Add test for samtools#871. Make F_MISSING and F_PASS aware of the necessacity to keep the FORMAT fields.

Author: pd3

filter.c

@@ -1028,7 +1028,6 @@ static int func_npass(filter_t *flt, bcf1_t *line, token_t *rtok, token_t **stac
     rtok->nvalues = 1;
     rtok->values[0] = rtok->tag[0]=='N' ? npass : (line->n_sample ? 1.0*npass/line->n_sample : 0);
     rtok->nsamples = 0;
-
     return 1;
 }
 static void filters_set_nalt(filter_t *flt, bcf1_t *line, token_t *tok)
@@ -2127,12 +2126,14 @@ static int filters_init1(filter_t *filter, char *str, int len, token_t *tok)
         }
         else if ( !strncasecmp(str,"N_MISSING",len) )
         {
+            filter->max_unpack |= BCF_UN_FMT;
             tok->setter = &filters_set_nmissing;
             tok->tag = strdup("N_MISSING");
             return 0;
         }
         else if ( !strncasecmp(str,"F_MISSING",len) )
         {
+            filter->max_unpack |= BCF_UN_FMT;
             tok->setter = &filters_set_nmissing;
             tok->tag = strdup("F_MISSING");
             return 0;
@@ -2519,8 +2520,18 @@ filter_t *filter_init(bcf_hdr_t *hdr, const char *str)
                 tok->hdr_id    = -1;
                 tok->pass_site = -1;
                 tok->threshold = -1.0;
-                if ( !strncasecmp(tmp-len,"N_PASS",6) ) { tok->func = func_npass; tok->tag = strdup("N_PASS"); }
-                else if ( !strncasecmp(tmp-len,"F_PASS",6) ) { tok->func = func_npass; tok->tag = strdup("F_PASS"); }
+                if ( !strncasecmp(tmp-len,"N_PASS",6) )
+                {
+                    filter->max_unpack |= BCF_UN_FMT;
+                    tok->func = func_npass;
+                    tok->tag = strdup("N_PASS");
+                }
+                else if ( !strncasecmp(tmp-len,"F_PASS",6) )
+                {
+                    filter->max_unpack |= BCF_UN_FMT;
+                    tok->func = func_npass;
+                    tok->tag = strdup("F_PASS");
+                }
                 else error("The function \"%s\" is not supported\n", tmp-len);
                 continue;
             }

test/filter.28.out

@@ -0,0 +1 @@
+3162007

test/filter.6.vcf

@@ -0,0 +1,6 @@
+##fileformat=VCFv4.2
+##FORMAT=<ID=GT,Number=1,Type=String,Description="genotype">
+##contig=<ID=1,assembly=b37,length=249250621>
+##reference=file:///lustre/scratch105/projects/g1k/ref/main_project/human_g1k_v37.fasta
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	A	B	C	D	E
+1	3162007	.	A	C	.	.	.	GT	0/0	./.	0/1	0/0	0/0

test/test.pl

@@ -240,6 +240,10 @@
 test_vcf_filter($opts,in=>'filter.5',out=>'filter.26.out',args=>q[-i'QUAL=="."'],fmt=>'%POS\\t%QUAL\\n');
 test_vcf_filter($opts,in=>'filter.2',out=>'filter.27.out',args=>q[-i'N_PASS(DP>32)=1'],fmt=>'[%POS\\t%SAMPLE\\t%DP\\n]');
 test_vcf_filter($opts,in=>'filter.2',out=>'filter.27.out',args=>q[-i'F_PASS(DP>32)=0.5'],fmt=>'[%POS\\t%SAMPLE\\t%DP\\n]');
+test_vcf_filter($opts,in=>'filter.6',out=>'filter.28.out',args=>q[-i'F_MISSING>=1/5'],fmt=>'%POS\\n');
+test_vcf_filter($opts,in=>'filter.6',out=>'filter.28.out',args=>q[-i'F_MISSING>=0.2'],fmt=>'%POS\\n');
+test_vcf_filter($opts,in=>'filter.6',out=>'filter.28.out',args=>q[-i'F_PASS(GT=="mis")>=1/5'],fmt=>'%POS\\n');
+test_vcf_filter($opts,in=>'filter.6',out=>'filter.28.out',args=>q[-i'F_PASS(GT=="mis")>=0.2'],fmt=>'%POS\\n');
 test_vcf_sort($opts,in=>'sort',out=>'sort.out',args=>q[-m 0],fmt=>'%CHROM\\t%POS\\t%REF,%ALT\\n');
 test_vcf_sort($opts,in=>'sort',out=>'sort.out',args=>q[-m 1000],fmt=>'%CHROM\\t%POS\\t%REF,%ALT\\n');
 test_vcf_regions($opts,in=>'regions');