annotate pyPRADA_1.2/tools/samtools-0.1.16/bcftools/bcf.c @ 0:acc2ca1a3ba4

Uploaded
author siyuan
date Thu, 20 Feb 2014 00:44:58 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
1 #include <string.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
2 #include <ctype.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
3 #include <stdio.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
4 #include "kstring.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
5 #include "bcf.h"
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
6
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
7 bcf_t *bcf_open(const char *fn, const char *mode)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
8 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
9 bcf_t *b;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
10 b = calloc(1, sizeof(bcf_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
11 if (strchr(mode, 'w')) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
12 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdout), mode);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
13 } else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
14 b->fp = strcmp(fn, "-")? bgzf_open(fn, mode) : bgzf_fdopen(fileno(stdin), mode);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
15 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
16 #ifndef BCF_LITE
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
17 b->fp->owned_file = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
18 #endif
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
19 return b;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
20 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
21
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
22 int bcf_close(bcf_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
23 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
24 int ret;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
25 if (b == 0) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
26 ret = bgzf_close(b->fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
27 free(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
28 return ret;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
29 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
30
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
31 int bcf_hdr_write(bcf_t *b, const bcf_hdr_t *h)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
32 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
33 if (b == 0 || h == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
34 bgzf_write(b->fp, "BCF\4", 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
35 bgzf_write(b->fp, &h->l_nm, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
36 bgzf_write(b->fp, h->name, h->l_nm);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
37 bgzf_write(b->fp, &h->l_smpl, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
38 bgzf_write(b->fp, h->sname, h->l_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
39 bgzf_write(b->fp, &h->l_txt, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
40 bgzf_write(b->fp, h->txt, h->l_txt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
41 bgzf_flush(b->fp);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
42 return 16 + h->l_nm + h->l_smpl + h->l_txt;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
43 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
44
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
45 bcf_hdr_t *bcf_hdr_read(bcf_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
46 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
47 uint8_t magic[4];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
48 bcf_hdr_t *h;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
49 if (b == 0) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
50 h = calloc(1, sizeof(bcf_hdr_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
51 bgzf_read(b->fp, magic, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
52 bgzf_read(b->fp, &h->l_nm, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
53 h->name = malloc(h->l_nm);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
54 bgzf_read(b->fp, h->name, h->l_nm);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
55 bgzf_read(b->fp, &h->l_smpl, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
56 h->sname = malloc(h->l_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
57 bgzf_read(b->fp, h->sname, h->l_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
58 bgzf_read(b->fp, &h->l_txt, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
59 h->txt = malloc(h->l_txt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
60 bgzf_read(b->fp, h->txt, h->l_txt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
61 bcf_hdr_sync(h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
62 return h;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
63 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
64
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
65 void bcf_hdr_destroy(bcf_hdr_t *h)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
66 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
67 if (h == 0) return;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
68 free(h->name); free(h->sname); free(h->txt); free(h->ns); free(h->sns);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
69 free(h);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
70 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
71
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
72 static inline char **cnt_null(int l, char *str, int *_n)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
73 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
74 int n = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
75 char *p, **list;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
76 *_n = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
77 if (l == 0 || str == 0) return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
78 for (p = str; p != str + l; ++p)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
79 if (*p == 0) ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
80 *_n = n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
81 list = calloc(n, sizeof(void*));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
82 list[0] = str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
83 for (p = str, n = 1; p < str + l - 1; ++p)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
84 if (*p == 0) list[n++] = p + 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
85 return list;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
86 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
87
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
88 int bcf_hdr_sync(bcf_hdr_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
89 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
90 if (b == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
91 if (b->ns) free(b->ns);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
92 if (b->sns) free(b->sns);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
93 if (b->l_nm) b->ns = cnt_null(b->l_nm, b->name, &b->n_ref);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
94 else b->ns = 0, b->n_ref = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
95 b->sns = cnt_null(b->l_smpl, b->sname, &b->n_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
96 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
97 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
98
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
99 int bcf_sync(bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
100 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
101 char *p, *tmp[5];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
102 int i, n, n_smpl = b->n_smpl;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
103 ks_tokaux_t aux;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
104 // set ref, alt, flt, info, fmt
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
105 b->ref = b->alt = b->flt = b->info = b->fmt = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
106 for (p = b->str, n = 0; p < b->str + b->l_str; ++p) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
107 if (*p == 0 && p+1 != b->str + b->l_str) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
108 if (n == 5) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
109 ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
110 break;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
111 } else tmp[n++] = p + 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
