bio-bigwig 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/LICENSE.txt +21 -0
- data/README.md +131 -0
- data/ext/bio/bigwig/bigwigext.c +747 -0
- data/ext/bio/bigwig/bigwigext.h +9 -0
- data/ext/bio/bigwig/extconf.rb +30 -0
- data/ext/bio/bigwig/libBigWig/LICENSE +22 -0
- data/ext/bio/bigwig/libBigWig/bigWig.h +606 -0
- data/ext/bio/bigwig/libBigWig/bigWigIO.h +110 -0
- data/ext/bio/bigwig/libBigWig/bwCommon.h +70 -0
- data/ext/bio/bigwig/libBigWig/bwRead.c +427 -0
- data/ext/bio/bigwig/libBigWig/bwStats.c +537 -0
- data/ext/bio/bigwig/libBigWig/bwValues.c +803 -0
- data/ext/bio/bigwig/libBigWig/bwValues.h +77 -0
- data/ext/bio/bigwig/libBigWig/bwWrite.c +1333 -0
- data/ext/bio/bigwig/libBigWig/io.c +296 -0
- data/ext/bio/bigwig/libBigWig/test/exampleWrite.c +76 -0
- data/ext/bio/bigwig/libBigWig/test/testBigBed.c +132 -0
- data/ext/bio/bigwig/libBigWig/test/testIterator.c +67 -0
- data/ext/bio/bigwig/libBigWig/test/testLocal.c +223 -0
- data/ext/bio/bigwig/libBigWig/test/testRemote.c +203 -0
- data/ext/bio/bigwig/libBigWig/test/testRemoteManyContigs.c +46 -0
- data/ext/bio/bigwig/libBigWig/test/testWrite.c +68 -0
- data/lib/bio/bigwig/version.rb +7 -0
- data/lib/bio/bigwig.rb +47 -0
- metadata +69 -0
@@ -0,0 +1,747 @@
|
|
1
|
+
#include "bigwigext.h"
|
2
|
+
|
3
|
+
#define SIZEOF_INT32 4
|
4
|
+
#define SIZEOF_INT64 8
|
5
|
+
|
6
|
+
#if SIZEOF_SHORT == SIZEOF_INT32
|
7
|
+
#define NUM2UINT32 NUM2USHORT
|
8
|
+
#define NUM2INT32 NUM2SHORT
|
9
|
+
#define UINT32_2NUM USHORT2NUM
|
10
|
+
#define INT32_2NUM SHORT2NUM
|
11
|
+
#elif SIZEOF_INT == SIZEOF_INT32
|
12
|
+
#define NUM2UINT32 NUM2UINT
|
13
|
+
#define NUM2INT32 NUM2INT
|
14
|
+
#define UINT32_2NUM UINT2NUM
|
15
|
+
#define INT32_2NUM INT2NUM
|
16
|
+
#elif SIZEOF_LONG == SIZEOF_INT32
|
17
|
+
#define NUM2UINT32 NUM2ULONG
|
18
|
+
#define NUM2INT32 NUM2LONG
|
19
|
+
#define UINT32_2NUM ULONG2NUM
|
20
|
+
#define INT32_2NUM LONG2NUM
|
21
|
+
#else
|
22
|
+
#error "Neither int, long, nor short is the same size as int32_t"
|
23
|
+
#endif
|
24
|
+
|
25
|
+
#if SIZEOF_INT == SIZEOF_INT64
|
26
|
+
#define NUM2UINT64 NUM2UINT
|
27
|
+
#define NUM2INT64 NUM2INT
|
28
|
+
#define UINT64_2NUM UINT2NUM
|
29
|
+
#define INT64_2NUM INT2NUM
|
30
|
+
#elif SIZEOF_LONG == SIZEOF_INT64
|
31
|
+
#define NUM2UINT64 NUM2ULONG
|
32
|
+
#define NUM2INT64 NUM2LONG
|
33
|
+
#define UINT64_2NUM ULONG2NUM
|
34
|
+
#define INT64_2NUM LONG2NUM
|
35
|
+
#elif SIZEOF_LONGLONG == SIZEOF_INT64
|
36
|
+
#define NUM2UINT64 NUM2ULL
|
37
|
+
#define NUM2INT64 NUM2LL
|
38
|
+
#define UINT64_2NUM ULL2NUM
|
39
|
+
#define INT64_2NUM LL2NUM
|
40
|
+
#else
|
41
|
+
#error "Neither int, long, nor short is the same size as int64_t"
|
42
|
+
#endif
|
43
|
+
|
44
|
+
VALUE rb_Bio;
|
45
|
+
VALUE rb_BigWig;
|
46
|
+
|
47
|
+
static void BigWig_free(void *ptr);
|
48
|
+
static size_t BigWig_memsize(const void *ptr);
|
49
|
+
|
50
|
+
