annotate SNV/SNVMix2_source/SNVMix2-v0.12.1-rc1/samtools-0.1.6/faidx.c @ 7:351b3acadd17 default tip

Uploaded
author ryanmorin
date Tue, 18 Oct 2011 18:33:15 -0400
parents 74f5ea818cea
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
1 #include <ctype.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
2 #include <string.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
3 #include <stdlib.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
4 #include <stdio.h>
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
5 #include "faidx.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
6 #include "khash.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
7
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
8 typedef struct {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
9 uint64_t len:32, line_len:16, line_blen:16;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
10 uint64_t offset;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
11 } faidx1_t;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
12 KHASH_MAP_INIT_STR(s, faidx1_t)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
13
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
14 #ifndef _NO_RAZF
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
15 #include "razf.h"
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
16 #else
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
17 #ifdef _WIN32
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
18 #define ftello(fp) ftell(fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
19 #define fseeko(fp, offset, whence) fseek(fp, offset, whence)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
20 #else
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
21 extern off_t ftello(FILE *stream);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
22 extern int fseeko(FILE *stream, off_t offset, int whence);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
23 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
24 #define RAZF FILE
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
25 #define razf_read(fp, buf, size) fread(buf, 1, size, fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
26 #define razf_open(fn, mode) fopen(fn, mode)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
27 #define razf_close(fp) fclose(fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
28 #define razf_seek(fp, offset, whence) fseeko(fp, offset, whence)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
29 #define razf_tell(fp) ftello(fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
30 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
31
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
32 struct __faidx_t {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
33 RAZF *rz;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
34 int n, m;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
35 char **name;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
36 khash_t(s) *hash;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
37 };
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
38
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
39 #ifndef kroundup32
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
40 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
41 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
42
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
43 static inline void fai_insert_index(faidx_t *idx, const char *name, int len, int line_len, int line_blen, uint64_t offset)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
44 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
45 khint_t k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
46 int ret;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
47 faidx1_t t;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
48 if (idx->n == idx->m) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
49 idx->m = idx->m? idx->m<<1 : 16;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
50 idx->name = (char**)realloc(idx->name, sizeof(void*) * idx->m);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
51 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
52 idx->name[idx->n] = strdup(name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
53 k = kh_put(s, idx->hash, idx->name[idx->n], &ret);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
54 t.len = len; t.line_len = line_len; t.line_blen = line_blen; t.offset = offset;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
55 kh_value(idx->hash, k) = t;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
56 ++idx->n;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
57 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
58
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
59 faidx_t *fai_build_core(RAZF *rz)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
60 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
61 char c, *name;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
62 int l_name, m_name, ret;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
63 int len, line_len, line_blen, state;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
64 int l1, l2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
65 faidx_t *idx;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
66 uint64_t offset;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
67
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
68 idx = (faidx_t*)calloc(1, sizeof(faidx_t));
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
69 idx->hash = kh_init(s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
70 name = 0; l_name = m_name = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
71 len = line_len = line_blen = -1; state = 0; l1 = l2 = -1; offset = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
72 while (razf_read(rz, &c, 1)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
73 if (c == '\n') { // an empty line
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
74 if (state == 1) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
75 offset = razf_tell(rz);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
76 continue;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
77 } else if ((state == 0 && len < 0) || state == 2) continue;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
78 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
79 if (c == '>') { // fasta header
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
80 if (len >= 0)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
81 fai_insert_index(idx, name, len, line_len, line_blen, offset);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
82 l_name = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
83 while ((ret = razf_read(rz, &c, 1)) != 0 && !isspace(c)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
84 if (m_name < l_name + 2) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
85 m_name = l_name + 2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
86 kroundup32(m_name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
87 name = (char*)realloc(name, m_name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
88 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
89 name[l_name++] = c;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
90 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
91 name[l_name] = '\0';
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
92 if (ret == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
93 fprintf(stderr, "[fai_build_core] the last entry has no sequence\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
94 free(name); fai_destroy(idx);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
95 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
96 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
97 if (c != '\n') while (razf_read(rz, &c, 1) && c != '\n');
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
98 state = 1; len = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
99 offset = razf_tell(rz);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
100 } else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
101 if (state == 3) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
102 fprintf(stderr, "[fai_build_core] inlined empty line is not allowed in sequence '%s'.\n", name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
103 free(name); fai_destroy(idx);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
104 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
105 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
