norm: avoid regression when ALT is a spanning deletion *

* add tests for #452 and #439 Resolves #452
samtools · Jul 1, 2016 · e43ca8c · e43ca8c
1 parent da5e8a0
commit e43ca8c
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 5 deletions.
diff --git a/test/norm.out b/test/norm.out
@@ -53,5 +53,6 @@
 20	275	.	A	C,G	999	PASS	INDEL;AN=2;AC=0,2	GT:PL:DP:FGF:FGI:FGS:FSTR	2:0,0,0:0:1e+06,2e+06,3e+06:1111,2222,3333:A,BB,CCC:WORD	2:0,0,0:0:1e+06,2e+06,3e+06:1111,2222,3333:A,BB,CCC:WORD
 3	10	.	GTGGAC	GTGGACACAC,GTGGACAC,GTGGACACACAC,GTGG,GTGGACACACACAC,ATGGACACACAC	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 3	17	.	CA	C	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
+4	25	.	T	TT,*	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 4	36	.	TC	C,TT,TTC	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 5	21	.	A	AAG	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
diff --git a/test/norm.vcf b/test/norm.vcf
@@ -54,4 +54,6 @@
 3	10	.	GTGGAC	GTGGACACAC,GTGGACAC,GTGGACACACAC,GTGG,GTGGACACACACAC,ATGGACACACAC	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 3	15	.	CACA	CAC	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 4	21	.	ATTTTTTTTTTTTTTTC	ATTTTTTTTTTTTTTC,ATTTTTTTTTTTTTTTT,ATTTTTTTTTTTTTTTTC	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
+4	25	.	T	TT,*	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
+4	37	.	C	I	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
 5	22	.	A	AGA	999	PASS	INDEL;AN=0	GT:DP	./.:0	./.:0
diff --git a/test/test.pl b/test/test.pl
@@ -95,7 +95,7 @@
 test_vcf_query($opts,in=>'missing',out=>'query.23.out',args=>q[-e'ISTR="."'  -f'%POS %ISTR\\n']);
 test_vcf_query($opts,in=>'missing',out=>'query.22.out',args=>q[-e'ISTR!="."' -f'%POS %ISTR\\n']);
 test_vcf_query($opts,in=>'missing',out=>'query.24.out',args=>q[-i'FILTER="q11"' -f'%POS %ISTR\\n']);
-test_vcf_norm($opts,in=>'norm',out=>'norm.out',fai=>'norm');
+test_vcf_norm($opts,in=>'norm',out=>'norm.out',fai=>'norm',args=>'-cx');
 test_vcf_norm($opts,in=>'norm.split',out=>'norm.split.out',args=>'-m-');
 test_vcf_norm($opts,in=>'norm.split.2',out=>'norm.split.2.out',args=>'-m-');
 test_vcf_norm($opts,in=>'norm.split',fai=>'norm',out=>'norm.split.and.norm.out',args=>'-m-');

diff --git a/vcfnorm.c b/vcfnorm.c
@@ -259,10 +259,11 @@ static void fix_dup_alt(args_t *args, bcf1_t *line)
     if ( changed ) bcf_update_genotypes(args->hdr,line,gts,ngts);
 }
 
-#define ERR_DUP_ALLELE      -2
-#define ERR_REF_MISMATCH    -1
-#define ERR_OK              0
-#define ERR_SYMBOLIC        1
+#define ERR_DUP_ALLELE       -2
+#define ERR_REF_MISMATCH     -1
+#define ERR_OK                0
+#define ERR_SYMBOLIC          1
+#define ERR_SPANNING_DELETION 2
 
 static int realign(args_t *args, bcf1_t *line)
 {
@@ -304,6 +305,7 @@ static int realign(args_t *args, bcf1_t *line)
     for (i=0; i<line->n_allele; i++)
     {
         if ( line->d.allele[i][0]=='<' ) return ERR_SYMBOLIC;  // symbolic allele
+        if ( line->d.allele[i][0]=='*' ) return ERR_SPANNING_DELETION;  // spanning deletion
         if ( has_non_acgtn(line->d.allele[i],0) )
         {
             if ( args->check_ref==CHECK_REF_EXIT )