annotate bwa-0.6.2/bamlite.c @ 2:a294fbfcb1db draft default tip

Uploaded BWA
author ashvark
date Fri, 18 Jul 2014 07:55:59 -0400
parents dd1186b11b3b
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
1 #include <stdlib.h>
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
2 #include <ctype.h>
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
3 #include <string.h>
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
4 #include <stdio.h>
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
5 #include "bamlite.h"
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
6
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
7 /*********************
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
8 * from bam_endian.c *
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
9 *********************/
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
10
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
11 static inline int bam_is_big_endian()
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
12 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
13 long one= 1;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
14 return !(*((char *)(&one)));
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
15 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
16 static inline uint16_t bam_swap_endian_2(uint16_t v)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
17 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
18 return (uint16_t)(((v & 0x00FF00FFU) << 8) | ((v & 0xFF00FF00U) >> 8));
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
19 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
20 static inline void *bam_swap_endian_2p(void *x)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
21 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
22 *(uint16_t*)x = bam_swap_endian_2(*(uint16_t*)x);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
23 return x;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
24 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
25 static inline uint32_t bam_swap_endian_4(uint32_t v)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
26 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
27 v = ((v & 0x0000FFFFU) << 16) | (v >> 16);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
28 return ((v & 0x00FF00FFU) << 8) | ((v & 0xFF00FF00U) >> 8);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
29 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
30 static inline void *bam_swap_endian_4p(void *x)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
31 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
32 *(uint32_t*)x = bam_swap_endian_4(*(uint32_t*)x);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
33 return x;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
34 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
35 static inline uint64_t bam_swap_endian_8(uint64_t v)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
36 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
37 v = ((v & 0x00000000FFFFFFFFLLU) << 32) | (v >> 32);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
38 v = ((v & 0x0000FFFF0000FFFFLLU) << 16) | ((v & 0xFFFF0000FFFF0000LLU) >> 16);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
39 return ((v & 0x00FF00FF00FF00FFLLU) << 8) | ((v & 0xFF00FF00FF00FF00LLU) >> 8);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
40 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
41 static inline void *bam_swap_endian_8p(void *x)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
42 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
43 *(uint64_t*)x = bam_swap_endian_8(*(uint64_t*)x);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
44 return x;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
45 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
46
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
47 /**************
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
48 * from bam.c *
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
49 **************/
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
50
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
51 int bam_is_be;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
52
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
53 bam_header_t *bam_header_init()
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
54 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
55 bam_is_be = bam_is_big_endian();
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
56 return (bam_header_t*)calloc(1, sizeof(bam_header_t));
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
57 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
58
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
59 void bam_header_destroy(bam_header_t *header)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
60 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
61 int32_t i;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
62 if (header == 0) return;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
63 if (header->target_name) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
64 for (i = 0; i < header->n_targets; ++i)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
65 free(header->target_name[i]);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
66 free(header->target_name);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
67 free(header->target_len);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
68 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
69 free(header->text);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
70 free(header);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
71 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
72
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
73 bam_header_t *bam_header_read(bamFile fp)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
74 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
75 bam_header_t *header;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
76 char buf[4];
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
77 int magic_len;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
78 int32_t i = 1, name_len;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
79 // read "BAM1"
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
80 magic_len = bam_read(fp, buf, 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
81 if (magic_len != 4 || strncmp(buf, "BAM\001", 4) != 0) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
82 fprintf(stderr, "[bam_header_read] invalid BAM binary header (this is not a BAM file).\n");
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
83 return 0;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
84 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
85 header = bam_header_init();
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
86 // read plain text and the number of reference sequences
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
87 bam_read(fp, &header->l_text, 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
88 if (bam_is_be) bam_swap_endian_4p(&header->l_text);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
89 header->text = (char*)calloc(header->l_text + 1, 1);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
90 bam_read(fp, header->text, header->l_text);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
91 bam_read(fp, &header->n_targets, 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
92 if (bam_is_be) bam_swap_endian_4p(&header->n_targets);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
93 // read reference sequence names and lengths
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
94 header->target_name = (char**)calloc(header->n_targets, sizeof(char*));
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
95 header->target_len = (uint32_t*)calloc(header->n_targets, 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
96 for (i = 0; i != header->n_targets; ++i) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
97 bam_read(fp, &name_len, 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
98 if (bam_is_be) bam_swap_endian_4p(&name_len);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
99 header->target_name[i] = (char*)calloc(name_len, 1);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
100 bam_read(fp, header->target_name[i], name_len);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
101 bam_read(fp, &header->target_len[i], 4);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
102 if (bam_is_be) bam_swap_endian_4p(&header->target_len[i]);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
103 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
104 return header;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
105 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
106
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
107 static void swap_endian_data(const bam1_core_t *c, int data_len, uint8_t *data)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
108 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
109 uint8_t *s;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
110 uint32_t i, *cigar = (uint32_t*)(data + c->l_qname);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
111 s = data + c->n_cigar*4 + c->l_qname + c->l_qseq + (c->l_qseq + 1)/2;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
112 for (i = 0; i < c->n_cigar; ++i) bam_swap_endian_4p(&cigar[i]);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
113 while (s < data + data_len) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
114 uint8_t type;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
115 s += 2; // skip key
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
116 type = toupper(*s); ++s; // skip type
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
117 if (type == 'C' || type == 'A') ++s;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
118 else if (type == 'S') { bam_swap_endian_2p(s); s += 2; }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
119 else if (type == 'I' || type == 'F') { bam_swap_endian_4p(s); s += 4; }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
120 else if (type == 'D') { bam_swap_endian_8p(s); s += 8; }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
121 else if (type == 'Z' || type == 'H') { while (*s) ++s; ++s; }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
122 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
123 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
124
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
125 int bam_read1(bamFile fp, bam1_t *b)
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
126 {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
127 bam1_core_t *c = &b->core;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
128 int32_t block_len, ret, i;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
129 uint32_t x[8];
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
130
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
131 if ((ret = bam_read(fp, &block_len, 4)) != 4) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
132 if (ret == 0) return -1; // normal end-of-file
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
133 else return -2; // truncated
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
134 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
135 if (bam_read(fp, x, sizeof(bam1_core_t)) != sizeof(bam1_core_t)) return -3;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
136 if (bam_is_be) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
137 bam_swap_endian_4p(&block_len);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
138 for (i = 0; i < 8; ++i) bam_swap_endian_4p(x + i);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
139 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
140 c->tid = x[0]; c->pos = x[1];
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
141 c->bin = x[2]>>16; c->qual = x[2]>>8&0xff; c->l_qname = x[2]&0xff;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
142 c->flag = x[3]>>16; c->n_cigar = x[3]&0xffff;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
143 c->l_qseq = x[4];
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
144 c->mtid = x[5]; c->mpos = x[6]; c->isize = x[7];
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
145 b->data_len = block_len - sizeof(bam1_core_t);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
146 if (b->m_data < b->data_len) {
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
147 b->m_data = b->data_len;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
148 kroundup32(b->m_data);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
149 b->data = (uint8_t*)realloc(b->data, b->m_data);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
150 }
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
151 if (bam_read(fp, b->data, b->data_len) != b->data_len) return -4;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
152 b->l_aux = b->data_len - c->n_cigar * 4 - c->l_qname - c->l_qseq - (c->l_qseq+1)/2;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
153 if (bam_is_be) swap_endian_data(c, b->data_len, b->data);
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
154 return 4 + block_len;
dd1186b11b3b Uploaded BWA
ashvark
parents:
diff changeset
155 }