annotate SNV/SNVMix2_source/SNVMix2-v0.12.1-rc1/samtools-0.1.6/bam.c @ 0:74f5ea818cea

Uploaded
author ryanmorin
date Wed, 12 Oct 2011 19:50:38 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
1 #include <stdio.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
2 #include <ctype.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
3 #include <assert.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
4 #include "bam.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
5 #include "bam_endian.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
6 #include "kstring.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
7
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
8 int bam_is_be = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
9 char *bam_flag2char_table = "pPuUrR12sfd\0\0\0\0\0";
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
10
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
11 /**************************
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
12 * CIGAR related routines *
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
13 **************************/
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
14
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
15 int bam_segreg(int32_t pos, const bam1_core_t *c, const uint32_t *cigar, bam_segreg_t *reg)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
16 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
17 unsigned k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
18 int32_t x = c->pos, y = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
19 int state = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
20 for (k = 0; k < c->n_cigar; ++k) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
21 int op = cigar[k] & BAM_CIGAR_MASK; // operation
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
22 int l = cigar[k] >> BAM_CIGAR_SHIFT; // length
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
23 if (state == 0 && (op == BAM_CMATCH || op == BAM_CDEL || op == BAM_CINS) && x + l > pos) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
24 reg->tbeg = x; reg->qbeg = y; reg->cbeg = k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
25 state = 1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
26 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
27 if (op == BAM_CMATCH) { x += l; y += l; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
28 else if (op == BAM_CDEL || op == BAM_CREF_SKIP) x += l;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
29 else if (op == BAM_CINS || op == BAM_CSOFT_CLIP) y += l;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
30 if (state == 1 && (op == BAM_CSOFT_CLIP || op == BAM_CHARD_CLIP || op == BAM_CREF_SKIP || k == c->n_cigar - 1)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
31 reg->tend = x; reg->qend = y; reg->cend = k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
32 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
33 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
34 return state? 0 : -1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
35 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
36
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
37 uint32_t bam_calend(const bam1_core_t *c, const uint32_t *cigar)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
38 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
39 uint32_t k, end;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
40 end = c->pos;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
41 for (k = 0; k < c->n_cigar; ++k) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
42 int op = cigar[k] & BAM_CIGAR_MASK;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
43 if (op == BAM_CMATCH || op == BAM_CDEL || op == BAM_CREF_SKIP)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
44 end += cigar[k] >> BAM_CIGAR_SHIFT;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
45 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
46 return end;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
47 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
48
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
49 int32_t bam_cigar2qlen(const bam1_core_t *c, const uint32_t *cigar)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
50 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
51 uint32_t k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
52 int32_t l = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
53 for (k = 0; k < c->n_cigar; ++k) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
54 int op = cigar[k] & BAM_CIGAR_MASK;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
55 if (op == BAM_CMATCH || op == BAM_CINS || op == BAM_CSOFT_CLIP)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
56 l += cigar[k] >> BAM_CIGAR_SHIFT;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
57 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
58 return l;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
59 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
60
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
61 /********************
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
62 * BAM I/O routines *
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
63 ********************/
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
64
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
65 bam_header_t *bam_header_init()
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
66 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
67 bam_is_be = bam_is_big_endian();
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
68 return (bam_header_t*)calloc(1, sizeof(bam_header_t));
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
69 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
70
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
71 void bam_header_destroy(bam_header_t *header)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
72 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
73 int32_t i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
74 extern void bam_destroy_header_hash(bam_header_t *header);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
75 if (header == 0) return;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
76 if (header->target_name) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
77 for (i = 0; i < header->n_targets; ++i)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
78 free(header->target_name[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
79 free(header->target_name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
80 free(header->target_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
81 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
82 free(header->text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
83 #ifndef BAM_NO_HASH
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
84 if (header->rg2lib) bam_strmap_destroy(header->rg2lib);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
85 bam_destroy_header_hash(header);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
86 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
87 free(header);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
88 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
89
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
90 bam_header_t *bam_header_read(bamFile fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
91 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
92 bam_header_t *header;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
93 char buf[4];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
94 int32_t i = 1, name_len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
95 // check EOF
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
96 i = bgzf_check_EOF(fp);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
97 if (i < 0) fprintf(stderr, "[bam_header_read] read from pipe; skip EOF checking.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
98 else if (i == 0) fprintf(stderr, "[bam_header_read] EOF marker is absent.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
99 // read "BAM1"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
100 if (bam_read(fp, buf, 4) != 4) return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
101 if (strncmp(buf, "BAM\001", 4)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
102 fprintf(stderr, "[bam_header_read] wrong header\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
103 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
104 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
105 header = bam_header_init();
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
106 // read plain text and the number of reference sequences
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
107 bam_read(fp, &header->l_text, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