static const rb_data_type_t BigWig_type = {
|
51
|
+
"BigWig",
|
52
|
+
{
|
53
|
+
0,
|
54
|
+
BigWig_free,
|
55
|
+
BigWig_memsize,
|
56
|
+
},
|
57
|
+
0,
|
58
|
+
0,
|
59
|
+
RUBY_TYPED_FREE_IMMEDIATELY};
|
60
|
+
|
61
|
+
static void
|
62
|
+
BigWig_free(void *ptr)
|
63
|
+
{
|
64
|
+
if (ptr)
|
65
|
+
{
|
66
|
+
bwClose((bigWigFile_t *)ptr);
|
67
|
+
}
|
68
|
+
}
|
69
|
+
|
70
|
+
static size_t BigWig_memsize(const void *ptr)
|
71
|
+
{
|
72
|
+
const bigWigFile_t *data = ptr;
|
73
|
+
|
74
|
+
return data ? sizeof(*data) : 0;
|
75
|
+
}
|
76
|
+
|
77
|
+
static bigWigFile_t *get_bigWigFile(VALUE obj)
|
78
|
+
{
|
79
|
+
bigWigFile_t *data;
|
80
|
+
TypedData_Get_Struct(obj, bigWigFile_t, &BigWig_type, data);
|
81
|
+
return data;
|
82
|
+
}
|
83
|
+
|
84
|
+
static VALUE
|
85
|
+
bigwig_allocate(VALUE klass)
|
86
|
+
{
|
87
|
+
bigWigFile_t *bw = NULL;
|
88
|
+
|
89
|
+
return TypedData_Wrap_Struct(klass, &BigWig_type, bw);
|
90
|
+
}
|
91
|
+
|
92
|
+
//Return 1 if there are any entries at all
|
93
|
+
int hasEntries(bigWigFile_t *bw)
|
94
|
+
{
|
95
|
+
if (bw->hdr->indexOffset != 0)
|
96
|
+
return 1; // No index, no entries pyBigWig issue #111
|
97
|
+
//if(bw->hdr->nBasesCovered > 0) return 1; // Sometimes headers are broken
|
98
|
+
return 0;
|
99
|
+
}
|
100
|
+
|
101
|
+
static VALUE
|
102
|
+
bigwig_init(VALUE self, VALUE rb_fname, VALUE rb_mode)
|
103
|
+
{
|
104
|
+
char *fname = NULL;
|
105
|
+
char *mode = "r";
|
106
|
+
bigWigFile_t *bw = NULL;
|
107
|
+
|
108
|
+
fname = StringValueCStr(rb_fname);
|
109
|
+
mode = StringValueCStr(rb_mode);
|
110
|
+
|
111
|
+
//Open the local/remote file
|
112
|
+
if (strchr(mode, 'w') != NULL || bwIsBigWig(fname, NULL))
|
113
|
+
{
|
114
|
+
bw = bwOpen(fname, NULL, mode);
|
115
|
+
}
|
116
|
+
else
|
117
|
+
{
|
118
|
+
bw = bbOpen(fname, NULL);
|
119
|
+
}
|
120
|
+
|
121
|
+
if (!bw)
|
122
|
+
{
|
123
|
+
fprintf(stderr, "bw is NULL!\n");
|
124
|
+
goto error;
|
125
|
+
}
|
126
|
+
|
127
|
+
if (!mode || !strchr(mode, 'w'))
|
128
|
+
{
|
129
|
+
if (!bw->cl)
|
130
|
+
goto error;
|
131
|
+
}
|
132
|
+
|
133
|
+
//Set the data pointer
|
134
|
+
DATA_PTR(self) = bw;
|
135
|
+
|
136
|
+
rb_ivar_set(self, rb_intern("@last_tid"), INT2NUM(-1));
|
137
|
+
rb_ivar_set(self, rb_intern("@last_type"), INT2NUM(-1));
|
138
|
+
rb_ivar_set(self, rb_intern("@last_span"), INT2NUM(-1));
|
139
|
+
rb_ivar_set(self, rb_intern("@last_step"), INT2NUM(-1));
|
140
|
+
rb_ivar_set(self, rb_intern("@last_start"), INT2NUM(-1));
|
141
|
+
|
142
|
+
return self;
|
143
|
+
|
144
|
+
error:
|
145
|
+
if (bw)
|
146
|
+
bwClose(bw);
|
147
|
+
rb_raise(rb_eRuntimeError, "Received an error during file opening!");
|
148
|
+
return Qnil;
|
149
|
+
}
|
150
|
+
|
151
|
+
static VALUE
|
152
|
+
bigwig_close(VALUE self)
|
153
|
+
{
|
154
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
155
|
+
|
156
|
+
if (bw)
|
157
|
+
{
|
158
|
+
bwClose(bw);
|
159
|
+
