V20210917

bioinfo-chru-strasbourg · Sep 17, 2021 · b704f54 · b704f54
1 parent b2cfdc0
commit b704f54
Showing 1 changed file with 9 additions and 5 deletions.
diff --git a/bin/HOWARD b/bin/HOWARD
@@ -713,23 +713,27 @@ if ((1)); then
 
  # CONTIG
  if ((1)); then
- existContig=$($BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "##contig=" -c)
+
+ #existContig=$($BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "##contig=" -c)
+ existContig=$($BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "##contig=<ID=[^,]*,length=[^,],assembly=[^,]>" -c)
  existVariants=$($BCFTOOLS view -H $INPUT_VCF_NUM 2>>$ERR | grep ^ -m1 -c)
 
  if (($existVariants)) && ! (($existContig)) ; then
  echo "#[INFO] Input VCF '$INPUT_VCF': No contig in header. Try to add contig in header";
  INPUT_TMP_CONTIG=$INPUT_VCF_NUM.contig.vcf
  INPUT_TMP_CONTIG_HEADER=$INPUT_VCF_NUM.contig.vcf.header
  # Create new header
- $BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "^##" >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
- $BCFTOOLS view -H $INPUT_VCF_NUM 2>>$ERR | cut -f1 | uniq | sort -k3,3V | awk '{print "##contig=<ID="$1">"}' >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
+ #$BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "^##" >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
+ $BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "^##" | grep -v "^##contig=" >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
+ $BCFTOOLS view -H $INPUT_VCF_NUM 2>>$ERR | cut -f1 | uniq | sort -k3,3V | awk '{print "##contig=<ID="$1",length=0,assembly=unknown>"}' >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
  $BCFTOOLS view -h $INPUT_VCF_NUM 2>>$ERR | grep "^#CHROM" >> $INPUT_TMP_CONTIG_HEADER 2>>$ERR
  # Reheader
  $BCFTOOLS reheader $INPUT_VCF_NUM -h $INPUT_TMP_CONTIG_HEADER > $INPUT_TMP_CONTIG 2>>$ERR
  # move file
  rm -f $INPUT_VCF_NUM
  mv $INPUT_TMP_CONTIG $INPUT_VCF_NUM
  rm -f $INPUT_TMP_CONTIG $INPUT_TMP_CONTIG_HEADER
+
  fi;
  fi;
 
@@ -1106,9 +1110,9 @@ if (($NB_VARIANT)); then
 
  # SPLIT VCF on ANNOTATIONS_SPLIT variants
  # HEAD
- $BCFTOOLS view -h $INPUT > $TMP_FOLDER/input.header
+ $BCFTOOLS view -h $INPUT > $TMP_FOLDER/input.header #2>/dev/null
  # VARIANTS
- $BCFTOOLS view -H $INPUT > $TMP_FOLDER/input.variants
+ $BCFTOOLS view -H $INPUT > $TMP_FOLDER/input.variants #2>/dev/null
  # SPLIT
  $BCFTOOLS view -H $INPUT | split - -a $SPLIT_SUFFIT_LENGTH -l $ANNOTATIONS_SPLIT $TMP_FOLDER/input.variants.splitted.
  VCF_SPLITTED_LIST=""