108 if (bam_is_be) bam_swap_endian_4p(&header->l_text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
109 header->text = (char*)calloc(header->l_text + 1, 1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
110 bam_read(fp, header->text, header->l_text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
111 bam_read(fp, &header->n_targets, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
112 if (bam_is_be) bam_swap_endian_4p(&header->n_targets);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
113 // read reference sequence names and lengths
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
114 header->target_name = (char**)calloc(header->n_targets, sizeof(char*));
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
115 header->target_len = (uint32_t*)calloc(header->n_targets, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
116 for (i = 0; i != header->n_targets; ++i) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
117 bam_read(fp, &name_len, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
118 if (bam_is_be) bam_swap_endian_4p(&name_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
119 header->target_name[i] = (char*)calloc(name_len, 1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
120 bam_read(fp, header->target_name[i], name_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
121 bam_read(fp, &header->target_len[i], 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
122 if (bam_is_be) bam_swap_endian_4p(&header->target_len[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
123 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
124 return header;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
125 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
126
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
127 int bam_header_write(bamFile fp, const bam_header_t *header)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
128 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
129 char buf[4];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
130 int32_t i, name_len, x;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
131 // write "BAM1"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
132 strncpy(buf, "BAM\001", 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
133 bam_write(fp, buf, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
134 // write plain text and the number of reference sequences
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
135 if (bam_is_be) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
136 x = bam_swap_endian_4(header->l_text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
137 bam_write(fp, &x, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
138 if (header->l_text) bam_write(fp, header->text, header->l_text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
139 x = bam_swap_endian_4(header->n_targets);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
140 bam_write(fp, &x, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
141 } else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
142 bam_write(fp, &header->l_text, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
143 if (header->l_text) bam_write(fp, header->text, header->l_text);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
144 bam_write(fp, &header->n_targets, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
145 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
146 // write sequence names and lengths
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
147 for (i = 0; i != header->n_targets; ++i) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
148 char *p = header->target_name[i];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
149 name_len = strlen(p) + 1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
150 if (bam_is_be) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
151 x = bam_swap_endian_4(name_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
152 bam_write(fp, &x, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
153 } else bam_write(fp, &name_len, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
154 bam_write(fp, p, name_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
155 if (bam_is_be) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
156 x = bam_swap_endian_4(header->target_len[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
157 bam_write(fp, &x, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
158 } else bam_write(fp, &header->target_len[i], 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
159 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
160 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
161 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
162
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
163 static void swap_endian_data(const bam1_core_t *c, int data_len, uint8_t *data)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
164 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
165 uint8_t *s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
166 uint32_t i, *cigar = (uint32_t*)(data + c->l_qname);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
167 s = data + c->n_cigar*4 + c->l_qname + c->l_qseq + (c->l_qseq + 1)/2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
168 for (i = 0; i < c->n_cigar; ++i) bam_swap_endian_4p(&cigar[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
169 while (s < data + data_len) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
170 uint8_t type;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
171 s += 2; // skip key
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
172 type = toupper(*s); ++s; // skip type
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
173 if (type == 'C' || type == 'A') ++s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
174 else if (type == 'S') { bam_swap_endian_2p(s); s += 2; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
175 else if (type == 'I' || type == 'F') { bam_swap_endian_4p(s); s += 4; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
176 else if (type == 'D') { bam_swap_endian_8p(s); s += 8; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
177 else if (type == 'Z' || type == 'H') { while (*s) ++s; ++s; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
178 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
179 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
180
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
181 int bam_read1(bamFile fp, bam1_t *b)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
182 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
183 bam1_core_t *c = &b->core;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
184 int32_t block_len, ret, i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
185 uint32_t x[8];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
186
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
187 assert(BAM_CORE_SIZE == 32);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
188 if ((ret = bam_read(fp, &block_len, 4)) != 4) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
189 if (ret == 0) return -1; // normal end-of-file
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
190 else return -2; // truncated
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
191 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
192 if (bam_read(fp, x, BAM_CORE_SIZE) != BAM_CORE_SIZE) return -3;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
193 if (bam_is_be) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
194 bam_swap_endian_4p(&block_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
195 for (i = 0; i < 8; ++i) bam_swap_endian_4p(x + i);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
196 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
197 c->tid = x[0]; c->pos = x[1];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
198 c->bin = x[2]>>16; c->qual = x[2]>>8&0xff; c->l_qname = x[2]&0xff;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
199 c->flag = x[3]>>16; c->n_cigar = x[3]&0xffff;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
200 c->l_qseq = x[4];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
201 c->mtid = x[5]; c->mpos = x[6]; c->isize = x[7];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
202 b->data_len = block_len - BAM_CORE_SIZE;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
203 if (b->m_data < b->data_len) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
204 b->m_data = b->data_len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
205 kroundup32(b->m_data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
206 b->data = (uint8_t*)realloc(b->data, b->m_data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
207 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
208 if (bam_read(fp, b->data, b->data_len) != b->data_len) return -4;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
209 b->l_aux = b->data_len - c->n_cigar * 4 - c->l_qname - c->l_qseq - (c->l_qseq+1)/2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