DATA_PTR(self) = NULL;
|
160
|
+
}
|
161
|
+
|
162
|
+
return Qnil;
|
163
|
+
}
|
164
|
+
|
165
|
+
static VALUE
|
166
|
+
bw_get_header(VALUE self)
|
167
|
+
{
|
168
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
169
|
+
VALUE rb_header;
|
170
|
+
|
171
|
+
if (!bw)
|
172
|
+
{
|
173
|
+
rb_raise(rb_eRuntimeError, "The bigWig file handle is not opened!");
|
174
|
+
return Qnil;
|
175
|
+
}
|
176
|
+
|
177
|
+
if (bw->isWrite == 1)
|
178
|
+
{
|
179
|
+
rb_raise(rb_eRuntimeError, "The header cannot be accessed in files opened for writing!");
|
180
|
+
return Qnil;
|
181
|
+
}
|
182
|
+
|
183
|
+
rb_header = rb_hash_new();
|
184
|
+
|
185
|
+
// FIXME return int or double?
|
186
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("version")), ULONG2NUM(bw->hdr->version));
|
187
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("levels")), ULONG2NUM(bw->hdr->nLevels));
|
188
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("bases_covered")), ULL2NUM(bw->hdr->nBasesCovered));
|
189
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("min_val")), INT2NUM((int)bw->hdr->minVal));
|
190
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("max_val")), INT2NUM((int)bw->hdr->maxVal));
|
191
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("sum_data")), INT2NUM((int)bw->hdr->sumData));
|
192
|
+
rb_hash_aset(rb_header, ID2SYM(rb_intern("sum_squared")), INT2NUM((int)bw->hdr->sumSquared));
|
193
|
+
|
194
|
+
return rb_header;
|
195
|
+
|
196
|
+
error:
|
197
|
+
rb_raise(rb_eRuntimeError, "Received an error while getting the bigWig header!");
|
198
|
+
return Qnil;
|
199
|
+
}
|
200
|
+
|
201
|
+
static VALUE
|
202
|
+
bw_get_chroms(int argc, VALUE *argv, VALUE self)
|
203
|
+
{
|
204
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
205
|
+
VALUE rb_chrom, val, ret;
|
206
|
+
char *chrom = NULL;
|
207
|
+
uint32_t i;
|
208
|
+
|
209
|
+
ret = Qnil; // return nil if no chrom is found
|
210
|
+
|
211
|
+
if (!bw)
|
212
|
+
{
|
213
|
+
rb_raise(rb_eRuntimeError, "The bigWig file handle is not opened!");
|
214
|
+
return Qnil;
|
215
|
+
}
|
216
|
+
|
217
|
+
if (bw->isWrite == 1)
|
218
|
+
{
|
219
|
+
rb_raise(rb_eRuntimeError, "Chromosomes cannot be accessed in files opened for writing!");
|
220
|
+
return Qnil;
|
221
|
+
}
|
222
|
+
|
223
|
+
rb_scan_args(argc, argv, "01", &rb_chrom);
|
224
|
+
|
225
|
+
if (argc == 0)
|
226
|
+
{
|
227
|
+
ret = rb_hash_new();
|
228
|
+
for (i = 0; i < bw->cl->nKeys; i++)
|
229
|
+
{
|
230
|
+
val = ULONG2NUM(bw->cl->len[i]);
|
231
|
+
rb_hash_aset(ret, rb_str_new2(bw->cl->chrom[i]), val);
|
232
|
+
}
|
233
|
+
}
|
234
|
+
else
|
235
|
+
{
|
236
|
+
chrom = StringValueCStr(rb_chrom);
|
237
|
+
for (i = 0; i < bw->cl->nKeys; i++)
|
238
|
+
{
|
239
|
+
if (strcmp(bw->cl->chrom[i], chrom) == 0)
|
240
|
+
{
|
241
|
+
ret = ULONG2NUM(bw->cl->len[i]);
|
242
|
+
break;
|
243
|
+
}
|
244
|
+
}
|
245
|
+
}
|
246
|
+
|
247
|
+
return ret;
