-    path "*.version.txt"                  , emit: version
+    tuple val(meta), path("*.vcf"), emit: vcf
+    path "*.version.txt"          , emit: version
 
 
     script:
@@ -34,7 +36,7 @@ process GATK4_VARIANTFILTRATION {
     gatk VariantFiltration \\
         -R $fasta \\
         -V $vcf \\
-        -O ${prefix}.filtered.vcf \\
+        -O ${prefix}.vcf \\
         $options.args
 
     gatk --version | grep Picard | sed "s/Picard Version: //g" > ${software}.version.txt

tests/software/bcftools/consensus/main.nf

@@ -6,9 +6,9 @@ include { BCFTOOLS_CONSENSUS } from '../../../../software/bcftools/consensus/mai
 
 workflow test_bcftools_consensus {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf.gz", checkIfExists: true) ],
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf.gz.tbi", checkIfExists: true) ],
-              [ file("${launchDir}/tests/data/genomics/sarscov2/genome/genome.fasta", checkIfExists: true) ]
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true) ],
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf_gz_tbi'], checkIfExists: true) ],
+              [ file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
             ]
 
     BCFTOOLS_CONSENSUS ( input )

tests/software/bcftools/consensus/test.yml

@@ -1,8 +1,8 @@
-- name: bcftools consensus
-  command: nextflow run ./tests/software/bcftools/consensus -entry test_bcftools_consensus -c tests/config/nextflow.config
+- name: bcftools consensus test_bcftools_consensus
+  command: nextflow run tests/software/bcftools/consensus -entry test_bcftools_consensus -c tests/config/nextflow.config
   tags:
-    - bcftools
     - bcftools_consensus
+    - bcftools
   files:
     - path: output/bcftools/test.fa
       md5sum: e57d3e4d113f989bc069c5fd61627091

tests/software/bcftools/filter/main.nf

@@ -7,7 +7,7 @@ include { BCFTOOLS_FILTER } from '../../../../software/bcftools/filter/main.nf'
 
 workflow test_bcftools_filter {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf", checkIfExists: true) ]
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf'], checkIfExists: true) ]
             ]
 
     BCFTOOLS_FILTER ( input )

tests/software/bcftools/filter/test.yml

@@ -1,8 +1,8 @@
-- name: bcftools filter
-  command: nextflow run ./tests/software/bcftools/filter -entry test_bcftools_filter -c tests/config/nextflow.config
+- name: bcftools filter test_bcftools_filter
+  command: nextflow run tests/software/bcftools/filter -entry test_bcftools_filter -c tests/config/nextflow.config
   tags:
-    - bcftools
     - bcftools_filter
+    - bcftools
   files:
     - path: output/bcftools/test.vcf.gz
       md5sum: 9d491cfa84067450342ba8e66c75e5b8

tests/software/bcftools/isec/main.nf

@@ -6,10 +6,10 @@ include { BCFTOOLS_ISEC } from '../../../../software/bcftools/isec/main.nf' addP
 
 workflow test_bcftools_isec {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf.gz", checkIfExists: true),
-                file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test2.vcf.gz", checkIfExists: true)],
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf.gz.tbi", checkIfExists: true),
-                file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test2.vcf.gz.tbi", checkIfExists: true)]
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test2_vcf_gz'], checkIfExists: true)],
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf_gz_tbi'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test2_vcf_gz_tbi'], checkIfExists: true)]
             ]
 
     BCFTOOLS_ISEC ( input )

tests/software/bcftools/isec/test.yml

@@ -1,18 +1,18 @@
-- name: bcftools isec
-  command: nextflow run ./tests/software/bcftools/isec -entry test_bcftools_isec -c tests/config/nextflow.config
+- name: bcftools isec test_bcftools_isec
+  command: nextflow run tests/software/bcftools/isec -entry test_bcftools_isec -c tests/config/nextflow.config
   tags:
     - bcftools
     - bcftools_isec
   files:
     - path: output/bcftools/test/0000.vcf.gz
       md5sum: fc178eb342a91dc0d1d568601ad8f8e2
+    - path: output/bcftools/test/0001.vcf.gz.tbi
+      md5sum: 36e11bf96ed0af4a92caa91a68612d64
+    - path: output/bcftools/test/sites.txt
+      md5sum: 1cea3fbde7f6d3c97f3d39036f9690df
     - path: output/bcftools/test/0000.vcf.gz.tbi
       md5sum: 36e11bf96ed0af4a92caa91a68612d64
     - path: output/bcftools/test/0001.vcf.gz
       md5sum: fc178eb342a91dc0d1d568601ad8f8e2
-    - path: output/bcftools/test/0001.vcf.gz.tbi
-      md5sum: 36e11bf96ed0af4a92caa91a68612d64
     - path: output/bcftools/test/README.txt
       md5sum: 10fc33b66522645600d44afbd41fb792
-    - path: output/bcftools/test/sites.txt
-      md5sum: 1cea3fbde7f6d3c97f3d39036f9690df

tests/software/bcftools/merge/main.nf

@@ -7,10 +7,10 @@ include { BCFTOOLS_MERGE } from '../../../../software/bcftools/merge/main.nf' ad
 
 workflow test_bcftools_merge {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test2.vcf.gz", checkIfExists: true),
-                file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test3.vcf.gz", checkIfExists: true) ],
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test2.vcf.gz.tbi", checkIfExists: true),
-                file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test3.vcf.gz.tbi", checkIfExists: true) ]
+              [ file(params.test_data['sarscov2']['illumina']['test2_vcf_gz'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test3_vcf_gz'], checkIfExists: true) ],
+              [ file(params.test_data['sarscov2']['illumina']['test2_vcf_gz_tbi'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test3_vcf_gz_tbi'], checkIfExists: true) ]
             ]
 
