* fixed a minor issue in printing VCFs

[samtools.git] / bcftools / bcfutils.c

diff --git a/bcftools/bcfutils.c b/bcftools/bcfutils.c
index 861e33fa6bf1a9e24955b9c3f459d6d248a83b5a..3c6ed735fdd352a850a56e91987d4d04ad71254d 100644 (file)
--- a/bcftools/bcfutils.c
+++ b/bcftools/bcfutils.c
@@ -1,4 +1,6 @@
+#include <string.h>
 #include "bcf.h"
+#include "kstring.h"
 #include "khash.h"
 KHASH_MAP_INIT_STR(str2id, int)
 
@@ -20,11 +22,6 @@ void *bcf_str2id_init()
        return kh_init(str2id);
 }
 
-int bcf_str2id_put(void *_hash, const char *str, int id)
-{
-       return 0;
-}
-
 void bcf_str2id_destroy(void *_hash)
 {
        khash_t(str2id) *hash = (khash_t(str2id)*)_hash;
@@ -40,3 +37,97 @@ int bcf_str2id(void *_hash, const char *str)
        return k == kh_end(hash)? -1 : kh_val(hash, k);
 }
 
+int bcf_str2id_add(void *_hash, const char *str)
+{
+       khint_t k;
+       int ret;
+       khash_t(str2id) *hash = (khash_t(str2id)*)_hash;
+       if (!hash) return -1;
+       k = kh_put(str2id, hash, str, &ret);
+       if (ret == 0) return kh_val(hash, k);
+       kh_val(hash, k) = kh_size(hash) - 1;
+       return kh_val(hash, k);
+}
+
+int bcf_shrink_alt(bcf1_t *b, int n)
+{
+       char *p;
+       int i, j, k, *z, n_smpl = b->n_smpl;
+       if (b->n_alleles <= n) return -1;
+       if (n > 1) {
+               for (p = b->alt, k = 1; *p; ++p)
+                       if (*p == ',' && ++k == n) break;
+               *p = '\0';
+       } else p = b->alt, *p = '\0';
+       ++p;
+       memmove(p, b->flt, b->str + b->l_str - b->flt);
+       b->l_str -= b->flt - p;
+       z = alloca(sizeof(int) / 2 * n * (n+1));
+       for (i = k = 0; i < n; ++i)
+               for (j = 0; j < n - i; ++j)
+                       z[k++] = i * b->n_alleles + j;
+       for (i = 0; i < b->n_gi; ++i) {
+               bcf_ginfo_t *g = b->gi + i;
+               if (g->fmt == bcf_str2int("PL", 2)) {
+                       int l, x = b->n_alleles * (b->n_alleles + 1) / 2;
+                       uint8_t *d = (uint8_t*)g->data;
+                       g->len = n * (n + 1) / 2;
+                       for (l = k = 0; l < n_smpl; ++l) {
+                               uint8_t *dl = d + l * x;
+                               for (j = 0; j < g->len; ++j) d[k++] = dl[z[j]];
+                       }
+               } // FIXME: to add GL
+       }
+       b->n_alleles = n;
+       bcf_sync(b);
+       return 0;
+}
+
+int bcf_gl2pl(bcf1_t *b)
+{
+       char *p;
+       int i, n_smpl = b->n_smpl;
+       bcf_ginfo_t *g;
+       float *d0;
+       uint8_t *d1;
+       if (strstr(b->fmt, "PL")) return -1;
+       if ((p = strstr(b->fmt, "GL")) == 0) return -1;
+       *p = 'P';
+       for (i = 0; i < b->n_gi; ++i)
+               if (b->gi[i].fmt == bcf_str2int("GL", 2))
+                       break;
+       g = b->gi + i;
+       g->fmt = bcf_str2int("PL", 2);
+       g->len /= 4; // 4 == sizeof(float)
+       d0 = (float*)g->data; d1 = (uint8_t*)g->data;
+       for (i = 0; i < n_smpl * g->len; ++i) {
+               int x = (int)(-10. * d0[i] + .499);
+               if (x > 255) x = 255;
+               if (x < 0) x = 0;
+               d1[i] = x;
+       }
+       return 0;
+}
+/* FIXME: this function will fail given AB:GTX:GT. BCFtools never
+ * produces such FMT, but others may do. */
+int bcf_fix_gt(bcf1_t *b)
+{
+       char *s;
+       int i;
+       uint32_t tmp;
+       bcf_ginfo_t gt;
+       // check the presence of the GT FMT
+       if ((s = strstr(b->fmt, ":GT")) == 0) return 0; // no GT or GT is already the first
+       if (s[3] != '\0' && s[3] != ':') return 0; // :GTX in fact
+       tmp = bcf_str2int("GT", 2);
+       for (i = 0; i < b->n_gi; ++i)
+               if (b->gi[i].fmt == tmp) break;
+       if (i == b->n_gi) return 0; // no GT in b->gi; probably a bug...
+       gt = b->gi[i];
+       // move GT to the first
+       for (; i > 0; --i) b->gi[i] = b->gi[i-1];
+       b->gi[0] = gt;
+       memmove(b->fmt + 3, b->fmt, s + 1 - b->fmt);
+       b->fmt[0] = 'G'; b->fmt[1] = 'T'; b->fmt[2] = ':';
+       return 0;
+}