From 0ea5423e9284fafe8d366fcd504c321e96869acb Mon Sep 17 00:00:00 2001
From: Dennis Hendriksen <d.hendriksen@umcg.nl>
Date: Thu, 14 Sep 2023 15:35:22 +0200
Subject: [PATCH] Fix: gvcf workflow results might be invalid for multi-sample
 projects (#465)

---
 config/nxf.config     |  2 +-
 modules/gvcf/utils.nf | 43 -------------------------------------------
 vip_gvcf.nf           |  2 +-
 3 files changed, 2 insertions(+), 45 deletions(-)
 delete mode 100644 modules/gvcf/utils.nf

diff --git a/config/nxf.config b/config/nxf.config
index 853366213..ba0302fb6 100644
--- a/config/nxf.config
+++ b/config/nxf.config
@@ -1,5 +1,5 @@
 env {
-  VIP_VERSION = "6.0.0"
+  VIP_VERSION = "6.0.1"
 
   TMPDIR = "\${TMPDIR:-\${NXF_TEMP:-\$(mktemp -d)}}"
   APPTAINER_BIND = "${APPTAINER_BIND}"
diff --git a/modules/gvcf/utils.nf b/modules/gvcf/utils.nf
deleted file mode 100644
index e2b4eca1c..000000000
--- a/modules/gvcf/utils.nf
+++ /dev/null
@@ -1,43 +0,0 @@
-include { parseFastaIndex } from '../utils'
-
-def determineChunks(meta) {
-  def fastaContigs = parseFastaIndex(params[meta.project.assembly].reference.fastaFai).collectEntries { record -> [record.contig, record] }
-  def records = meta.sample.gvcf.stats.readLines().collect { line -> line.split('\t') }
-
-  int chunkSize = 10000
-  int maxNrRecords = records.size() > 0 ? Math.max((records.max { record -> record[2] as int })[2] as int, chunkSize) : chunkSize
-
-  int regionNrRecords=0
-  def regions=[]
-  def chunks=[]
-  records.each { record ->
-    def vcfContig = record[0]
-    def fastaContig = fastaContigs[vcfContig]
-    if(!fastaContig) {
-        def fasta = params[meta.project.assembly].reference.fasta
-        throw new IllegalArgumentException("vcf chromosome '${vcfContig}' does not exist in reference genome '${fasta}' (assembly '${meta.project.assembly}'). are you using the correct reference genome?")
-    }
-    int contigNrRecords = record[2] as int
-    if(regionNrRecords + contigNrRecords <= maxNrRecords) {
-      regions.add([chrom: fastaContig.contig, chromStart: 0, chromEnd: fastaContig.size])
-      regionNrRecords += contigNrRecords
-    }
-    else {
-      chunks.add(regions)
-      regions=[]
-      regions.add([chrom: fastaContig.contig, chromStart: 0, chromEnd: fastaContig.size])
-      regionNrRecords = contigNrRecords
-    }
-  }
-  if(regions.size > 0) {
-     chunks.add(regions)
-  }
-
-  return chunks
-}
-
-def scatter(meta) {
-    def chunks = determineChunks(meta)
-    def index = 0
-    return !chunks.isEmpty() ? chunks.collect(chunk -> [*:meta, chunk: [index: index++, regions: chunk, total: chunks.size()] ]) : [[*:meta, chunk: [index: 0, regions: [], total: 0] ]]
-}
\ No newline at end of file
diff --git a/vip_gvcf.nf b/vip_gvcf.nf
index 05dc846ad..5f5be8cc3 100644
--- a/vip_gvcf.nf
+++ b/vip_gvcf.nf
@@ -3,7 +3,7 @@ nextflow.enable.dsl=2
 include { parseCommonSampleSheet; getAssemblies } from './modules/sample_sheet'
 include { getCramRegex; getGenomeVcfRegex } from './modules/utils'
 include { validate } from './modules/gvcf/validate'
-include { scatter } from './modules/gvcf/utils'
+include { scatter } from './modules/utils'
 include { merge } from './modules/gvcf/merge'
 include { vcf; validateVcfParams } from './vip_vcf'