update to have proper binary representation

samtools · Dec 6, 2024 · b833e31 · b833e31
1 parent 5177190
commit b833e31
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 10 deletions.
diff --git a/htslib/vcf.h b/htslib/vcf.h
@@ -1617,12 +1617,15 @@ static inline int bcf_format_gt1(const bcf_hdr_t *hdr, bcf_fmt_t *fmt, int isamp
              need to specify explicitly */
             e |= (ploidy > 1 && anyunphased) ?
                     (kputc('|', &tmp2) < 0) :
+                        (ploidy <= 1 && !((val0 >> 1)) ? //|. needs explicit o/p
+                            (kputc('|', &tmp2) < 0) :
+                            0);
                     0;
         } else {
             /* 1st allele is unphased, if ploidy is = 1 or allele is '.' or
              ploidy > 1 and no other unphased allele exist, need to specify
              explicitly */
-            e |= ((ploidy <= 1) || (ploidy > 1 && !anyunphased)) ?
+            e |= ((ploidy <= 1 && val0 != 0) || (ploidy > 1 && !anyunphased)) ?
                     (kputc('/', &tmp2) < 0) :
                     0;
         }

diff --git a/test/vcf44_1.expected b/test/vcf44_1.expected
@@ -1,6 +1,7 @@
 ##fileformat=VCFv4.4
 ##FILTER=<ID=PASS,Description="All filters passed">
 ##contig=<ID=1,length=1000>
+##reference=file://test
 ##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
 ##failue="test file on explicit and implicit phasing markers in 4.4"
 #CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	HG00096	HG00097
@@ -22,6 +23,13 @@
 1	61494	rs56992752	T	A	100	PASS	.	GT	/0|0	1/0
 1	61495	rs56992752	T	A	100	PASS	.	GT	0|0	|1/0
 1	61496	rs56992752	T	A	100	PASS	.	GT	.	.
-1	61497	rs56992752	T	A	100	PASS	.	GT	./1	.|1
-1	61498	rs56992752	T	A	100	PASS	.	GT	1/.	1|.
-1	61499	rs56992752	T	A	100	PASS	.	GT	./.	.|.
+1	61497	rs56992752	T	A	100	PASS	.	GT	.	|.
+1	61498	rs56992752	T	A	100	PASS	.	GT	./1	.|1
+1	61499	rs56992752	T	A	100	PASS	.	GT	./1	.|1
+1	61500	rs56992752	T	A	100	PASS	.	GT	|./1	/.|1
+1	61501	rs56992752	T	A	100	PASS	.	GT	1/.	1|.
+1	61502	rs56992752	T	A	100	PASS	.	GT	1/.	/1|.
+1	61503	rs56992752	T	A	100	PASS	.	GT	|1/.	1|.
+1	61504	rs56992752	T	A	100	PASS	.	GT	./.	.|.
+1	61505	rs56992752	T	A	100	PASS	.	GT	./.	.|.
+1	61506	rs56992752	T	A	100	PASS	.	GT	|./.	/.|.
diff --git a/test/vcf44_1.vcf b/test/vcf44_1.vcf
@@ -1,5 +1,6 @@
 ##fileformat=VCFv4.4
 ##contig=<ID=1,length=1000>
+##reference=file://test
 ##FORMAT=<ID=GT,Number=1,Type=String,Description="Genotype">
 ##failue="test file on explicit and implicit phasing markers in 4.4"
 #CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	HG00096	HG00097
@@ -21,6 +22,13 @@
 1	61494	rs56992752	T	A	100	PASS	.	GT	/0|0	1/0
 1	61495	rs56992752	T	A	100	PASS	.	GT	0|0	|1/0
 1	61496	rs56992752	T	A	100	PASS	.	GT	.	.
-1	61497	rs56992752	T	A	100	PASS	.	GT	./1	.|1
-1	61498	rs56992752	T	A	100	PASS	.	GT	1/.	1|.
-1	61499	rs56992752	T	A	100	PASS	.	GT	./.	.|.
+1	61497	rs56992752	T	A	100	PASS	.	GT	/.	|.
+1	61498	rs56992752	T	A	100	PASS	.	GT	./1	.|1
+1	61499	rs56992752	T	A	100	PASS	.	GT	/./1	|.|1
+1	61500	rs56992752	T	A	100	PASS	.	GT	|./1	/.|1
+1	61501	rs56992752	T	A	100	PASS	.	GT	1/.	1|.
+1	61502	rs56992752	T	A	100	PASS	.	GT	/1/.	/1|.
+1	61503	rs56992752	T	A	100	PASS	.	GT	|1/.	|1|.
+1	61504	rs56992752	T	A	100	PASS	.	GT	./.	.|.
+1	61505	rs56992752	T	A	100	PASS	.	GT	/./.	|.|.
+1	61506	rs56992752	T	A	100	PASS	.	GT	|./.	/.|.
diff --git a/vcf.c b/vcf.c
@@ -3111,7 +3111,7 @@ static int vcf_parse_format_fill5(kstring_t *s, const bcf_hdr_t *h, bcf1_t *v,
                     uint32_t unreadable = 0;
                     uint32_t max = 0;
                     int overflow = 0, ploidy = 0, anyunphased = 0, \
-                        phasingprfx = 0;
+                        phasingprfx = 0, unknown1 = 0;
 
                     /* with prefixed phasing, it is explicitly given for 1st one
                     with non-prefixed, set based on ploidy and phasing of other
@@ -3126,6 +3126,9 @@ static int vcf_parse_format_fill5(kstring_t *s, const bcf_hdr_t *h, bcf1_t *v,
                         ploidy++;
                         if (*t == '.') {
                             ++t, x[l++] = is_phased;
+                            if (l==1) {   //for 1st allele only
+                                unknown1 = 1;
+                            }
                         } else {
                             const char *tt = t;
                             uint32_t val;
@@ -3151,9 +3154,11 @@ static int vcf_parse_format_fill5(kstring_t *s, const bcf_hdr_t *h, bcf1_t *v,
                         /* no explicit phasing for 1st allele, set based on
                          other alleles and ploidy */
                         if (ploidy == 1) {  //implicitly phased
-                            x[0]|= 1;
+                            if (!unknown1) {
+                                x[0] |= 1;
+                            }
                         } else {            //set by other unphased alleles
-                            x[0] |= anyunphased ? 0 : 1;
+                            x[0] |= (anyunphased)? 0 : 1;
                         }
                     }
                     // Possibly check max against v->n_allele instead?