Ngscheckmate (nf-core#3094)

* Fix version string spurious text in singularity tests * Add NGSCheckMate subworkflow * Remember to save files * Fix bracket * Update meta * Update meta/channel name * Update NCM * Update ncm test * Add the config file * Remove tests on bcftools files as failing conda * Update subworkflows/nf-core/bam_ngscheckmate/main.nf Co-authored-by: Maxime U Garcia <[email protected]> * Update subworkflows/nf-core/bam_ngscheckmate/main.nf Co-authored-by: Maxime U Garcia <[email protected]> * Update subworkflows/nf-core/bam_ngscheckmate/main.nf --------- Co-authored-by: SPearce <[email protected]> Co-authored-by: Maxime U Garcia <[email protected]> Co-authored-by: Maxime U Garcia <[email protected]>
jvfe · Mar 30, 2023 · 3cfd245 · 3cfd245
1 parent aa90540
commit 3cfd245
Show file tree

Hide file tree

Showing 11 changed files with 217 additions and 24 deletions.
diff --git a/modules/nf-core/ngscheckmate/ncm/main.nf b/modules/nf-core/ngscheckmate/ncm/main.nf
@@ -16,7 +16,7 @@ process NGSCHECKMATE_NCM {
     path "*_corr_matrix.txt", emit: corr_matrix
     path "*_matched.txt"    , emit: matched
     path "*_all.txt"        , emit: all
-    path "*.vcf"            , emit: vcfs, optional: true
+    path "*.vcf"            , emit: vcf, optional: true
     path "versions.yml"     , emit: versions
 
     when:

diff --git a/modules/nf-core/ngscheckmate/ncm/meta.yml b/modules/nf-core/ngscheckmate/ncm/meta.yml
@@ -55,9 +55,9 @@ output:
       description: A txt file containing all the sample comparisons, whether they match or not
       pattern: "*all.txt"
 
-  - vcfs:
+  - vcf:
       type: file
-      description: If ran in bam mode, vcf files for each sample giving the SNP calls
+      description: If ran in bam mode, vcf files for each sample giving the SNP calls used
       pattern: "*.vcf"
 
 authors:

diff --git a/subworkflows/nf-core/bam_ngscheckmate/main.nf b/subworkflows/nf-core/bam_ngscheckmate/main.nf
@@ -0,0 +1,39 @@
+include { BCFTOOLS_MPILEUP     } from '../../../modules/nf-core/bcftools/mpileup/main'
+include { NGSCHECKMATE_NCM } from '../../../modules/nf-core/ngscheckmate/ncm/main.nf'
+
+workflow BAM_NGSCHECKMATE {
+
+    take:
+    ch_bam              // channel: [ val(meta), bam ]
+    ch_bed              // channel: [ bed ]
+    ch_fasta            // channel: [ fasta ]
+
+    main:
+
+    ch_versions = Channel.empty()
+
+    ch_bam_bed  = ch_bam.combine(ch_bed)
+
+    BCFTOOLS_MPILEUP (ch_bam_bed, ch_fasta.collect(), false)
+    ch_versions = ch_versions.mix(BCFTOOLS_MPILEUP.out.versions)
+
+    BCFTOOLS_MPILEUP
+    .out
+    .vcf
+    .map{meta, vcf -> vcf}
+    .collect()
+    .set {ch_flat_vcfs}
+
+    NGSCHECKMATE_NCM (ch_flat_vcfs, ch_bed, ch_fasta)
+    ch_versions = ch_versions.mix(NGSCHECKMATE_NCM.out.versions)
+
+    emit:
+    pdf          = NGSCHECKMATE_NCM.out.pdf          // channel: [ pdf ]
+    corr_matrix  = NGSCHECKMATE_NCM.out.corr_matrix  // channel: [ corr_matrix ]
+    matched      = NGSCHECKMATE_NCM.out.matched      // channel: [ matched ]
+    all          = NGSCHECKMATE_NCM.out.all          // channel: [ all ]
+    vcf          = BCFTOOLS_MPILEUP.out.vcf          // channel: [ meta, vcf ]
+    versions     = ch_versions                       // channel: [ versions.yml ]
+
+}
+
diff --git a/subworkflows/nf-core/bam_ngscheckmate/meta.yml b/subworkflows/nf-core/bam_ngscheckmate/meta.yml
@@ -0,0 +1,55 @@
+name: "bam_ngscheckmate"
+description: Take a set of bam files and run NGSCheckMate to determine whether samples match with each other, using a set of SNPs.
+keywords:
+  - ngscheckmate
+  - qc
+modules:
+  - bcftools/mpileup
+  - ngscheckmate/ncm
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test' ]
+  - bam:
+      type: file
+      description: BAM files for each sample
+      pattern: "*.{bam}"
+  - snp_bed:
+      type: file
+      description: BED file containing the SNPs to analyse. NGSCheckMate provides some default ones for hg19/hg38.
+      pattern: "*.{bed}"
+
+  - fasta:
+      type: file
+      description: fasta file for the genome
+      pattern: "*.{fasta}"
+
+output:
+  - pdf:
+      type: file
+      description: A pdf containing a dendrogram showing how the samples match up
+      pattern: "*.{pdf}"
+  - corr_matrix:
+      type: file
+      description: A text file containing the correlation matrix between each sample
+      pattern: "*corr_matrix.txt"
+  - matched:
+      type: file
+      description: A txt file containing only the samples that match with each other
+      pattern: "*matched.txt"
+  - all:
+      type: file
+      description: A txt file containing all the sample comparisons, whether they match or not
+      pattern: "*all.txt"
+  - vcf:
+      type: file
+      description: vcf files for each sample giving the SNP calls
+      pattern: "*.vcf"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+authors:
+  - "@SPPearce"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
@@ -3216,6 +3216,10 @@ subworkflows/bam_markduplicates_picard:
   - subworkflows/nf-core/bam_markduplicates_picard/**
   - tests/subworkflows/nf-core/bam_markduplicates_picard/**
 