106 if (state == 2) state = 3;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
107 l1 = l2 = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
108 do {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
109 ++l1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
110 if (isgraph(c)) ++l2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
111 } while ((ret = razf_read(rz, &c, 1)) && c != '\n');
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
112 if (state == 3 && l2) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
113 fprintf(stderr, "[fai_build_core] different line length in sequence '%s'.\n", name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
114 free(name); fai_destroy(idx);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
115 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
116 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
117 ++l1; len += l2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
118 if (l2 >= 0x10000) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
119 fprintf(stderr, "[fai_build_core] line length exceeds 65535 in sequence '%s'.\n", name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
120 free(name); fai_destroy(idx);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
121 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
122 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
123 if (state == 1) line_len = l1, line_blen = l2, state = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
124 else if (state == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
125 if (l1 != line_len || l2 != line_blen) state = 2;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
126 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
127 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
128 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
129 fai_insert_index(idx, name, len, line_len, line_blen, offset);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
130 free(name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
131 return idx;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
132 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
133
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
134 void fai_save(const faidx_t *fai, FILE *fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
135 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
136 khint_t k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
137 int i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
138 for (i = 0; i < fai->n; ++i) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
139 faidx1_t x;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
140 k = kh_get(s, fai->hash, fai->name[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
141 x = kh_value(fai->hash, k);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
142 #ifdef _WIN32
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
143 fprintf(fp, "%s\t%d\t%ld\t%d\t%d\n", fai->name[i], (int)x.len, (long)x.offset, (int)x.line_blen, (int)x.line_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
144 #else
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
145 fprintf(fp, "%s\t%d\t%lld\t%d\t%d\n", fai->name[i], (int)x.len, (long long)x.offset, (int)x.line_blen, (int)x.line_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
146 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
147 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
148 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
149
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
150 faidx_t *fai_read(FILE *fp)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
151 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
152 faidx_t *fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
153 char *buf, *p;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
154 int len, line_len, line_blen;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
155 #ifdef _WIN32
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
156 long offset;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
157 #else
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
158 long long offset;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
159 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
160 fai = (faidx_t*)calloc(1, sizeof(faidx_t));
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
161 fai->hash = kh_init(s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
162 buf = (char*)calloc(0x10000, 1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
163 while (!feof(fp) && fgets(buf, 0x10000, fp)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
164 for (p = buf; *p && isgraph(*p); ++p);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
165 *p = 0; ++p;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
166 #ifdef _WIN32
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
167 sscanf(p, "%d%ld%d%d", &len, &offset, &line_blen, &line_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
168 #else
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
169 sscanf(p, "%d%lld%d%d", &len, &offset, &line_blen, &line_len);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
170 #endif
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
171 fai_insert_index(fai, buf, len, line_len, line_blen, offset);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
172 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
173 free(buf);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
174 return fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
175 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
176
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
177 void fai_destroy(faidx_t *fai)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
178 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
179 int i;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
180 for (i = 0; i < fai->n; ++i) free(fai->name[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
181 free(fai->name);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
182 kh_destroy(s, fai->hash);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
183 if (fai->rz) razf_close(fai->rz);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
184 free(fai);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
185 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
186
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
187 int fai_build(const char *fn)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
188 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
189 char *str;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
190 RAZF *rz;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
191 FILE *fp;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
192 faidx_t *fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
193 str = (char*)calloc(strlen(fn) + 5, 1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
194 sprintf(str, "%s.fai", fn);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
195 rz = razf_open(fn, "r");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
196 if (rz == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
197 fprintf(stderr, "[fai_build] fail to open the FASTA file.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
198 free(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
199 return -1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
200 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
201 fai = fai_build_core(rz);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
202 razf_close(rz);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
203 fp = fopen(str, "wb");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
204 if (fp == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
205 fprintf(stderr, "[fai_build] fail to write FASTA index.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
206 fai_destroy(fai); free(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
207 return -1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
208 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
209 fai_save(fai, fp);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
210 fclose(fp);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
211 free(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
212 fai_destroy(fai);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
213 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
214 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
215
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
