Do not split multiallelic variants

config/nxf_vcf.config

@@ -7,9 +7,9 @@ env {
   CMD_VEP = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vep-111.0.sif vep"
   CMD_FILTERVEP = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vep-111.0.sif filter_vep"
   CMD_STRANGER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/stranger-0.8.1_v2.sif stranger"
-  CMD_VCFREPORT="apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-report-7.0.2.sif"
+  CMD_VCFREPORT="apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-report-7.0.3.sif"
   CMD_VCFDECISIONTREE = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-decision-tree-4.1.4.sif"
-  CMD_VCFINHERITANCEMATCHER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-inheritance-matcher-3.2.0.sif"
+  CMD_VCFINHERITANCEMATCHER = "apptainer exec --no-mount home --bind \${TMPDIR} ${APPTAINER_CACHEDIR}/vcf-inheritance-matcher-3.2.1.sif"
 
   // workaround for SAMtools https://github.com/samtools/samtools/issues/1366#issuecomment-769170935
   REF_PATH = ":"
@@ -37,7 +37,7 @@ process {
   }
 
   withLabel: 'vcf_report' {
-    memory = '4GB'
+    memory = '6GB'
   }
 }
 

docs/home/key_features.md

@@ -7,6 +7,7 @@ Features include:
 - Long-read sequencing support (Oxford Nanopore, PacBio HiFi)
 - Short-read sequencing support (Illumina, both single and paired-end reads)
 - Supports GRCh38, supports GRCh37 and T2T via liftover
+- Supports multiallelic variants
 - Short variant detection
 - Structural variant detection
 - Short tandem repeat detection

install.sh

@@ -88,8 +88,8 @@ download_files() {
   urls+=("8f6e06847776448e004df8b863571109" "images/straglr-1.4.4_vip_v3.sif")
   urls+=("9c69ac645e04b91c8f480289c536429c" "images/stranger-0.8.1_v2.sif")
   urls+=("3855ced7eb0c72f283d098d7938d8586" "images/vcf-decision-tree-4.1.4.sif")
-  urls+=("d2ab19ecccd9c596e2ec07442d1d551d" "images/vcf-inheritance-matcher-3.2.0.sif")
-  urls+=("14c1904c601cf051ce4c9dc4b0aa4607" "images/vcf-report-7.0.2.sif")
+  urls+=("f238b75e85e8a097447bad471369d0b2" "images/vcf-inheritance-matcher-3.2.1.sif")
+  urls+=("87b2d9031b1b8351d2da14dd0095fbea" "images/vcf-report-7.0.3.sif")
   urls+=("7bffc236a7c65b2b2e2e5f7d64beaa87" "images/vep-111.0.sif")
   urls+=("82be3c18406e7c027ee4cec83a723d71" "nextflow-24.04.2-all")
   if [ "${assembly}" == "ALL" ] || [ "${assembly}" == "GRCh37" ]; then

modules/vcf/templates/classify.sh

@@ -7,14 +7,14 @@ classify () {
   args+=("-XX:ParallelGCThreads=2")
   args+=("-Xmx!{task.memory.toMega() - 512}m")
   args+=("-jar" "/opt/vcf-decision-tree/lib/vcf-decision-tree.jar")
-  args+=("--input" "!{vcf}")
+  args+=("--input" "!{vcf}_replaced.vcf.gz")
   args+=("--metadata" "!{metadata}")
   args+=("--config" "!{decisionTree}")
   if [ !{annotatePath} -eq 1 ]; then
     args+=("--path")
   fi
 
-  args+=("--output" "!{vcfOut}")
+  args+=("--output" "!{vcfOut}_replaced.vcf.gz")
 
   ${CMD_VCFDECISIONTREE} java "${args[@]}"
 }
@@ -25,7 +25,7 @@ index () {
 }
 
 
-#Workaround for https://github.com/samtools/htsjdk/issues/500https://github.com/samtools/htsjdk/issues/500
+#Workaround for https://github.com/samtools/htsjdk/issues/500
 store_alt(){
   #store ALT headers
   zcat "!{vcf}" | sed --quiet --expression='/^##ALT/p' > header.tmp
@@ -38,15 +38,32 @@ insert_alt(){
     #re-insert the ALT headers
     f1=$(<header.tmp)
     awk -vf1="$f1" '/^#CHROM/{print f1;print;next}1' "!{vcfOut}".tmp | ${CMD_BGZIP} -c > "!{vcfOut}"
-    #rm header.tmp
   fi
 }
 