+subworkflows/bam_ngscheckmate:
+  - subworkflows/nf-core/bam_ngscheckmate/**
+  - tests/subworkflows/nf-core/bam_ngscheckmate/**
+
 subworkflows/bam_qc_picard:
   - subworkflows/nf-core/bam_qc_picard/**
   - tests/subworkflows/nf-core/bam_qc_picard/**

diff --git a/tests/modules/nf-core/ngscheckmate/ncm/main.nf b/tests/modules/nf-core/ngscheckmate/ncm/main.nf
@@ -8,7 +8,7 @@ include { NGSCHECKMATE_NCM as NGSCHECKMATE_NCM_VCF} from '../../../../../modules
 include { BEDTOOLS_MAKEWINDOWS } from '../../../../../modules/nf-core/bedtools/makewindows/main.nf'
 
 include { BCFTOOLS_MPILEUP } from '../../../../../modules/nf-core/bcftools/mpileup/main.nf'
-include { BCFTOOLS_MPILEUP as BCFTOOLS_MPILEUP2 } from '../../../../../modules/nf-core/bcftools/mpileup/main.nf'
+include { BCFTOOLS_MPILEUP as BCFTOOLS_MPILEUP_TWO } from '../../../../../modules/nf-core/bcftools/mpileup/main.nf'
 
 workflow test_ngscheckmate_ncm_bam {
     input    = [ file(params.test_data['sarscov2']['illumina']['test_paired_end_methylated_sorted_bam'], checkIfExists: true),
@@ -21,8 +21,8 @@ workflow test_ngscheckmate_ncm_bam {
     inputBed = [ [ id:'test'],
                  file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)]
 
-    BEDTOOLS_MAKEWINDOWS(inputBed, true).
-    tab.
+    BEDTOOLS_MAKEWINDOWS(inputBed).
+    bed.
     map{it[1]}.
     view().
     set{snp_channel}
@@ -31,28 +31,30 @@ workflow test_ngscheckmate_ncm_bam {
 }
 
 workflow test_ngscheckmate_ncm_vcf {
-    input1   = [ [ id:'test1' ], // meta map
-                 [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ]
-               ]
-
-    input2   = [ [ id:'test2' ], // meta map
-                 [ file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true) ]
-               ]
-
     fasta    = [ file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
 
     inputBed = [ [ id:'test'],
-                 file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)]
+                file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)]
+
+    input1   = [ [ id:'test1' ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)
+                ]
+
+    input2   = [ [ id:'test2' ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)
+                ]
 
