Enable PureCN with tumor-only Mutect too

As long as Mutect has been run with a PoN, PureCN can work with it.
nf-core · Oct 30, 2024 · 82d477e · 82d477e
1 parent 9e713bb
commit 82d477e
Show file tree

Hide file tree

Showing 2 changed files with 32 additions and 1 deletion.
diff --git a/subworkflows/local/bam_variant_calling_tumor_only_all/main.nf b/subworkflows/local/bam_variant_calling_tumor_only_all/main.nf
@@ -11,6 +11,8 @@ include { BAM_VARIANT_CALLING_TUMOR_ONLY_CONTROLFREEC } from '../bam_variant_cal
 include { BAM_VARIANT_CALLING_TUMOR_ONLY_MANTA        } from '../bam_variant_calling_tumor_only_manta/main'
 include { BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2      } from '../bam_variant_calling_tumor_only_mutect2/main'
 include { BAM_VARIANT_CALLING_TUMOR_ONLY_LOFREQ       } from '../bam_variant_calling_tumor_only_lofreq/main'
+include { PURECN_INTERVALFILE                         } from '../../../modules/nf-core/purecn/intervalfile/main'
+include { BAM_VARIANT_CALLING_SOMATIC_PURECN          } from '../bam_variant_calling_somatic_purecn/main.nf'
 
 workflow BAM_VARIANT_CALLING_TUMOR_ONLY_ALL {
     take:
@@ -36,6 +38,8 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_ALL {
     panel_of_normals_tbi          // channel: [optional]  panel_of_normals_tbi
     joint_mutect2                 // boolean: [mandatory] [default: false] run mutect2 in joint mode
     wes                           // boolean: [mandatory] [default: false] whether targeted data is processed
+    purecn_gatk_pon               // channel: [optional] PoN made from GATK4 for PureCN
+    purecn_normaldb               // channel: [optional] NormalDB generated by PureCN
 
     main:
     versions = Channel.empty()
@@ -173,6 +177,31 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_ALL {
         versions = versions.mix(BAM_VARIANT_CALLING_SINGLE_TIDDIT.out.versions)
     }
 
+    // Needs Mutect and a PoN specified
+    if (tools.split(',').contains('purecn') &&
+            tools.split('.').contains("mutect2") &&
+            panel_of_normals) {
+
+        ch_bed = [["id": "intervals_bed"], intervals]
+        ch_fasta = [["id": "fasta"], fasta]
+        PURECN_INTERVALFILE(ch_bed, ch_fasta)
+        versions = versions.mix(PURECN_INTERVALFILE.out.versions)
+
+        // As long as there's a PoN, it can work
+        BAM_VARIANT_CALLING_SOMATIC_PURECN(
+            cram,
+            fasta,
+            fasta_fai,
+            dict,
+            file(purecn_normaldb),
+            BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2.out.vcf, // PureCN will do filtering itself
+            file(purecn_gatk_pon),
+            PURECN_INTERVALFILE.out.txt,
+            PURECN_INTERVALFILE.out.bed
+            )
+        versions = versions.mix(BAM_VARIANT_CALLING_SOMATIC_PURECN.out.versions)
+    }
+
     vcf_all = Channel.empty().mix(
         vcf_freebayes,
         vcf_lofreq,

diff --git a/workflows/sarek/main.nf b/workflows/sarek/main.nf
@@ -796,7 +796,9 @@ workflow SAREK {
             pon,
             pon_tbi,
             params.joint_mutect2,
-            params.wes
+            params.wes,
+            params.purecn_gatk_pon,
+            params.purecn_normaldb
         )
 
         // PAIR VARIANT CALLING