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