     BCFTOOLS_MERGE ( input )

tests/software/bcftools/merge/test.yml

@@ -1,8 +1,8 @@
-- name: bcftools merge
-  command: nextflow run ./tests/software/bcftools/merge -entry test_bcftools_merge -c tests/config/nextflow.config
+- name: bcftools merge test_bcftools_merge
+  command: nextflow run tests/software/bcftools/merge -entry test_bcftools_merge -c tests/config/nextflow.config
   tags:
-    - bcftools
     - bcftools_merge
+    - bcftools
   files:
     - path: output/bcftools/test.vcf.gz
       md5sum: 52165fc4149c61547e63800b876c9661

tests/software/bcftools/mpileup/main.nf

@@ -7,8 +7,8 @@ include { BCFTOOLS_MPILEUP } from '../../../../software/bcftools/mpileup/main.nf
 
 workflow test_bcftools_mpileup {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/bam/test_paired_end.sorted.bam", checkIfExists: true) ]]
-    fasta = [ file("${launchDir}/tests/data/genomics/sarscov2/genome/genome.fasta", checkIfExists: true) ]
+              [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ]]
+    fasta = [ file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
 
     BCFTOOLS_MPILEUP ( input, fasta )
 }

tests/software/bcftools/mpileup/test.yml

@@ -1,12 +1,12 @@
-- name: bcftools mpileup
-  command: nextflow run ./tests/software/bcftools/mpileup -entry test_bcftools_mpileup -c tests/config/nextflow.config
+- name: bcftools mpileup test_bcftools_mpileup
+  command: nextflow run tests/software/bcftools/mpileup -entry test_bcftools_mpileup -c tests/config/nextflow.config
   tags:
-    - bcftools
     - bcftools_mpileup
+    - bcftools
   files:
-    - path: output/bcftools/test.vcf.gz
-      md5sum: 518f770a9d04004c4493fc2167280653
-    - path: output/bcftools/test.vcf.gz.tbi
-      md5sum: f4be4d707434b3b960fde100b7ec3bf3
     - path: output/bcftools/test.bcftools_stats.txt
       md5sum: 2d506e32837a53a01fea0fc90402632a
+    - path: output/bcftools/test.vcf.gz.tbi
+      md5sum: 11d90b5b35e4adf6b44fc53bec93bed3
+    - path: output/bcftools/test.vcf.gz
+      md5sum: 2cf273a9fa3784383799b6b24df2f88c

tests/software/bcftools/stats/main.nf

@@ -6,7 +6,7 @@ include { BCFTOOLS_STATS } from '../../../../software/bcftools/stats/main.nf' ad
 
 workflow test_bcftools_stats {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf", checkIfExists: true) ]
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf'], checkIfExists: true) ]
             ]
 
     BCFTOOLS_STATS ( input )

tests/software/bcftools/stats/test.yml

@@ -1,5 +1,5 @@
-- name: bcftools stats
-  command: nextflow run ./tests/software/bcftools/stats -entry test_bcftools_stats -c tests/config/nextflow.config
+- name: bcftools stats test_bcftools_stats
+  command: nextflow run tests/software/bcftools/stats -entry test_bcftools_stats -c tests/config/nextflow.config
   tags:
     - bcftools
     - bcftools_stats

tests/software/gatk4/variantfiltration/main.nf

@@ -2,17 +2,17 @@
 
 nextflow.enable.dsl = 2
 
-test_options = ['args': '--filter-name "test_filter" --filter-expression "MQ0 > 0"']
+test_options = ['args': '--filter-name "test_filter" --filter-expression "MQ0 > 0"', 'suffix': '.filtered']
 include { GATK4_VARIANTFILTRATION } from '../../../../software/gatk4/variantfiltration/main.nf' addParams( options: test_options )
 
 workflow test_gatk4_variantfiltration {
     input = [ [ id:'test' ], // meta map
-              [ file("${launchDir}/tests/data/genomics/sarscov2/illumina/vcf/test.vcf", checkIfExists: true) ] 
-            ]
-    fasta = [ file("tests/data/genomics/sarscov2/genome/genome.fasta", checkIfExists: true),
-              file("tests/data/genomics/sarscov2/genome/genome.fasta.fai", checkIfExists: true),
-              file("tests/data/genomics/sarscov2/genome/genome.dict", checkIfExists: true)
+              [ file(params.test_data['sarscov2']['illumina']['test_vcf'], 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) ]
+    genome_dict = [ file(params.test_data['sarscov2']['genome']['genome_dict'], checkIfExists: true) ]
 
-    GATK4_VARIANTFILTRATION ( input, fasta )
+
+    GATK4_VARIANTFILTRATION ( input, fasta, fai, genome_dict )
 }

tests/software/gatk4/variantfiltration/test.yml

@@ -1,8 +1,9 @@
-- name: Run tests for gatk4/variantfiltration - test_gatk4_variantfiltration
-  command: nextflow run tests/software/gatk4/variantfiltration -entry test_gatk4_variantfiltration
-    -c tests/config/nextflow.config
+- name: gatk4 variantfiltration test_gatk4_variantfiltration
+  command: nextflow run tests/software/gatk4/variantfiltration -entry test_gatk4_variantfiltration -c tests/config/nextflow.config
   tags:
-    - gatk4
     - gatk4_variantfiltration
+    - gatk4
   files:
     - path: output/gatk4/test.filtered.vcf
+      contains:
+        - "AC=2;AN=2;DP=1;DP4=0,0,1,0;MQ=60;MQ0F=0;SGB=-0.379885"