Updated samtools to 0.1.19

[rsem.git] / sam / bcftools / bcf.c

diff --git a/sam/bcftools/bcf.c b/sam/bcftools/bcf.c
index 84a8e767ec58adb397630f609a52670779f99c89..24728dbe1515e7cc1171a8cc08f842bafa167d5e 100644 (file)
--- a/sam/bcftools/bcf.c
+++ b/sam/bcftools/bcf.c
@@ -13,9 +13,6 @@ bcf_t *bcf_open(const char *fn, const char *mode)
        } else {
                b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
        }
-#ifndef BCF_LITE
-       b->fp->owned_file = 1;
-#endif
        return b;
 }
 
@@ -140,7 +137,7 @@ int bcf_sync(bcf1_t *b)
        for (i = 0; i < b->n_gi; ++i) {
                if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
                        b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
-               } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2)) {
+               } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2) || b->gi[i].fmt == bcf_str2int("DV", 2)) {
                        b->gi[i].len = 2;
                } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) {
                        b->gi[i].len = 1;
@@ -233,32 +230,59 @@ void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
        }
        x = b->n_alleles * (b->n_alleles + 1) / 2;
        if (b->n_gi == 0) return;
+    int iPL = -1;
+    if ( b->n_alleles > 2 ) {
+        for (i=0; i<b->n_gi; i++) {
+            if ( b->gi[i].fmt == bcf_str2int("PL", 2) ) {
+                iPL = i;
+                break;
+            }
+        }
+    }
        for (j = 0; j < h->n_smpl; ++j) {
+        int ploidy = b->ploidy ? b->ploidy[j] : 2;
                kputc('\t', s);
                for (i = 0; i < b->n_gi; ++i) {
                        if (i) kputc(':', s);
                        if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
                                uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
                                int k;
-                               for (k = 0; k < x; ++k) {
-                                       if (k > 0) kputc(',', s);
-                                       kputw(d[k], s);
-                               }
-                       } else if (b->gi[i].fmt == bcf_str2int("DP", 2)) {
+                if ( ploidy==1 )
+                    for (k=0; k<b->n_alleles; k++)
+                    {
+                        if (k>0) kputc(',', s);
+                        kputw(d[(k+1)*(k+2)/2-1], s);
+                    }
+                else
+                    for (k = 0; k < x; ++k) {
+                        if (k > 0) kputc(',', s);
+                        kputw(d[k], s);
+                    }
+                       } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("DV", 2)) {
                                kputw(((uint16_t*)b->gi[i].data)[j], s);
                        } else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
                                kputw(((uint8_t*)b->gi[i].data)[j], s);
                        } else if (b->gi[i].fmt == bcf_str2int("SP", 2)) {
                                kputw(((int32_t*)b->gi[i].data)[j], s);
                        } else if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
-                               int y = ((uint8_t*)b->gi[i].data)[j];
-                               if (y>>7&1) {
-                                       kputsn("./.", 3, s);
-                               } else {
-                                       kputc('0' + (y>>3&7), s);
-                                       kputc("/|"[y>>6&1], s);
-                                       kputc('0' + (y&7), s);
-                               }
+                int y = ((uint8_t*)b->gi[i].data)[j];
+                if ( ploidy==1 )
+                {
+                    if ( y>>7&1 )
+                        kputc('.', s);
+                    else 
+                        kputc('0' + (y>>3&7), s);
+                }
+                else
+                {
+                    if ( y>>7&1 )
+                        kputsn("./.", 3, s);
+                    else { 
+                        kputc('0' + (y>>3&7), s);
+                        kputc("/|"[y>>6&1], s);
+                        kputc('0' + (y&7), s);
+                    }
+                }
                        } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
                                float *d = (float*)b->gi[i].data + j * x;
                                int k;
@@ -299,6 +323,50 @@ int bcf_append_info(bcf1_t *b, const char *info, int l)
        return 0;
 }
 
+int remove_tag(char *str, const char *tag, char delim)
+{
+    char *tmp = str, *p;
+    int len_diff = 0, ori_len = strlen(str);
+    while ( *tmp && (p = strstr(tmp,tag)) )
+    {
+        if ( p>str )
+        {
+            if ( *(p-1)!=delim ) { tmp=p+1; continue; } // shared substring
+            p--;
+        }
+        char *q=p+1;
+        while ( *q && *q!=delim ) q++;
+        if ( p==str && *q ) q++;        // the tag is first, don't move the delim char
+        len_diff += q-p;
+        if ( ! *q ) { *p = 0; break; }  // the tag was last, no delim follows
+        else
+            memmove(p,q,ori_len-(int)(p-str)-(int)(q-p));  // *q==delim
+    }
+    if ( len_diff==ori_len )
+        str[0]='.', str[1]=0, len_diff--;
+
+    return len_diff;
+}
+
+
+void rm_info(kstring_t *s, const char *key)
+{
+    char *p = s->s; 
+    int n = 0;
+    while ( n<4 )
+    {
+        if ( !*p ) n++;
+        p++;
+    }
+    char *q = p+1; 
+    while ( *q && q-s->s<s->l ) q++;
+
+    int nrm = remove_tag(p, key, ';');
+    if ( nrm )
+        memmove(q-nrm, q, s->s+s->l-q+1);
+    s->l -= nrm;
+}
+
 int bcf_cpy(bcf1_t *r, const bcf1_t *b)
 {
        char *t1 = r->str;