From d98d58f07136bfcfa8dd8854df8ac43a2e23da53 Mon Sep 17 00:00:00 2001
From: Petr Danecek <pd3@sanger.ac.uk>
Date: Tue, 3 Dec 2024 14:47:34 +0100
Subject: [PATCH] Fix BCF output

The command `convert --gvcf2vcf` was not filling the REF allele when BCF was output
because VCF API was not used, the alleles were rewritten directly. That works for
VCF, but not BCF output.

Fixes #243, finally
---
 NEWS         | 4 ++++
 test/test.pl | 1 +
 vcfconvert.c | 3 ++-
 3 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/NEWS b/NEWS
index e0a725cc..46793eec 100644
--- a/NEWS
+++ b/NEWS
@@ -12,6 +12,10 @@ Changes affecting specific commands:
     - New experimental plugin for scoring variants and assess site noisiness (allelic frequency profiles)
       from a large number of unaffected parental samples
 
+* bcftools convert
+
+    - The command `convert --gvcf2vcf` was not filling the REF allele when BCF was output (#243)
+
 * bcftools query
 
     - The functions used in -i/-e filtering expressions (such as SUM, MEDIAN, etc) can be
diff --git a/test/test.pl b/test/test.pl
index 96105ff8..6609d04a 100755
--- a/test/test.pl
+++ b/test/test.pl
@@ -1541,6 +1541,7 @@ sub test_vcf_convert_gvcf
     my ($opts,%args) = @_;
     bgzip_tabix_vcf($opts,$args{in});
     test_cmd($opts,%args,cmd=>"$$opts{bin}/bcftools convert --no-version $args{args} -f $$opts{path}/$args{fa} $$opts{tmp}/$args{in}.vcf.gz");
+    test_cmd($opts,%args,cmd=>"$$opts{bin}/bcftools convert --no-version $args{args} -f $$opts{path}/$args{fa} $$opts{tmp}/$args{in}.vcf.gz -Ou | $$opts{bin}/bcftools view --no-version");
     test_cmd($opts,%args,cmd=>"$$opts{bin}/bcftools view -Ob $$opts{tmp}/$args{in}.vcf.gz | $$opts{bin}/bcftools convert $args{args} -f $$opts{path}/$args{fa} | grep -v ^##bcftools");
 }
 sub test_vcf_convert_tsv2vcf
diff --git a/vcfconvert.c b/vcfconvert.c
index f75085aa..32764401 100644
--- a/vcfconvert.c
+++ b/vcfconvert.c
@@ -1,6 +1,6 @@
 /*  vcfconvert.c -- convert between VCF/BCF and related formats.
 
-    Copyright (C) 2013-2023 Genome Research Ltd.
+    Copyright (C) 2013-2024 Genome Research Ltd.
 
     Author: Petr Danecek <pd3@sanger.ac.uk>
 
@@ -1583,6 +1583,7 @@ static void gvcf_to_vcf(args_t *args)
             char *ref = faidx_fetch_seq(args->ref, (char*)bcf_hdr_id2name(hdr,line->rid), line->pos, line->pos, &len);
             if ( !ref ) error("faidx_fetch_seq failed at %s:%"PRId64"\n", bcf_hdr_id2name(hdr,line->rid),(int64_t) line->pos+1);
             strncpy(line->d.allele[0],ref,len);
+            bcf_update_alleles(hdr,line,(const char**)line->d.allele,line->n_allele);
             if ( bcf_write(out_fh,hdr,line)!=0 ) error("[%s] Error: cannot write to %s\n", __func__,args->outfname);
             free(ref);
         }