0
|
1 #ifndef BWA_H_
|
|
2 #define BWA_H_
|
|
3
|
|
4 #include <stdint.h>
|
|
5
|
|
6 #define BWA_DEF_MAX_SCORE 2048
|
|
7 #define BWA_MAX_QUERY_LEN 1024
|
|
8
|
|
9 // BWA index
|
|
10 struct bwa_idx_t;
|
|
11 typedef struct bwa_idx_t bwa_idx_t;
|
|
12
|
|
13 // Buffer for BWA alignment
|
|
14 struct bwa_buf_t;
|
|
15 typedef struct bwa_buf_t bwa_buf_t;
|
|
16
|
|
17 // BWA alignment options
|
|
18 typedef struct {
|
|
19 int s_gapo, s_gape; // gap open and extension penalties; the mismatch penalty is fixed at 3
|
|
20 int max_diff, max_gapo, max_gape; // max differences (-1 to use fnr for length-adjusted max diff), gap opens and gap extensions
|
|
21 int seed_len, max_seed_diff; // seed length and max differences allowed in the seed
|
|
22 float fnr; // parameter for automatic length-adjusted max differences
|
|
23 } bwa_opt_t;
|
|
24
|
|
25 // default BWA alignment options
|
|
26 extern bwa_opt_t bwa_def_opt; // = { 11, 4, -1, 1, 6, 32, 2, 0.04 }
|
|
27
|
|
28 // an interval hit in the SA coordinate; basic unit in .sai files
|
|
29 typedef struct {
|
|
30 uint32_t n_mm:16, n_gapo:8, n_gape:8;
|
|
31 int score;
|
|
32 uint64_t k, l; // [k,l] is the SA interval; each interval has l-k+1 hits
|
|
33 } bwa_sai1_t;
|
|
34
|
|
35 // all interval hits in the SA coordinate
|
|
36 typedef struct {
|
|
37 int n; // number of interval hits
|
|
38 bwa_sai1_t *sai;
|
|
39 } bwa_sai_t;
|
|
40
|
|
41 // an alignment
|
|
42 typedef struct {
|
|
43 uint32_t n_n:8, n_gap:12, n_mm:12; // number of ambiguous bases, gaps and mismatches in the alignment
|
|
44 int32_t ref_id; // referece sequence index (the first seq is indexed by 0)
|
|
45 uint32_t offset; // coordinate on the reference; zero-based
|
|
46 uint32_t n_cigar:16, flag:16; // number of CIGAR operations; SAM flag
|
|
47 uint32_t *cigar; // CIGAR in the BAM 28+4 encoding; having n_cigar operations
|
|
48 } bwa_aln_t;
|
|
49
|
|
50 typedef struct {
|
|
51 int mapQs, mapQ, c1, c2;
|
|
52 uint64_t sa;
|
|
53 bwa_sai1_t *which;
|
|
54 bwa_sai_t sai;
|
|
55 bwa_aln_t one;
|
|
56 } bwa_one_t;
|
|
57
|
|
58 typedef struct {
|
|
59 double avg, std, ap_prior;
|
|
60 uint64_t low, high, high_bayesian;
|
|
61 } bwa_pestat_t;
|
|
62
|
|
63 #ifdef __cplusplus
|
|
64 extern "C" {
|
|
65 #endif
|
|
66
|
|
67 // load a BWA index
|
|
68 bwa_idx_t *bwa_idx_load(const char *prefix);
|
|
69 void bwa_idx_destroy(bwa_idx_t *p);
|
|
70
|
|
71 // allocate a BWA alignment buffer; if unsure, set opt to &bwa_def_opt and max_score to BWA_DEF_MAX_SCORE
|
|
72 bwa_buf_t *bwa_buf_init(const bwa_opt_t *opt, int max_score);
|
|
73 void bwa_buf_destroy(bwa_buf_t *p);
|
|
74
|
|
75 /**
|
|
76 * Find all the SA intervals
|
|
77 *
|
|
78 * @param idx BWA index; multiple threads can share the same index
|
|
79 * @param buf BWA alignment buffer; each thread should have its own buffer
|
|
80 * @param seq NULL terminated C string, consisting of A/C/G/T/N only
|
|
81 *
|
|
82 * @return SA intervals seq is matched to
|
|
83 */
|
|
84 bwa_sai_t bwa_sai(const bwa_idx_t *idx, bwa_buf_t *buf, const char *seq);
|
|
85
|
|
86 /**
|
|
87 * Construct an alignment in the base-pair coordinate
|
|
88 *
|
|
89 * @param idx BWA index
|
|
90 * @param buf BWA alignment buffer
|
|
91 * @param seq NULL terinated C string
|
|
92 * @param sa Suffix array value
|
|
93 * @param n_gaps Number of gaps (typically equal to bwa_sai1_t::n_gapo + bwa_sai1_t::n_gape
|
|
94 *
|
|
95 * @return An alignment
|
|
96 */
|
|
97 void bwa_sa2aln(const bwa_idx_t *idx, bwa_buf_t *buf, const char *seq, uint64_t sa, int n_gaps, bwa_aln_t *aln);
|
|
98
|
|
99 bwa_one_t *bwa_se(const bwa_idx_t *idx, bwa_buf_t *buf, const char *seq, int gen_cigar);
|
|
100
|
|
101 void bwa_one_destroy(bwa_one_t *one);
|
|
102
|
|
103 #ifdef __cplusplus
|
|
104 }
|
|
105 #endif
|
|
106
|
|
107 #endif
|