     BCFTOOLS_MPILEUP ( input1, fasta, false )
-    BCFTOOLS_MPILEUP2 ( input2, fasta, false )
+    BCFTOOLS_MPILEUP_TWO ( input2, fasta, false )
 
-    BCFTOOLS_MPILEUP2.out.vcf.
+    BCFTOOLS_MPILEUP_TWO.out.vcf.
         combine( BCFTOOLS_MPILEUP.out.vcf ).
         map { [ it[1], it[3] ] }.
         set { vcf_channel }
 
-    BEDTOOLS_MAKEWINDOWS( inputBed, true ).tab.
+    BEDTOOLS_MAKEWINDOWS( inputBed).bed.
         map { it[1] }.
         view().
         set { snp_channel }

diff --git a/tests/modules/nf-core/ngscheckmate/ncm/nextflow.config b/tests/modules/nf-core/ngscheckmate/ncm/nextflow.config
@@ -4,14 +4,15 @@ process {
 
     withName: BEDTOOLS_MAKEWINDOWS {
         ext.args = '-w 1'
+        ext.prefix = 'test_split'
     }
 
     withName: BCFTOOLS_MPILEUP {
         ext.args2 = '--no-version --ploidy 1 --multiallelic-caller'
         ext.args3 = '--no-version'
     }
 
-    withName: BCFTOOLS_MPILEUP2 {
+    withName: BCFTOOLS_MPILEUP_TWO {
         ext.args2 = '--no-version --ploidy 1 --multiallelic-caller'
         ext.args3 = '--no-version'
     }
@@ -24,4 +25,4 @@ process {
         ext.args = '-B'
     }
 
-}
+}
diff --git a/tests/modules/nf-core/ngscheckmate/ncm/test.yml b/tests/modules/nf-core/ngscheckmate/ncm/test.yml
@@ -1,29 +1,44 @@
 - name: ngscheckmate ncm test_ngscheckmate_ncm_bam
-  command: nextflow run ./tests/modules/nf-core/ngscheckmate/ncm -entry test_ngscheckmate_ncm_bam -c ./tests/config/nextflow.config
+  command: nextflow run ./tests/modules/nf-core/ngscheckmate/ncm -entry test_ngscheckmate_ncm_bam -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/ngscheckmate/ncm/nextflow.config
   tags:
     - ngscheckmate/ncm
     - ngscheckmate
   files:
+    - path: output/bedtools/test_split.bed
+      md5sum: 5058157987313b598f1c260e6965d6f7
+    - path: output/bedtools/versions.yml
+    - path: output/ngscheckmate/output.pdf
     - path: output/ngscheckmate/output_all.txt
       md5sum: f71a712c3f6ecf64dd526365212f1b7c
     - path: output/ngscheckmate/output_corr_matrix.txt
       md5sum: 6777377aa9ae3d57f841b12896318db0
     - path: output/ngscheckmate/output_matched.txt
       md5sum: f71a712c3f6ecf64dd526365212f1b7c
+    - path: output/ngscheckmate/test.paired_end.methylated.sorted.vcf
+    - path: output/ngscheckmate/test.paired_end.sorted.vcf
     - path: output/ngscheckmate/versions.yml
-      md5sum: fbb2bebd65b4f4e1e93c6bf5c08a6829
 