|
248
|
+
}
|
249
|
+
|
250
|
+
enum bwStatsType char2enum(char *s)
|
251
|
+
{
|
252
|
+
if (strcmp(s, "mean") == 0)
|
253
|
+
return mean;
|
254
|
+
if (strcmp(s, "std") == 0)
|
255
|
+
return stdev;
|
256
|
+
if (strcmp(s, "dev") == 0)
|
257
|
+
return dev;
|
258
|
+
if (strcmp(s, "max") == 0)
|
259
|
+
return max;
|
260
|
+
if (strcmp(s, "min") == 0)
|
261
|
+
return min;
|
262
|
+
if (strcmp(s, "cov") == 0)
|
263
|
+
return cov;
|
264
|
+
if (strcmp(s, "coverage") == 0)
|
265
|
+
return cov;
|
266
|
+
if (strcmp(s, "sum") == 0)
|
267
|
+
return sum;
|
268
|
+
return -1;
|
269
|
+
};
|
270
|
+
|
271
|
+
// double *bwStats (bigWigFile_t *fp, char *chrom, uint32_t start, uint32_t end, uint32_t nBins, enum bwStatsType type);
|
272
|
+
// double *bwStatsFromFull(bigWigFile_t *fp, char *chrom, uint32_t start, uint32_t end, uint32_t nBins, enum bwStatsType type);
|
273
|
+
|
274
|
+
static VALUE
|
275
|
+
bw_get_stats(VALUE self, VALUE rb_chrom, VALUE rb_start, VALUE rb_end, VALUE rb_nBins, VALUE rb_type, VALUE rb_exact)
|
276
|
+
{
|
277
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
278
|
+
double *val;
|
279
|
+
unsigned long startl = 0, endl = -1;
|
280
|
+
uint32_t start, end = -1, tid;
|
281
|
+
int nBins = 1, i;
|
282
|
+
char *chrom = NULL, *type = "mean";
|
283
|
+
VALUE ret;
|
284
|
+
|
285
|
+
if (!bw)
|
286
|
+
{
|
287
|
+
rb_raise(rb_eRuntimeError, "The bigWig file handle is not opened!");
|
288
|
+
return Qnil;
|
289
|
+
}
|
290
|
+
|
291
|
+
if (bw->isWrite == 1)
|
292
|
+
{
|
293
|
+
rb_raise(rb_eRuntimeError, "Statistics cannot be accessed in files opened for writing!");
|
294
|
+
return Qnil;
|
295
|
+
}
|
296
|
+
|
297
|
+
if (bw->type == 1)
|
298
|
+
{
|
299
|
+
rb_raise(rb_eRuntimeError, "bigBed files have no statistics!");
|
300
|
+
return Qnil;
|
301
|
+
}
|
302
|
+
|
303
|
+
if (rb_chrom != Qnil)
|
304
|
+
{
|
305
|
+
chrom = StringValueCStr(rb_chrom);
|
306
|
+
}
|
307
|
+
|
308
|
+
if (rb_start != Qnil)
|
309
|
+
startl = NUM2LONG(rb_start);
|
310
|
+
|
311
|
+
if (rb_end != Qnil)
|
312
|
+
endl = NUM2LONG(rb_end);
|
313
|
+
|
314
|
+
if (rb_nBins != Qnil)
|
315
|
+
nBins = NUM2INT(rb_nBins);
|
316
|
+
|
317
|
+
if (rb_type != Qnil)
|
318
|
+
type = StringValueCStr(rb_type);
|
319
|
+
|
320
|
+
if (rb_exact != Qnil)
|
321
|
+
{
|
322
|
+
if (RTEST(rb_exact) == 1)
|
323
|
+
endl = startl + nBins - 1;
|
324
|
+
}
|
325
|
+
|
326
|
+
tid = bwGetTid(bw, chrom);
|
327
|
+
|
328
|
+
if (endl == (unsigned long)-1 && tid != (uint32_t)-1)
|
329
|
+
endl = bw->cl->len[tid];
|
330
|
+
|
331
|
+
if (tid == (uint32_t)-1 || startl > end || endl > end)
|
332
|
+
{
|
333
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
334
|
+
return Qnil;
|
335
|
+
}
|
336
|
+
|
337
|
+
start = (uint32_t)startl;
|
338
|
+
end = (uint32_t)endl;
|
339
|
+
|
340
|
+
if (end <= start || end > bw->cl->len[tid] || start >= end)
|
341
|
+
{
|
342
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