216 faidx_t *fai_load(const char *fn)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
217 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
218 char *str;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
219 FILE *fp;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
220 faidx_t *fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
221 str = (char*)calloc(strlen(fn) + 5, 1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
222 sprintf(str, "%s.fai", fn);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
223 fp = fopen(str, "rb");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
224 if (fp == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
225 fprintf(stderr, "[fai_load] build FASTA index.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
226 fai_build(fn);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
227 fp = fopen(str, "r");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
228 if (fp == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
229 fprintf(stderr, "[fai_load] fail to open FASTA index.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
230 free(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
231 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
232 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
233 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
234 fai = fai_read(fp);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
235 fclose(fp);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
236 fai->rz = razf_open(fn, "rb");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
237 free(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
238 if (fai->rz == 0) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
239 fprintf(stderr, "[fai_load] fail to open FASTA file.\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
240 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
241 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
242 return fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
243 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
244
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
245 char *fai_fetch(const faidx_t *fai, const char *str, int *len)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
246 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
247 char *s, *p, c;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
248 int i, l, k;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
249 khiter_t iter;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
250 faidx1_t val;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
251 khash_t(s) *h;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
252 int beg, end;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
253
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
254 beg = end = -1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
255 h = fai->hash;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
256 l = strlen(str);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
257 p = s = (char*)malloc(l+1);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
258 /* squeeze out "," */
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
259 for (i = k = 0; i != l; ++i)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
260 if (str[i] != ',' && !isspace(str[i])) s[k++] = str[i];
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
261 s[k] = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
262 for (i = 0; i != k; ++i) if (s[i] == ':') break;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
263 s[i] = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
264 iter = kh_get(s, h, s); /* get the ref_id */
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
265 if (iter == kh_end(h)) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
266 *len = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
267 free(s); return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
268 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
269 val = kh_value(h, iter);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
270 if (i == k) { /* dump the whole sequence */
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
271 beg = 0; end = val.len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
272 } else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
273 for (p = s + i + 1; i != k; ++i) if (s[i] == '-') break;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
274 beg = atoi(p);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
275 if (i < k) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
276 p = s + i + 1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
277 end = atoi(p);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
278 } else end = val.len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
279 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
280 if (beg > 0) --beg;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
281 if (beg >= val.len) beg = val.len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
282 if (end >= val.len) end = val.len;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
283 if (beg > end) beg = end;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
284 free(s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
285
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
286 // now retrieve the sequence
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
287 l = 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
288 s = (char*)malloc(end - beg + 2);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
289 razf_seek(fai->rz, val.offset + beg / val.line_blen * val.line_len + beg % val.line_blen, SEEK_SET);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
290 while (razf_read(fai->rz, &c, 1) == 1 && l < end - beg)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
291 if (isgraph(c)) s[l++] = c;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
292 s[l] = '\0';
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
293 *len = l;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
294 return s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
295 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
296
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
297 int faidx_main(int argc, char *argv[])
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
298 {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
299 if (argc == 1) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
300 fprintf(stderr, "Usage: faidx <in.fasta> [<reg> [...]]\n");
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
301 return 1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
302 } else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
303 if (argc == 2) fai_build(argv[1]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
304 else {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
305 int i, j, k, l;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
306 char *s;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
307 faidx_t *fai;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
308 fai = fai_load(argv[1]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
309 if (fai == 0) return 1;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
310 for (i = 2; i != argc; ++i) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
311 printf(">%s\n", argv[i]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
312 s = fai_fetch(fai, argv[i], &l);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
313 for (j = 0; j < l; j += 60) {
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
314 for (k = 0; k < 60 && k < l - j; ++k)
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
315 putchar(s[j + k]);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
316 putchar('\n');
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
317 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
318 free(s);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
319 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
320 fai_destroy(fai);
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
321 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
322 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
323 return 0;
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
324 }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
325
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
326 #ifdef FAIDX_MAIN
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
327 int main(int argc, char *argv[]) { return faidx_main(argc, argv); }
74f5ea818cea Uploaded
ryanmorin
parents:
diff changeset
328 #endif