 - name: ngscheckmate ncm test_ngscheckmate_ncm_vcf
-  command: nextflow run ./tests/modules/nf-core/ngscheckmate/ncm -entry test_ngscheckmate_ncm_vcf -c ./tests/config/nextflow.config
+  command: nextflow run ./tests/modules/nf-core/ngscheckmate/ncm -entry test_ngscheckmate_ncm_vcf -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/ngscheckmate/ncm/nextflow.config
   tags:
     - ngscheckmate/ncm
     - ngscheckmate
   files:
+    - path: output/bcftools/test1.bcftools_stats.txt
+    - path: output/bcftools/test1.vcf.gz
+    - path: output/bcftools/test1.vcf.gz.tbi
+    - path: output/bcftools/test2.bcftools_stats.txt
+    - path: output/bcftools/test2.vcf.gz
+    - path: output/bcftools/test2.vcf.gz.tbi
+    - path: output/bcftools/versions.yml
+    - path: output/bedtools/test_split.bed
+      md5sum: 5058157987313b598f1c260e6965d6f7
+    - path: output/bedtools/versions.yml
+    - path: output/ngscheckmate/output.pdf
     - path: output/ngscheckmate/output_all.txt
       md5sum: fd74956dcac279b6f58e82ea73e344f8
     - path: output/ngscheckmate/output_corr_matrix.txt
       md5sum: 0c86bdad2721c470fe6be119f291c8e5
     - path: output/ngscheckmate/output_matched.txt
       md5sum: fd74956dcac279b6f58e82ea73e344f8
     - path: output/ngscheckmate/versions.yml
-      md5sum: f06910b83dde194a47870c553cefe193
diff --git a/tests/subworkflows/nf-core/bam_ngscheckmate/main.nf b/tests/subworkflows/nf-core/bam_ngscheckmate/main.nf
@@ -0,0 +1,33 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { BAM_NGSCHECKMATE } from '../../../../subworkflows/nf-core/bam_ngscheckmate/main.nf'
+include { BEDTOOLS_MAKEWINDOWS } from '../../../../modules/nf-core/bedtools/makewindows/main.nf'
+
+workflow test_bam_ngscheckmate {
+
+    inputBed = [ [ id:'test_bed'],
+                file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)]
+
+    BEDTOOLS_MAKEWINDOWS(Channel.of(inputBed))
+    snp_bed  = BEDTOOLS_MAKEWINDOWS.out.bed
+
+    input   = [[
+            [ id:'test1' ], // meta map
+            file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true)
+        ],[
+            [ id:'test2' ], // meta map
+            file(params.test_data['sarscov2']['illumina']['test_paired_end_methylated_sorted_bam'], checkIfExists: true)
+        ]
+    ]
+
+    fasta    = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    ch_input = Channel.fromList(input)
+    ch_snp_bed = BEDTOOLS_MAKEWINDOWS.out.bed.map{it[1]}
+    ch_fasta = Channel.of(fasta)
+
+    BAM_NGSCHECKMATE( ch_input, ch_snp_bed , ch_fasta)
+
+}
diff --git a/tests/subworkflows/nf-core/bam_ngscheckmate/nextflow.config b/tests/subworkflows/nf-core/bam_ngscheckmate/nextflow.config
@@ -0,0 +1,19 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+   withName: BEDTOOLS_MAKEWINDOWS {
+        ext.args = '-w 1'
+        ext.prefix = 'test_split'
+    }
+
+    withName: ".*BAM_NGSCHECKMATE:BCFTOOLS_MPILEUP" {
+        ext.args2 = '--no-version --ploidy 1 -c'
+        ext.args3 = '--no-version'
+    }
+
+    withName: ".*BAM_NGSCHECKMATE:NGSCHECKMATE_NCM" {
+        ext.args = '-V'
+    }
+
+}
diff --git a/tests/subworkflows/nf-core/bam_ngscheckmate/test.yml b/tests/subworkflows/nf-core/bam_ngscheckmate/test.yml
@@ -0,0 +1,25 @@
+- name: bam_ngscheckmate test_bam_ngscheckmate
+  command: nextflow run ./tests/subworkflows/nf-core/bam_ngscheckmate -entry test_bam_ngscheckmate -c ./tests/config/nextflow.config
+  tags:
+    - bcftools
+    - bcftools/mpileup
+    - ngscheckmate
+    - ngscheckmate/ncm
+    - subworkflows
+    - subworkflows/bam_ngscheckmate
+  files:
+    - path: output/bcftools/test1.bcftools_stats.txt
+    - path: output/bcftools/test1.vcf.gz
+    - path: output/bcftools/test1.vcf.gz.tbi
+    - path: output/bcftools/test2.bcftools_stats.txt
+    - path: output/bcftools/test2.vcf.gz
+    - path: output/bcftools/test2.vcf.gz.tbi
+    - path: output/bedtools/test_split.bed
+      md5sum: 5058157987313b598f1c260e6965d6f7
+    - path: output/ngscheckmate/output.pdf
+    - path: output/ngscheckmate/output_all.txt
+      md5sum: efdecd402fb7f1425d96624a73ba33b2
+    - path: output/ngscheckmate/output_corr_matrix.txt
+      md5sum: 8b8acb28a3a2bc7c450a15eed397d8d8
+    - path: output/ngscheckmate/output_matched.txt
+      md5sum: efdecd402fb7f1425d96624a73ba33b2