343
|
+
return Qnil;
|
344
|
+
}
|
345
|
+
|
346
|
+
if (char2enum(type) == doesNotExist)
|
347
|
+
{
|
348
|
+
rb_raise(rb_eRuntimeError, "Invalid type! : %s", type);
|
349
|
+
return Qnil;
|
350
|
+
}
|
351
|
+
|
352
|
+
if (!hasEntries(bw))
|
353
|
+
{
|
354
|
+
ret = rb_ary_new2(nBins);
|
355
|
+
for (i = 0; i < nBins; i++)
|
356
|
+
{
|
357
|
+
rb_ary_store(ret, i, Qnil);
|
358
|
+
}
|
359
|
+
return ret;
|
360
|
+
}
|
361
|
+
|
362
|
+
if (RTEST(rb_exact))
|
363
|
+
{
|
364
|
+
val = bwStatsFromFull(bw, chrom, start, end, nBins, char2enum(type));
|
365
|
+
}
|
366
|
+
else
|
367
|
+
{
|
368
|
+
val = bwStats(bw, chrom, start, end, nBins, char2enum(type));
|
369
|
+
}
|
370
|
+
|
371
|
+
if (!val)
|
372
|
+
{
|
373
|
+
rb_raise(rb_eRuntimeError, "Error getting statistics!An error was encountered while fetching statistics.");
|
374
|
+
return Qnil;
|
375
|
+
}
|
376
|
+
|
377
|
+
ret = rb_ary_new2(nBins);
|
378
|
+
for (i = 0; i < nBins; i++)
|
379
|
+
{
|
380
|
+
rb_ary_store(ret, i, rb_float_new(val[i]));
|
381
|
+
}
|
382
|
+
free(val);
|
383
|
+
|
384
|
+
return ret;
|
385
|
+
}
|
386
|
+
|
387
|
+
static VALUE
|
388
|
+
bw_get_values(VALUE self, VALUE rb_chrom, VALUE rb_start, VALUE rb_end)
|
389
|
+
{
|
390
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
391
|
+
int i;
|
392
|
+
uint32_t start, end = -1, tid;
|
393
|
+
unsigned long startl = 0, endl = -1;
|
394
|
+
char *chrom = NULL;
|
395
|
+
VALUE ret;
|
396
|
+
bwOverlappingIntervals_t *o;
|
397
|
+
|
398
|
+
if (!bw)
|
399
|
+
{
|
400
|
+
rb_raise(rb_eRuntimeError, "The bigWig file handle is not opened!");
|
401
|
+
return Qnil;
|
402
|
+
}
|
403
|
+
|
404
|
+
if (bw->type == 1)
|
405
|
+
{
|
406
|
+
rb_raise(rb_eRuntimeError, "bigBed files have no values! Use 'entries' instead.");
|
407
|
+
return Qnil;
|
408
|
+
}
|
409
|
+
|
410
|
+
if (rb_chrom != Qnil)
|
411
|
+
{
|
412
|
+
chrom = StringValueCStr(rb_chrom);
|
413
|
+
}
|
414
|
+
|
415
|
+
if (rb_start != Qnil)
|
416
|
+
startl = NUM2LONG(rb_start);
|
417
|
+
|
418
|
+
if (rb_end != Qnil)
|
419
|
+
endl = NUM2LONG(rb_end);
|
420
|
+
|
421
|
+
tid = bwGetTid(bw, chrom);
|
422
|
+
|
423
|
+
if (endl == (unsigned long)-1 && tid != (uint32_t)-1)
|
424
|
+
endl = bw->cl->len[tid];
|
425
|
+
if (tid == (uint32_t)-1 || startl > end || endl > end)
|
426
|
+
{
|
427
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
428
|
+
return Qnil;
|
429
|
+
}
|
430
|
+
|
431
|
+
start = (uint32_t)startl;
|
432
|
+
end = (uint32_t)endl;
|
433
|
+
|
434
|
+
if (end <= start || end > bw->cl->len[tid] || start >= end)
|
435
|
+
{
|
436
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
437
|
+
return Qnil;
|
438
|
+
}
|
439
|
+
|
440
|
+
if (!hasEntries(bw))
|
441
|
+
{
|
442
|
+
return rb_ary_new2(0);
|
443
|
+
}
|
444
|
+
|
445
|
+
o = bwGetValues(bw, chrom, start, end, 1);
|
446
|
+
if (!o)
|
447
|
+
{
|
448
|
+