210 if (bam_is_be) swap_endian_data(c, b->data_len, b->data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
211 return 4 + block_len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
212 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
213
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
214 inline int bam_write1_core(bamFile fp, const bam1_core_t *c, int data_len, uint8_t *data)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
215 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
216 uint32_t x[8], block_len = data_len + BAM_CORE_SIZE, y;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
217 int i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
218 assert(BAM_CORE_SIZE == 32);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
219 x[0] = c->tid;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
220 x[1] = c->pos;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
221 x[2] = (uint32_t)c->bin<<16 | c->qual<<8 | c->l_qname;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
222 x[3] = (uint32_t)c->flag<<16 | c->n_cigar;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
223 x[4] = c->l_qseq;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
224 x[5] = c->mtid;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
225 x[6] = c->mpos;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
226 x[7] = c->isize;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
227 if (bam_is_be) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
228 for (i = 0; i < 8; ++i) bam_swap_endian_4p(x + i);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
229 y = block_len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
230 bam_write(fp, bam_swap_endian_4p(&y), 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
231 swap_endian_data(c, data_len, data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
232 } else bam_write(fp, &block_len, 4);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
233 bam_write(fp, x, BAM_CORE_SIZE);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
234 bam_write(fp, data, data_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
235 if (bam_is_be) swap_endian_data(c, data_len, data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
236 return 4 + block_len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
237 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
238
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
239 int bam_write1(bamFile fp, const bam1_t *b)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
240 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
241 return bam_write1_core(fp, &b->core, b->data_len, b->data);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
242 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
243
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
244 char *bam_format1_core(const bam_header_t *header, const bam1_t *b, int of)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
245 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
246 uint8_t *s = bam1_seq(b), *t = bam1_qual(b);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
247 int i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
248 const bam1_core_t *c = &b->core;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
249 kstring_t str;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
250 str.l = str.m = 0; str.s = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
251
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
252 ksprintf(&str, "%s\t", bam1_qname(b));
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
253 if (of == BAM_OFDEC) ksprintf(&str, "%d\t", c->flag);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
254 else if (of == BAM_OFHEX) ksprintf(&str, "0x%x\t", c->flag);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
255 else { // BAM_OFSTR
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
256 for (i = 0; i < 16; ++i)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
257 if ((c->flag & 1<<i) && bam_flag2char_table[i])
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
258 kputc(bam_flag2char_table[i], &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
259 kputc('\t', &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
260 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
261 if (c->tid < 0) kputs("*\t", &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
262 else ksprintf(&str, "%s\t", header->target_name[c->tid]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
263 ksprintf(&str, "%d\t%d\t", c->pos + 1, c->qual);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
264 if (c->n_cigar == 0) kputc('*', &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
265 else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
266 for (i = 0; i < c->n_cigar; ++i)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
267 ksprintf(&str, "%d%c", bam1_cigar(b)[i]>>BAM_CIGAR_SHIFT, "MIDNSHP"[bam1_cigar(b)[i]&BAM_CIGAR_MASK]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
268 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
269 kputc('\t', &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
270 if (c->mtid < 0) kputs("*\t", &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
271 else if (c->mtid == c->tid) kputs("=\t", &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
272 else ksprintf(&str, "%s\t", header->target_name[c->mtid]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
273 ksprintf(&str, "%d\t%d\t", c->mpos + 1, c->isize);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
274 if (c->l_qseq) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
275 for (i = 0; i < c->l_qseq; ++i) kputc(bam_nt16_rev_table[bam1_seqi(s, i)], &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
276 kputc('\t', &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
277 if (t[0] == 0xff) kputc('*', &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
278 else for (i = 0; i < c->l_qseq; ++i) kputc(t[i] + 33, &str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
279 } else ksprintf(&str, "*\t*");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
280 s = bam1_aux(b);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
281 while (s < b->data + b->data_len) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
282 uint8_t type, key[2];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
283 key[0] = s[0]; key[1] = s[1];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
284 s += 2; type = *s; ++s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
285 ksprintf(&str, "\t%c%c:", key[0], key[1]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
286 if (type == 'A') { ksprintf(&str, "A:%c", *s); ++s; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
287 else if (type == 'C') { ksprintf(&str, "i:%u", *s); ++s; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
288 else if (type == 'c') { ksprintf(&str, "i:%d", *s); ++s; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
289 else if (type == 'S') { ksprintf(&str, "i:%u", *(uint16_t*)s); s += 2; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
290 else if (type == 's') { ksprintf(&str, "i:%d", *(int16_t*)s); s += 2; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
291 else if (type == 'I') { ksprintf(&str, "i:%u", *(uint32_t*)s); s += 4; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
292 else if (type == 'i') { ksprintf(&str, "i:%d", *(int32_t*)s); s += 4; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
293 else if (type == 'f') { ksprintf(&str, "f:%g", *(float*)s); s += 4; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
294 else if (type == 'd') { ksprintf(&str, "d:%lg", *(double*)s); s += 8; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
295 else if (type == 'Z' || type == 'H') { ksprintf(&str, "%c:", type); while (*s) kputc(*s++, &str); ++s; }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
296 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
297 return str.s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
298 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
299
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
300 char *bam_format1(const bam_header_t *header, const bam1_t *b)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
301 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
302 return bam_format1_core(header, b, BAM_OFDEC);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
303 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
304
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
305 void bam_view1(const bam_header_t *header, const bam1_t *b)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
306 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
307 char *s = bam_format1(header, b);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
308 printf("%s\n", s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
309 free(s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
310 }