+#Workaround for https://github.com/samtools/htsjdk/issues/1718
+replace_cnv_tr(){
+  zcat "!{vcf}" | awk 'BEGIN{FS=OFS="\t"} {i=0; while(sub(/<CNV:TR>/,"<CNV:TR"++i">",$5));}1' | ${CMD_BGZIP} -c > "!{vcf}_replaced.vcf.gz"
+}
+
+restore_cnv_tr(){
+  zcat "!{vcfOut}_replaced.vcf.gz" | awk 'BEGIN{FS=OFS="\t"} {gsub(/<CNV:TR[0-9]+>/,"<CNV:TR>",$5);}1' | ${CMD_BGZIP} -c > "!{vcfOut}"
+}
+
+cleanup(){
+  rm "!{vcf}_replaced.vcf.gz"
+  rm "!{vcfOut}_replaced.vcf.gz"
+  rm header.tmp
+}
+
 main () {
   store_alt
+  replace_cnv_tr
   classify
+  restore_cnv_tr
   insert_alt
   index
+  cleanup
 }
 
 main "$@"

modules/vcf/templates/classify_samples.sh

@@ -7,7 +7,7 @@ classify_samples() {
   args+=("-XX:ParallelGCThreads=2")
   args+=("-Xmx!{task.memory.toMega() - 512}m")
   args+=("-jar" "/opt/vcf-decision-tree/lib/vcf-decision-tree.jar")
-  args+=("--input" "!{vcf}")
+  args+=("--input" "!{vcf}_replaced.vcf.gz")
   args+=("--metadata" "!{metadata}")
   args+=("--type" "sample")
   args+=("--config" "!{decisionTree}")
@@ -17,7 +17,7 @@ classify_samples() {
   if [ -n "!{probands}" ]; then
     args+=("--probands" "!{probands}")
   fi
-  args+=("--output" "!{vcfOut}")
+  args+=("--output" "!{vcfOut}_replaced.vcf.gz")
 
   ${CMD_VCFDECISIONTREE} java "${args[@]}"
 }
@@ -28,7 +28,7 @@ index () {
 }
 
 
-#Workaround for https://github.com/samtools/htsjdk/issues/500https://github.com/samtools/htsjdk/issues/500
+#Workaround for https://github.com/samtools/htsjdk/issues/500
 store_alt(){
   #store ALT headers
   zcat "!{vcf}" | sed --quiet --expression='/^##ALT/p' > header.tmp
@@ -41,15 +41,32 @@ insert_alt(){
     #re-insert the ALT headers
     f1=$(<header.tmp)
     awk -vf1="$f1" '/^#CHROM/{print f1;print;next}1' "!{vcfOut}".tmp | ${CMD_BGZIP} -c > "!{vcfOut}"
-    #rm header.tmp
   fi
 }
 
+#Workaround for https://github.com/samtools/htsjdk/issues/1718
+replace_cnv_tr(){
+  zcat "!{vcf}" | awk 'BEGIN{FS=OFS="\t"} {i=0; while(sub(/<CNV:TR>/,"<CNV:TR"++i">",$5));}1' | ${CMD_BGZIP} -c > "!{vcf}_replaced.vcf.gz"
+}
+
+restore_cnv_tr(){
+  zcat "!{vcfOut}_replaced.vcf.gz" | awk 'BEGIN{FS=OFS="\t"} {gsub(/<CNV:TR[0-9]+>/,"<CNV:TR>",$5);}1' | ${CMD_BGZIP} -c > "!{vcfOut}"
+}
+
+cleanup(){
+  rm "!{vcf}_replaced.vcf.gz"
+  rm "!{vcfOut}_replaced.vcf.gz"
+  rm header.tmp
+}
+
 main() {
+  replace_cnv_tr
   store_alt
   classify_samples
+  restore_cnv_tr
   insert_alt
   index
+  cleanup
 }
 
 main "$@"

modules/vcf/templates/inheritance.sh

@@ -11,8 +11,8 @@ inheritance () {
   args+=("-XX:ParallelGCThreads=2")
   args+=("-Xmx!{task.memory.toMega() - 512}m")
   args+=("-jar" "/opt/vcf-inheritance-matcher/lib/vcf-inheritance-matcher.jar")
-  args+=("--input" "!{vcf}")
-  args+=("--output" "!{vcfOut}")
+  args+=("--input" "!{vcf}_replaced.vcf.gz")
+  args+=("--output" "!{vcfOut}_replaced.vcf.gz")
   if [ -n "!{pedigree}" ]; then
     args+=("--pedigree" "!{pedigree}")
   fi
@@ -28,7 +28,7 @@ index () {
   ${CMD_BCFTOOLS} index --stats "!{vcfOut}" > "!{vcfOutStats}"
 }
 
-#Workaround for https://github.com/samtools/htsjdk/issues/500https://github.com/samtools/htsjdk/issues/500
+#Workaround for https://github.com/samtools/htsjdk/issues/500
 store_alt(){
   #store ALT headers
   zcat "!{vcf}" | sed --quiet --expression='/^##ALT/p' > header.tmp
@@ -41,16 +41,33 @@ insert_alt(){
     #re-insert the ALT headers
     f1=$(<header.tmp)
     awk -vf1="$f1" '/^#CHROM/{print f1;print;next}1' "!{vcfOut}".tmp | ${CMD_BGZIP} -c > "!{vcfOut}"
-    #rm header.tmp
   fi
 }
 
