nf-core · ramprasadn · Jul 26, 2023 · Jul 21, 2023 · Jul 25, 2023 · Jul 25, 2023
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -13,6 +13,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### `Fixed`
 
+## v1.1.1 - Abu (Patch) [2023-07-26]
+
+### `Fixed`
+
+- Avoids errors thrown by bcftools concat due to sample names in input vcf files not being in same order [#388](https://github.com/nf-core/raredisease/pull/388)
+
 ## v1.1.0 - Abu [2023-07-21]
 
 ### `Added`

diff --git a/conf/modules/merge_annotate_MT.config b/conf/modules/merge_annotate_MT.config
@@ -36,6 +36,7 @@ process {
     }
 
     withName: '.*ANALYSE_MT:MERGE_ANNOTATE_MT:BCFTOOLS_MERGE_MT' {
+        ext.args = '--output-type z'
         ext.prefix = { "${meta.id}_merge_mt" }
     }
 

diff --git a/conf/modules/raredisease.config b/conf/modules/raredisease.config
@@ -71,9 +71,8 @@ process {
 //
 
 process {
-    withName: '.*RAREDISEASE:BCFTOOLS_CONCAT' {
+    withName: '.*RAREDISEASE:GATK4_MERGEVCFS' {
         ext.prefix = { "${meta.id}_mito_genome_merged" }
-        ext.args = " -a "
     }
 }
 

diff --git a/lib/WorkflowRaredisease.groovy b/lib/WorkflowRaredisease.groovy
@@ -16,6 +16,25 @@ class WorkflowRaredisease {
 
     }
 
+    //
+    // Replace spaces in vcf INFO fields with underscore
+    //
+    public static String replaceSpacesInInfoColumn(vcf_file, parent_dir, base_name) {
+        def outfile = new File(parent_dir + '/' + base_name + '_formatted.vcf')
+        def writer  = outfile.newWriter()
+        vcf_file.eachLine { line ->
+            if (line.startsWith("#")) {
+                writer << line + "\n"
+            } else {
+                def split_str = line.tokenize("\t")
+                split_str[7] = split_str.getAt(7).replaceAll(" ","_")
+                writer << split_str.join("\t") + "\n"
+            }
+        }
+        writer.close()
+        return outfile
+    }
+
     //
     // Get workflow summary for MultiQC
     //

diff --git a/subworkflows/local/mitochondria/merge_annotate_MT.nf b/subworkflows/local/mitochondria/merge_annotate_MT.nf
@@ -143,7 +143,11 @@ workflow MERGE_ANNOTATE_MT {
 
         // HMTNOTE ANNOTATE
         HMTNOTE_ANNOTATE(VCFANNO_MT.out.vcf)
-        ZIP_TABIX_HMTNOTE(HMTNOTE_ANNOTATE.out.vcf)
+        HMTNOTE_ANNOTATE.out.vcf.map{meta, vcf ->
+            return [meta, WorkflowRaredisease.replaceSpacesInInfoColumn(vcf, vcf.parent.toString(), vcf.baseName)]
+            }
+            .set { ch_hmtnote_reformatted }
+        ZIP_TABIX_HMTNOTE(ch_hmtnote_reformatted)
 
         // Prepare output
         ch_vcf_out = ZIP_TABIX_HMTNOTE.out.gz_tbi.map{meta, vcf, tbi -> return [meta, vcf] }

diff --git a/workflows/raredisease.nf b/workflows/raredisease.nf
@@ -97,7 +97,7 @@ include { FILTER_VEP as FILTER_VEP_SV           } from '../modules/local/filter_
 // MODULE: Installed directly from nf-core/modules
 //
 
-include { BCFTOOLS_CONCAT                       } from '../modules/nf-core/bcftools/concat/main'
+include { GATK4_MERGEVCFS                       } from '../modules/nf-core/gatk4/mergevcfs/main'
 include { CUSTOM_DUMPSOFTWAREVERSIONS           } from '../modules/nf-core/custom/dumpsoftwareversions/main'
 include { FASTQC                                } from '../modules/nf-core/fastqc/main'
 include { GATK4_SELECTVARIANTS                  } from '../modules/nf-core/gatk4/selectvariants/main'
@@ -483,16 +483,9 @@ workflow RAREDISEASE {
                 .groupTuple()
                 .set { ch_merged_vcf }
 
-            ANNOTATE_SNVS.out.tbi
-                .concat(ANALYSE_MT.out.tbi)
-                .groupTuple()
-                .set { ch_merged_tbi }
-
-            ch_merged_vcf.join(ch_merged_tbi, failOnMismatch:true, failOnDuplicate:true).set {ch_concat_in}
-
-            BCFTOOLS_CONCAT (ch_concat_in)
-            ch_snv_annotate = BCFTOOLS_CONCAT.out.vcf
-            ch_versions = ch_versions.mix(BCFTOOLS_CONCAT.out.versions)
+            GATK4_MERGEVCFS (ch_merged_vcf, ch_genome_dictionary)
+            ch_snv_annotate = GATK4_MERGEVCFS.out.vcf
+            ch_versions = ch_versions.mix(GATK4_MERGEVCFS.out.versions)
         }
 
         ANN_CSQ_PLI_SNV (
-Original file line number
+Diff line change
@@ Expand Up / @@ -36,6 +36,7 @@ process { @@
         }
         withName: '.*ANALYSE_MT:MERGE_ANNOTATE_MT:BCFTOOLS_MERGE_MT' {
+            ext.args = '--output-type z'
             ext.prefix = { "${meta.id}_merge_mt" }
         }
@@ Expand Down @@