annotate pyPRADA_1.2/tools/bwa-0.5.7-mh/bwt.h @ 3:f17965495ec9 draft default tip

Uploaded
author siyuan
date Tue, 11 Mar 2014 12:14:01 -0400
parents acc2ca1a3ba4
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
1 /* The MIT License
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
2
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
3 Copyright (c) 2008 Genome Research Ltd (GRL).
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
4
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
5 Permission is hereby granted, free of charge, to any person obtaining
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
6 a copy of this software and associated documentation files (the
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
7 "Software"), to deal in the Software without restriction, including
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
8 without limitation the rights to use, copy, modify, merge, publish,
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
9 distribute, sublicense, and/or sell copies of the Software, and to
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
10 permit persons to whom the Software is furnished to do so, subject to
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
11 the following conditions:
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
12
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
13 The above copyright notice and this permission notice shall be
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
14 included in all copies or substantial portions of the Software.
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
15
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
16 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
17 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
18 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
19 NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
20 BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
21 ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
22 CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
23 SOFTWARE.
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
24 */
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
25
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
26 /* Contact: Heng Li <lh3@sanger.ac.uk> */
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
27
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
28 #ifndef BWA_BWT_H
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
29 #define BWA_BWT_H
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
30
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
31 #include <stdint.h>
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
32
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
33 // requirement: (OCC_INTERVAL%16 == 0)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
34 #define OCC_INTERVAL 0x80
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
35
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
36 #ifndef BWA_UBYTE
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
37 #define BWA_UBYTE
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
38 typedef unsigned char ubyte_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
39 #endif
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
40 typedef uint32_t bwtint_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
41
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
42 typedef struct {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
43 bwtint_t primary; // S^{-1}(0), or the primary index of BWT
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
44 bwtint_t L2[5]; // C(), cumulative count
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
45 bwtint_t seq_len; // sequence length
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
46 bwtint_t bwt_size; // size of bwt, about seq_len/4
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
47 uint32_t *bwt; // BWT
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
48 // occurance array, separated to two parts
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
49 uint32_t cnt_table[256];
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
50 // suffix array
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
51 int sa_intv;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
52 bwtint_t n_sa;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
53 bwtint_t *sa;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
54 } bwt_t;
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
55
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
56 #define bwt_bwt(b, k) ((b)->bwt[(k)/OCC_INTERVAL*12 + 4 + (k)%OCC_INTERVAL/16])
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
57
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
58 /* retrieve a character from the $-removed BWT string. Note that
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
59 * bwt_t::bwt is not exactly the BWT string and therefore this macro is
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
60 * called bwt_B0 instead of bwt_B */
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
61 #define bwt_B0(b, k) (bwt_bwt(b, k)>>((~(k)&0xf)<<1)&3)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
62
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
63 #define bwt_occ_intv(b, k) ((b)->bwt + (k)/OCC_INTERVAL*12)
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
64
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
65 // inverse Psi function
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
66 #define bwt_invPsi(bwt, k) \
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
67 (((k) == (bwt)->primary)? 0 : \
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
68 ((k) < (bwt)->primary)? \
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
69 (bwt)->L2[bwt_B0(bwt, k)] + bwt_occ(bwt, k, bwt_B0(bwt, k)) \
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
70 : (bwt)->L2[bwt_B0(bwt, (k)-1)] + bwt_occ(bwt, k, bwt_B0(bwt, (k)-1)))
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
71
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
72 #ifdef __cplusplus
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
73 extern "C" {
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
74 #endif
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
75
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
76 void bwt_dump_bwt(const char *fn, const bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
77 void bwt_dump_sa(const char *fn, const bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
78
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
79 bwt_t *bwt_restore_bwt(const char *fn);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
80 void bwt_restore_sa(const char *fn, bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
81
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
82 void bwt_destroy(bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
83
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
84 void bwt_bwtgen(const char *fn_pac, const char *fn_bwt); // from BWT-SW
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
85 void bwt_cal_sa(bwt_t *bwt, int intv);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
86
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
87 void bwt_bwtupdate_core(bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
88
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
89 inline bwtint_t bwt_occ(const bwt_t *bwt, bwtint_t k, ubyte_t c);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
90 inline void bwt_occ4(const bwt_t *bwt, bwtint_t k, bwtint_t cnt[4]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
91 bwtint_t bwt_sa(const bwt_t *bwt, bwtint_t k);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
92
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
93 // more efficient version of bwt_occ/bwt_occ4 for retrieving two close Occ values
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
94 void bwt_gen_cnt_table(bwt_t *bwt);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
95 inline void bwt_2occ(const bwt_t *bwt, bwtint_t k, bwtint_t l, ubyte_t c, bwtint_t *ok, bwtint_t *ol);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
96 inline void bwt_2occ4(const bwt_t *bwt, bwtint_t k, bwtint_t l, bwtint_t cntk[4], bwtint_t cntl[4]);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
97
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
98 int bwt_match_exact(const bwt_t *bwt, int len, const ubyte_t *str, bwtint_t *sa_begin, bwtint_t *sa_end);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
99 int bwt_match_exact_alt(const bwt_t *bwt, int len, const ubyte_t *str, bwtint_t *k0, bwtint_t *l0);
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
100
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
101 #ifdef __cplusplus
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
102 }
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
103 #endif
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
104
acc2ca1a3ba4 Uploaded
siyuan
parents:
diff changeset
105 #endif