rb_raise(rb_eRuntimeError, "An error occurred while fetching values!");
|
449
|
+
return Qnil;
|
450
|
+
}
|
451
|
+
|
452
|
+
ret = rb_ary_new2(end - start);
|
453
|
+
for (i = 0; i < (int)o->l; i++)
|
454
|
+
rb_ary_store(ret, i, DBL2NUM(o->value[i]));
|
455
|
+
bwDestroyOverlappingIntervals(o);
|
456
|
+
|
457
|
+
return ret;
|
458
|
+
}
|
459
|
+
|
460
|
+
static VALUE
|
461
|
+
bw_get_intervals(VALUE self, VALUE rb_chrom, VALUE rb_start, VALUE rb_end)
|
462
|
+
{
|
463
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
464
|
+
uint32_t start, end = -1, tid, i;
|
465
|
+
unsigned long startl = 0, endl = -1;
|
466
|
+
bwOverlappingIntervals_t *intervals = NULL;
|
467
|
+
char *chrom;
|
468
|
+
VALUE ret;
|
469
|
+
|
470
|
+
if (!bw)
|
471
|
+
{
|
472
|
+
rb_raise(rb_eRuntimeError, "The bigWig file handle is not opened!");
|
473
|
+
return Qnil;
|
474
|
+
}
|
475
|
+
|
476
|
+
if (bw->isWrite == 1)
|
477
|
+
{
|
478
|
+
rb_raise(rb_eRuntimeError, "Intervals cannot be accessed in files opened for writing!");
|
479
|
+
return Qnil;
|
480
|
+
}
|
481
|
+
|
482
|
+
if (bw->type == 1)
|
483
|
+
{
|
484
|
+
rb_raise(rb_eRuntimeError, "bigBed files have no intervals! Use 'entries' instead.");
|
485
|
+
return Qnil;
|
486
|
+
}
|
487
|
+
|
488
|
+
if (rb_chrom != Qnil)
|
489
|
+
{
|
490
|
+
chrom = StringValueCStr(rb_chrom);
|
491
|
+
}
|
492
|
+
|
493
|
+
if (rb_start != Qnil)
|
494
|
+
startl = NUM2LONG(rb_start);
|
495
|
+
|
496
|
+
if (rb_end != Qnil)
|
497
|
+
endl = NUM2LONG(rb_end);
|
498
|
+
|
499
|
+
tid = bwGetTid(bw, chrom);
|
500
|
+
|
501
|
+
//Sanity check
|
502
|
+
if (endl == (unsigned long)-1 && tid != (uint32_t)-1)
|
503
|
+
endl = bw->cl->len[tid];
|
504
|
+
|
505
|
+
if (tid == (uint32_t)-1 || startl > end || endl > end)
|
506
|
+
{
|
507
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
508
|
+
return Qnil;
|
509
|
+
}
|
510
|
+
|
511
|
+
start = (uint32_t)startl;
|
512
|
+
end = (uint32_t)endl;
|
513
|
+
|
514
|
+
if (end <= start || end > bw->cl->len[tid] || start >= end)
|
515
|
+
{
|
516
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
517
|
+
return Qnil;
|
518
|
+
}
|
519
|
+
|
520
|
+
if (!hasEntries(bw))
|
521
|
+
{
|
522
|
+
return rb_ary_new2(0);
|
523
|
+
}
|
524
|
+
|
525
|
+
//Get the intervals
|
526
|
+
intervals = bwGetOverlappingIntervals(bw, chrom, start, end);
|
527
|
+
if (!intervals)
|
528
|
+
{
|
529
|
+
rb_raise(rb_eRuntimeError, "An error occurred while fetching the overlapping intervals!");
|
530
|
+
return Qnil;
|
531
|
+
}
|
532
|
+
if (!intervals->l)
|
533
|
+
{
|
534
|
+
return rb_ary_new2(0);
|
535
|
+
}
|
536
|
+
|
537
|
+
ret = rb_ary_new2(intervals->l);
|
538
|
+
for (i = 0; i < intervals->l; i++)
|
539
|
+
{
|
540
|
+
VALUE tupl = rb_ary_new3(3, UINT32_2NUM(intervals->start[i]), UINT32_2NUM(intervals->end[i]), DBL2NUM(intervals->value[i]));
|
541
|
+
rb_ary_store(ret, i, tupl);
|
542
|
+
}
|
543
|
+
bwDestroyOverlappingIntervals(intervals);