112 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
113 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
114 if (n != 5) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
115 fprintf(stderr, "[%s] incorrect number of fields (%d != 5) at %d:%d\n", __func__, n, b->tid, b->pos);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
116 return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
117 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
118 b->ref = tmp[0]; b->alt = tmp[1]; b->flt = tmp[2]; b->info = tmp[3]; b->fmt = tmp[4];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
119 // set n_alleles
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
120 if (*b->alt == 0) b->n_alleles = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
121 else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
122 for (p = b->alt, n = 1; *p; ++p)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
123 if (*p == ',') ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
124 b->n_alleles = n + 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
125 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
126 // set n_gi and gi[i].fmt
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
127 for (p = b->fmt, n = 1; *p; ++p)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
128 if (*p == ':') ++n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
129 if (n > b->m_gi) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
130 int old_m = b->m_gi;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
131 b->m_gi = n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
132 kroundup32(b->m_gi);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
133 b->gi = realloc(b->gi, b->m_gi * sizeof(bcf_ginfo_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
134 memset(b->gi + old_m, 0, (b->m_gi - old_m) * sizeof(bcf_ginfo_t));
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
135 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
136 b->n_gi = n;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
137 for (p = kstrtok(b->fmt, ":", &aux), n = 0; p; p = kstrtok(0, 0, &aux))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
138 b->gi[n++].fmt = bcf_str2int(p, aux.p - p);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
139 // set gi[i].len
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
140 for (i = 0; i < b->n_gi; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
141 if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
142 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
143 } else if (b->gi[i].fmt == bcf_str2int("DP", 2) || b->gi[i].fmt == bcf_str2int("HQ", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
144 b->gi[i].len = 2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
145 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2) || b->gi[i].fmt == bcf_str2int("GT", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
146 b->gi[i].len = 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
147 } else if (b->gi[i].fmt == bcf_str2int("SP", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
148 b->gi[i].len = 4;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
149 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
150 b->gi[i].len = b->n_alleles * (b->n_alleles + 1) / 2 * 4;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
151 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
152 b->gi[i].data = realloc(b->gi[i].data, n_smpl * b->gi[i].len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
153 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
154 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
155 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
156
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
157 int bcf_write(bcf_t *bp, const bcf_hdr_t *h, const bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
158 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
159 int i, l = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
160 if (b == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
161 bgzf_write(bp->fp, &b->tid, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
162 bgzf_write(bp->fp, &b->pos, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
163 bgzf_write(bp->fp, &b->qual, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
164 bgzf_write(bp->fp, &b->l_str, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
165 bgzf_write(bp->fp, b->str, b->l_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
166 l = 12 + b->l_str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
167 for (i = 0; i < b->n_gi; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
168 bgzf_write(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
169 l += b->gi[i].len * h->n_smpl;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
170 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
171 return l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
172 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
173
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
174 int bcf_read(bcf_t *bp, const bcf_hdr_t *h, bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
175 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
176 int i, l = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
177 if (b == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
178 if (bgzf_read(bp->fp, &b->tid, 4) == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
179 b->n_smpl = h->n_smpl;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
180 bgzf_read(bp->fp, &b->pos, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
181 bgzf_read(bp->fp, &b->qual, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
182 bgzf_read(bp->fp, &b->l_str, 4);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
183 if (b->l_str > b->m_str) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
184 b->m_str = b->l_str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
185 kroundup32(b->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
186 b->str = realloc(b->str, b->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
187 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
188 bgzf_read(bp->fp, b->str, b->l_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
189 l = 12 + b->l_str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
190 if (bcf_sync(b) < 0) return -2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
191 for (i = 0; i < b->n_gi; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
192 bgzf_read(bp->fp, b->gi[i].data, b->gi[i].len * h->n_smpl);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
193 l += b->gi[i].len * h->n_smpl;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
194 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
195 return l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
196 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
197
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
198 int bcf_destroy(bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
199 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
200 int i;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
201 if (b == 0) return -1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
202 free(b->str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
203 for (i = 0; i < b->m_gi; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
204 free(b->gi[i].data);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
205 free(b->gi);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
206 free(b);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
207 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
208 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
209
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
210 static inline void fmt_str(const char *p, kstring_t *s)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
211 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
212 if (*p == 0) kputc('.', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
213 else kputs(p, s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
214 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
215
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
216 void bcf_fmt_core(const bcf_hdr_t *h, bcf1_t *b, kstring_t *s)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
217 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
218 int i, j, x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
219 s->l = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
220 if (h->n_ref) kputs(h->ns[b->tid], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
