From e57a6c67c5bc4a6d927e937b5cf6595154d48742 Mon Sep 17 00:00:00 2001
From: Dave Lawrence <davmlaw@gmail.com>
Date: Wed, 7 Oct 2026 17:10:49 +1030
Subject: [PATCH] norm: left-align <DUP> with INFO/END but no INFO/SVLEN
 (#2601)

A <DUP> took its length only from INFO/SVLEN, so one described by
INFO/END alone was returned unchanged rather than realigned. The <DEL>
branch uses line->rlen, which htslib already derives from INFO/END or
INFO/SVLEN for both types, so do the same for <DUP>.

Assisted-by: Claude:claude-fable-5-1
Signed-off-by: Dave Lawrence <davmlaw@gmail.com>
---
 test/norm.symbolic.4.1.out | 10 ++++++++++
 test/norm.symbolic.4.vcf   |  8 ++++++++
 test/test.pl               |  1 +
 vcfnorm.c                  |  5 +----
 4 files changed, 20 insertions(+), 4 deletions(-)
 create mode 100644 test/norm.symbolic.4.1.out
 create mode 100644 test/norm.symbolic.4.vcf

diff --git a/test/norm.symbolic.4.1.out b/test/norm.symbolic.4.1.out
new file mode 100644
index 00000000..64cc40be
--- /dev/null
+++ b/test/norm.symbolic.4.1.out
@@ -0,0 +1,10 @@
+##fileformat=VCFv4.2
+##FILTER=<ID=PASS,Description="All filters passed">
+##INFO=<ID=END,Number=1,Type=Integer,Description="End position of the variant">
+##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="Difference in length between REF and ALT alleles">
+##INFO=<ID=SVTYPE,Number=1,Type=String,Description="Type of structural variant">
+##contig=<ID=1>
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+1	11	.	G	GGATTACA	.	PASS	SVTYPE=DUP
+1	11	.	G	<DUP>	.	PASS	SVTYPE=DUP;END=18
+1	11	.	G	<DUP>	.	PASS	SVTYPE=DUP;SVLEN=7;END=18
diff --git a/test/norm.symbolic.4.vcf b/test/norm.symbolic.4.vcf
new file mode 100644
index 00000000..1530a568
--- /dev/null
+++ b/test/norm.symbolic.4.vcf
@@ -0,0 +1,8 @@
+##fileformat=VCFv4.2
+##INFO=<ID=END,Number=1,Type=Integer,Description="End position of the variant">
+##INFO=<ID=SVLEN,Number=1,Type=Integer,Description="Difference in length between REF and ALT alleles">
+##INFO=<ID=SVTYPE,Number=1,Type=String,Description="Type of structural variant">
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO
+1	26	.	G	GATTACAG	.	PASS	SVTYPE=DUP
+1	26	.	G	<DUP>	.	PASS	SVTYPE=DUP;END=33
+1	26	.	G	<DUP>	.	PASS	SVTYPE=DUP;SVLEN=7;END=33
diff --git a/test/test.pl b/test/test.pl
index c0e600c4..060c73a2 100755
--- a/test/test.pl
+++ b/test/test.pl
@@ -312,6 +312,7 @@
 run_test(\&test_vcf_norm,$opts,in=>'norm.join-missing-ploidy',out=>'norm.join-missing-ploidy.1.out',args=>'-m +both');
 run_test(\&test_vcf_norm,$opts,in=>'norm.split.5',out=>'norm.split.5.1.out',args=>'-m - --multi-overlaps .');
 run_test(\&test_vcf_norm,$opts,in=>'norm.symbolic.3',out=>'norm.symbolic.3.1.out',fai=>'norm.symbolic.3',args=>'');
+run_test(\&test_vcf_norm,$opts,in=>'norm.symbolic.4',out=>'norm.symbolic.4.1.out',fai=>'norm.symbolic.3',args=>'');
 run_test(\&test_vcf_norm,$opts,in=>'norm',out=>'norm.out',fai=>'norm',args=>'-cx');
 run_test(\&test_vcf_norm,$opts,in=>'norm.split',out=>'norm.split.out',args=>'-m-');
 run_test(\&test_vcf_norm,$opts,in=>'norm.split',out=>'norm.split.1.out',args=>'-m- -a');
diff --git a/vcfnorm.c b/vcfnorm.c
index b32d521b..1b734694 100644
--- a/vcfnorm.c
+++ b/vcfnorm.c
@@ -697,10 +697,7 @@ static int realign(args_t *args, bcf1_t *line)
             // TODO: there should be check for symbolic allele length. If too big, perhaps should not attempt realignment
             int32_t sv_len = 0;
             if ( !strncmp("<DEL",line->d.allele[i],4) ) sv_len = -line->rlen;
-            else if ( !strncmp("<DUP",line->d.allele[i],4) )
-            {
-                if ( bcf_get_info_int32(args->hdr,line,"SVLEN",&args->int32_arr,&args->nint32_arr)==1 ) sv_len = args->int32_arr[0];
-            }
+            else if ( !strncmp("<DUP",line->d.allele[i],4) ) sv_len = line->rlen - 1;   // like <DEL>, htslib sets rlen from INFO/END or INFO/SVLEN
             if ( !sv_len ) return ERR_SYMBOLIC;
 
             als[i].l = 0;