|
544
|
+
|
545
|
+
return ret;
|
546
|
+
}
|
547
|
+
|
548
|
+
static VALUE
|
549
|
+
bb_get_entries(VALUE self, VALUE rb_chrom, VALUE rb_start, VALUE rb_end, VALUE rb_with_string)
|
550
|
+
{
|
551
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
552
|
+
uint32_t start, end = -1, tid, i;
|
553
|
+
unsigned long startl, endl;
|
554
|
+
char *chrom;
|
555
|
+
VALUE ret, t;
|
556
|
+
int withString = 1;
|
557
|
+
bbOverlappingEntries_t *o;
|
558
|
+
|
559
|
+
if (!bw)
|
560
|
+
{
|
561
|
+
rb_raise(rb_eRuntimeError, "The bigBed file handle is not opened!");
|
562
|
+
return Qnil;
|
563
|
+
}
|
564
|
+
|
565
|
+
if (bw->type == 0)
|
566
|
+
{
|
567
|
+
rb_raise(rb_eRuntimeError, "bigWig files have no entries! Use 'values' or 'intervals' instead.");
|
568
|
+
return Qnil;
|
569
|
+
}
|
570
|
+
|
571
|
+
if (rb_chrom != Qnil)
|
572
|
+
{
|
573
|
+
chrom = StringValueCStr(rb_chrom);
|
574
|
+
}
|
575
|
+
|
576
|
+
if (rb_start != Qnil)
|
577
|
+
startl = NUM2LONG(rb_start);
|
578
|
+
|
579
|
+
if (rb_end != Qnil)
|
580
|
+
endl = NUM2LONG(rb_end);
|
581
|
+
|
582
|
+
if (rb_with_string != Qnil)
|
583
|
+
{
|
584
|
+
if (RTEST(rb_with_string))
|
585
|
+
withString = 1;
|
586
|
+
else
|
587
|
+
withString = 0;
|
588
|
+
}
|
589
|
+
|
590
|
+
tid = bwGetTid(bw, chrom);
|
591
|
+
|
592
|
+
//Sanity check
|
593
|
+
if (endl == (unsigned long)-1 && tid != (uint32_t)-1)
|
594
|
+
endl = bw->cl->len[tid];
|
595
|
+
if (tid == (uint32_t)-1 || startl > end || endl > end)
|
596
|
+
{
|
597
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
598
|
+
return Qnil;
|
599
|
+
}
|
600
|
+
|
601
|
+
start = (uint32_t)startl;
|
602
|
+
end = (uint32_t)endl;
|
603
|
+
|
604
|
+
if (end <= start || end > bw->cl->len[tid] || start >= end)
|
605
|
+
{
|
606
|
+
rb_raise(rb_eRuntimeError, "Invalid interval bounds!");
|
607
|
+
return Qnil;
|
608
|
+
}
|
609
|
+
|
610
|
+
o = bbGetOverlappingEntries(bw, chrom, start, end, withString);
|
611
|
+
if (!o)
|
612
|
+
{
|
613
|
+
rb_raise(rb_eRuntimeError, "An error occurred while fetching the overlapping entries!\n");
|
614
|
+
return Qnil;
|
615
|
+
}
|
616
|
+
if (!o->l)
|
617
|
+
{
|
618
|
+
return rb_ary_new2(0);
|
619
|
+
}
|
620
|
+
|
621
|
+
ret = rb_ary_new2(o->l);
|
622
|
+
if (!ret)
|
623
|
+
goto error;
|
624
|
+
|
625
|
+
for (i = 0; i < o->l; i++)
|
626
|
+
{
|
627
|
+
if (withString)
|
628
|
+
{
|
629
|
+
t = rb_ary_new3(3, UINT32_2NUM(o->start[i]), UINT32_2NUM(o->end[i]), rb_str_new2(o->str[i]));
|
630
|
+
}
|
631
|
+
else
|
632
|
+
{
|
633
|
+
t = rb_ary_new3(2, UINT32_2NUM(o->start[i]), UINT32_2NUM(o->end[i]));
|
634
|
+
}
|
635
|
+
if (!t)
|
636
|
+
goto error;
|
637
|
+
rb_ary_store(ret, i, t);
|
638
|
+
}
|
639
|
+
|
640
|
+
bbDestroyOverlappingEntries(o);
|
641
|
+
return ret;
|
642
|
+
|
643
|
+
error:
|
644
|
+
bbDestroyOverlappingEntries(o);
|
645
|
+
rb_raise(rb_eRuntimeError, "An error occurred while constructing the output!\n");