+#Workaround for https://github.com/samtools/htsjdk/issues/1718
+replace_cnv_tr(){
+  zcat "!{vcf}" | awk 'BEGIN{FS=OFS="\t"} {i=0; while(sub(/<CNV:TR>/,"<CNV:TR"++i">",$5));}1' | ${CMD_BGZIP} -c > "!{vcf}_replaced.vcf.gz"
+}
+
+restore_cnv_tr(){
+  zcat "!{vcfOut}_replaced.vcf.gz" | awk 'BEGIN{FS=OFS="\t"} {gsub(/<CNV:TR[0-9]+>/,"<CNV:TR>",$5);}1' | ${CMD_BGZIP} -c > "!{vcfOut}"
+}
+
+cleanup(){
+  rm "!{vcf}_replaced.vcf.gz"
+  rm "!{vcfOut}_replaced.vcf.gz"
+  rm header.tmp
+}
+
 main() {
+  replace_cnv_tr
   create_ped
   store_alt
   inheritance
+  restore_cnv_tr
   insert_alt
   index
+  cleanup
 }
 
 main "$@"

modules/vcf/templates/normalize.sh

@@ -4,14 +4,13 @@ set -euo pipefail
 normalize () {
   local args=()
   args+=("norm")
-  # split multi-allelic sites into bi-allelic records (both SNPs and indels are merged separately into two records)
-  args+=("--multiallelics" "-both")
   # warn when incorrect or missing REF allele is encountered or when alternate allele is non-ACGTN (e.g. structural variant)
   args+=("--check-ref" "w")
   args+=("--fasta-ref" "!{refSeqPath}")
   args+=("--output-type" "z")
   args+=("--output" "!{vcfOut}")
   args+=("--no-version")
+  args+=("--old-rec-tag" "OLD_REC") # if variant is normalized, keep the original location in this field
   args+=("--threads" "!{task.cpus}")
   args+=("!{vcf}")
 

test/suites/cram/resources/single.cfg

@@ -1,4 +1,4 @@
 params {
-    vcf.filter.classes = "LB,VUS,LP,P"
-    vcf.filter_samples.classes = "U1,U2"
+    vcf.filter.classes = "B,LB,VUS,LP,P"
+    vcf.filter_samples.classes = "U1,U2,U3"
 }

test/suites/vcf/resources/corner_cases.vcf

@@ -19,13 +19,13 @@
 ##FILTER=<ID=q40,Description="Quality below 40">
 ##FILTER=<ID=c5,Description="Coverage below 5">
 #CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	SAMPLE0
-chr1	1701	.	T	C	93	PASS	DP=131	GT	0/1
-chr1	1821	.	G	T	93	PASS	DP=133	GT	0/1
-chr1	1977	.	A	G	93	PASS	DP=134	GT	0/1
-chr1	4013	.	T	C	93	PASS	DP=122	GT	0/1
+chr1	1701	.	T	<CNV:TR>,<CNV:TR>	93	PASS	DP=131	GT	1/2
+chr1	1821	.	G	<CNV:TR>	93	PASS	DP=133	GT	0/1
+chr1	1977	.	A	G,C	93	PASS	DP=134	GT	0/1
+chr1	4013	.	T	C,G	93	PASS	DP=122	GT	0/1
 chr1	7362	.	G	C	93	PASS	DP=111	GT	0/1
 chr1	1455576	.	G	GGCCGAGGCCCGGGCGCGC	99	PASS	.	GT	1|1
-chr2	321682	.	T	<DEL>	6	PASS	SVTYPE=DEL;END=321887;SVLEN=-205	GT	0/1
+chr2	321682	.	T	A,<DEL>	6	PASS	SVTYPE=DEL;END=321887;SVLEN=-205	GT	0/1
 chr2	6693415	.	T	<DEL>	6	PASS	SVTYPE=DEL;END=6707844;SVLEN=-14429	GT	0/1
 chr2	9959221	.	C	<DEL>	6	PASS	SVTYPE=DEL;END=9959571;SVLEN=-350	GT	0/1
 chr2	28983075	.	C	<DUP:TANDEM>	.	.	END=105279483;SVTYPE=DUP;SVLEN=76296408	GT	1/1
@@ -36,4 +36,4 @@ chr13	51551934	.	C	<DEL>	6	PASS	SVTYPE=DEL;END=51552263;SVLEN=-329	GT	0/1
 chr17	14093499	.	T	<DEL>	PASS	.	SVTYPE=DEL;END=15486000;SVLEN=1392501	GT	0/1
 chrX	53513441	.	G	<DEL>	6	PASS	SVTYPE=DEL;END=53520776;SVLEN=-7335	GT	0/1
 chrX	53820721	.	T	<DUP>	6	PASS	SVTYPE=DUP;END=53862221;SVLEN=41500	GT	0/1
-chr11	119076999	.	C	<STR12>	.	PASS	END=119077032;REF=11;RL=33;RU=CGG;REPID=CBL;OLD_MULTIALLELIC=11:119076999:C/<STR12>/<STR20>	GT	1/0
+chr11	119076999	.	C	<STR12>,<STR14>	.	PASS	END=119077032;REF=11;RL=33;RU=CGG;REPID=CBL;OLD_MULTIALLELIC=11:119076999:C/<CNV:TR>/<CNV:TR>	GT	1/2