221 else kputw(b->tid, s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
222 kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
223 kputw(b->pos + 1, s); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
224 fmt_str(b->str, s); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
225 fmt_str(b->ref, s); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
226 fmt_str(b->alt, s); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
227 ksprintf(s, "%.3g", b->qual); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
228 fmt_str(b->flt, s); kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
229 fmt_str(b->info, s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
230 if (b->fmt[0]) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
231 kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
232 fmt_str(b->fmt, s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
233 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
234 x = b->n_alleles * (b->n_alleles + 1) / 2;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
235 if (b->n_gi == 0) return;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
236 for (j = 0; j < h->n_smpl; ++j) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
237 kputc('\t', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
238 for (i = 0; i < b->n_gi; ++i) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
239 if (i) kputc(':', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
240 if (b->gi[i].fmt == bcf_str2int("PL", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
241 uint8_t *d = (uint8_t*)b->gi[i].data + j * x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
242 int k;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
243 for (k = 0; k < x; ++k) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
244 if (k > 0) kputc(',', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
245 kputw(d[k], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
246 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
247 } else if (b->gi[i].fmt == bcf_str2int("DP", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
248 kputw(((uint16_t*)b->gi[i].data)[j], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
249 } else if (b->gi[i].fmt == bcf_str2int("GQ", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
250 kputw(((uint8_t*)b->gi[i].data)[j], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
251 } else if (b->gi[i].fmt == bcf_str2int("SP", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
252 kputw(((int32_t*)b->gi[i].data)[j], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
253 } else if (b->gi[i].fmt == bcf_str2int("GT", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
254 int y = ((uint8_t*)b->gi[i].data)[j];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
255 if (y>>7&1) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
256 kputsn("./.", 3, s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
257 } else {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
258 kputc('0' + (y>>3&7), s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
259 kputc("/|"[y>>6&1], s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
260 kputc('0' + (y&7), s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
261 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
262 } else if (b->gi[i].fmt == bcf_str2int("GL", 2)) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
263 float *d = (float*)b->gi[i].data + j * x;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
264 int k;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
265 //printf("- %lx\n", d);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
266 for (k = 0; k < x; ++k) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
267 if (k > 0) kputc(',', s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
268 ksprintf(s, "%.2f", d[k]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
269 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
270 } else kputc('.', s); // custom fields
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
271 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
272 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
273 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
274
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
275 char *bcf_fmt(const bcf_hdr_t *h, bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
276 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
277 kstring_t s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
278 s.l = s.m = 0; s.s = 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
279 bcf_fmt_core(h, b, &s);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
280 return s.s;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
281 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
282
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
283 int bcf_append_info(bcf1_t *b, const char *info, int l)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
284 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
285 int shift = b->fmt - b->str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
286 int l_fmt = b->l_str - shift;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
287 char *ori = b->str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
288 if (b->l_str + l > b->m_str) { // enlarge if necessary
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
289 b->m_str = b->l_str + l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
290 kroundup32(b->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
291 b->str = realloc(b->str, b->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
292 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
293 memmove(b->str + shift + l, b->str + shift, l_fmt); // move the FORMAT field
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
294 memcpy(b->str + shift - 1, info, l); // append to the INFO field
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
295 b->str[shift + l - 1] = '\0';
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
296 b->fmt = b->str + shift + l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
297 b->l_str += l;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
298 if (ori != b->str) bcf_sync(b); // synchronize when realloc changes the pointer
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
299 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
300 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
301
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
302 int bcf_cpy(bcf1_t *r, const bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
303 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
304 char *t1 = r->str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
305 bcf_ginfo_t *t2 = r->gi;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
306 int i, t3 = r->m_str, t4 = r->m_gi;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
307 *r = *b;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
308 r->str = t1; r->gi = t2; r->m_str = t3; r->m_gi = t4;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
309 if (r->m_str < b->m_str) {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
310 r->m_str = b->m_str;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
311 r->str = realloc(r->str, r->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
312 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
313 memcpy(r->str, b->str, r->m_str);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
314 bcf_sync(r); // calling bcf_sync() is simple but inefficient
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
315 for (i = 0; i < r->n_gi; ++i)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
316 memcpy(r->gi[i].data, b->gi[i].data, r->n_smpl * r->gi[i].len);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
317 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
318 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
319
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
320 int bcf_is_indel(const bcf1_t *b)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
321 {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
322 char *p;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
323 if (strlen(b->ref) > 1) return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
324 for (p = b->alt; *p; ++p)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
325 if (*p != ',' && p[1] != ',' && p[1] != '\0')
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
326 return 1;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
327 return 0;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
328 }