|
646
|
+
return Qnil;
|
647
|
+
}
|
648
|
+
|
649
|
+
static VALUE
|
650
|
+
bb_get_sql(VALUE self)
|
651
|
+
{
|
652
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
653
|
+
VALUE ret;
|
654
|
+
char *str;
|
655
|
+
|
656
|
+
if (!bw)
|
657
|
+
{
|
658
|
+
rb_raise(rb_eRuntimeError, "The bigBed file handle is not opened!");
|
659
|
+
return Qnil;
|
660
|
+
}
|
661
|
+
|
662
|
+
if (bw->type == 0)
|
663
|
+
{
|
664
|
+
rb_raise(rb_eRuntimeError, "bigWig files have no entries!");
|
665
|
+
return Qnil;
|
666
|
+
}
|
667
|
+
|
668
|
+
str = bbGetSQL(bw);
|
669
|
+
if (!str)
|
670
|
+
{
|
671
|
+
return Qnil;
|
672
|
+
}
|
673
|
+
|
674
|
+
ret = rb_str_new2(str);
|
675
|
+
if (str)
|
676
|
+
free(str);
|
677
|
+
|
678
|
+
return ret;
|
679
|
+
}
|
680
|
+
|
681
|
+
static VALUE
|
682
|
+
bw_get_file_type(VALUE self)
|
683
|
+
{
|
684
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
685
|
+
|
686
|
+
if (bw->type == 0)
|
687
|
+
{
|
688
|
+
return rb_str_new2("BigWig");
|
689
|
+
}
|
690
|
+
else if (bw->type == 1)
|
691
|
+
{
|
692
|
+
return rb_str_new2("BigBed");
|
693
|
+
}
|
694
|
+
|
695
|
+
return rb_str_new2("Unknown");
|
696
|
+
}
|
697
|
+
|
698
|
+
static VALUE
|
699
|
+
bw_is_bigwig_q(VALUE self)
|
700
|
+
{
|
701
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
702
|
+
|
703
|
+
if (bw->type == 0)
|
704
|
+
{
|
705
|
+
return Qtrue;
|
706
|
+
}
|
707
|
+
else
|
708
|
+
{
|
709
|
+
return Qfalse;
|
710
|
+
}
|
711
|
+
}
|
712
|
+
|
713
|
+
static VALUE
|
714
|
+
bw_is_bigbed_q(VALUE self)
|
715
|
+
{
|
716
|
+
bigWigFile_t *bw = get_bigWigFile(self);
|
717
|
+
|
718
|
+
if (bw->type == 1)
|
719
|
+
{
|
720
|
+
return Qtrue;
|
721
|
+
}
|
722
|
+
else
|
723
|
+
{
|
724
|
+
return Qfalse;
|
725
|
+
}
|
726
|
+
}
|
727
|
+
|
728
|
+
void Init_bigwigext()
|
729
|
+
{
|
730
|
+
rb_Bio = rb_define_module("Bio");
|
731
|
+
rb_BigWig = rb_define_class_under(rb_Bio, "BigWig", rb_cObject);
|
732
|
+
|
733
|
+
rb_define_alloc_func(rb_BigWig, bigwig_allocate);
|
734
|
+
|
735
|
+
rb_define_private_method(rb_BigWig, "initialize_raw", bigwig_init, 2);
|
736
|
+
rb_define_method(rb_BigWig, "close", bigwig_close, 0);
|
737
|
+
rb_define_method(rb_BigWig, "header", bw_get_header, 0);
|
738
|
+
rb_define_method(rb_BigWig, "chroms", bw_get_chroms, -1);
|
739
|
+
rb_define_private_method(rb_BigWig, "stats_raw", bw_get_stats, 6);
|
740
|
+
rb_define_private_method(rb_BigWig, "values_raw", bw_get_values, 3);
|
741
|
+
rb_define_private_method(rb_BigWig, "intervals_raw", bw_get_intervals, 3);
|
742
|
+
rb_define_private_method(rb_BigWig, "entries_raw", bb_get_entries, 4);
|
743
|
+
rb_define_method(rb_BigWig, "sql", bb_get_sql, 0);
|
744
|
+
rb_define_method(rb_BigWig, "file_type", bw_get_file_type, 0);
|
745
|
+
rb_define_method(rb_BigWig, "is_bigwig?", bw_is_bigwig_q, 0);
|
746
|
+
rb_define_method(rb_BigWig, "is_bigbed?", bw_is_bigbed_q, 0);
|
747
|
+
}
|