htslib 0.0.8 → 0.2.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/README.md +40 -33
- data/lib/hts/bam/aux.rb +40 -0
- data/lib/hts/bam/cigar.rb +10 -14
- data/lib/hts/bam/flag.rb +27 -58
- data/lib/hts/bam/header.rb +3 -5
- data/lib/hts/bam/record.rb +97 -56
- data/lib/hts/bam.rb +134 -64
- data/lib/hts/bcf/format.rb +85 -14
- data/lib/hts/bcf/header.rb +5 -4
- data/lib/hts/bcf/info.rb +58 -32
- data/lib/hts/bcf/record.rb +67 -38
- data/lib/hts/bcf.rb +155 -38
- data/lib/hts/faidx.rb +28 -19
- data/lib/hts/ffi_ext/pointer.rb +18 -0
- data/lib/hts/hts.rb +121 -0
- data/lib/hts/libhts/bgzf.rb +10 -5
- data/lib/hts/libhts/constants.rb +46 -7
- data/lib/hts/libhts/cram.rb +300 -0
- data/lib/hts/libhts/hfile.rb +37 -11
- data/lib/hts/libhts/hts.rb +158 -25
- data/lib/hts/libhts/sam.rb +683 -94
- data/lib/hts/libhts/sam_funcs.rb +92 -587
- data/lib/hts/libhts/vcf.rb +433 -220
- data/lib/hts/libhts/vcf_funcs.rb +232 -424
- data/lib/hts/libhts.rb +9 -3
- data/lib/hts/tbx.rb +45 -0
- data/lib/hts/version.rb +1 -1
- data/lib/htslib.rb +5 -5
- metadata +8 -4
- data/lib/hts/tabix.rb +0 -28
data/lib/hts/libhts/constants.rb
CHANGED
@@ -1,8 +1,8 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
2
|
|
3
3
|
module HTS
|
4
|
+
# Module for working with C HTSlib.
|
4
5
|
module LibHTS
|
5
|
-
typedef :pointer, :HFILE
|
6
6
|
typedef :int64, :hts_pos_t
|
7
7
|
typedef :pointer, :bam_plp_auto_f
|
8
8
|
|
@@ -25,6 +25,25 @@ module HTS
|
|
25
25
|
:f, :pointer # kstream_t
|
26
26
|
end
|
27
27
|
|
28
|
+
# HFile
|
29
|
+
|
30
|
+
class HFile < FFI::BitStruct
|
31
|
+
layout \
|
32
|
+
:buffer, :string,
|
33
|
+
:begin, :string,
|
34
|
+
:end, :string,
|
35
|
+
:limit, :string,
|
36
|
+
:backend, :pointer,
|
37
|
+
:offset, :size_t,
|
38
|
+
:_flags, :uint,
|
39
|
+
:has_errno, :int
|
40
|
+
|
41
|
+
bit_fields :_flags,
|
42
|
+
:at_eof, 1,
|
43
|
+
:mobile, 1,
|
44
|
+
:readonly, 1
|
45
|
+
end
|
46
|
+
|
28
47
|
# BGZF
|
29
48
|
class BGZF < FFI::BitStruct
|
30
49
|
layout \
|
@@ -38,7 +57,7 @@ module HTS
|
|
38
57
|
:uncompressed_block, :pointer,
|
39
58
|
:compressed_block, :pointer,
|
40
59
|
:cache, :pointer,
|
41
|
-
:fp,
|
60
|
+
:fp, HFile.ptr,
|
42
61
|
:mt, :pointer,
|
43
62
|
:idx, :pointer,
|
44
63
|
:idx_build_otf, :int,
|
@@ -171,6 +190,16 @@ module HTS
|
|
171
190
|
)
|
172
191
|
end
|
173
192
|
|
193
|
+
class HtsReglist < FFI::Struct
|
194
|
+
layout \
|
195
|
+
:reg, :string,
|
196
|
+
:intervals, :pointer, # hts_pair_pos_t
|
197
|
+
:tid, :int,
|
198
|
+
:count, :uint32_t,
|
199
|
+
:min_beg, :hts_pos_t,
|
200
|
+
:max_end, :hts_pos_t
|
201
|
+
end
|
202
|
+
|
174
203
|
# HtsFile
|
175
204
|
class SamHdr < FFI::Struct
|
176
205
|
layout \
|
@@ -198,8 +227,8 @@ module HTS
|
|
198
227
|
:fp,
|
199
228
|
union_layout(
|
200
229
|
:bgzf, BGZF.ptr,
|
201
|
-
:cram, :pointer,
|
202
|
-
:hfile,
|
230
|
+
:cram, :pointer, # cram_fd
|
231
|
+
:hfile, HFile.ptr
|
203
232
|
),
|
204
233
|
:state, :pointer,
|
205
234
|
:format, HtsFormat,
|
@@ -245,7 +274,7 @@ module HTS
|
|
245
274
|
:n_reg, :int,
|
246
275
|
:beg, :int64,
|
247
276
|
:end, :int64,
|
248
|
-
:reg_list, :pointer,
|
277
|
+
:reg_list, :pointer, # HtsReglist.ptr,
|
249
278
|
:curr_tid, :int,
|
250
279
|
:curr_reg, :int,
|
251
280
|
:curr_intv, :int,
|
@@ -461,8 +490,8 @@ module HTS
|
|
461
490
|
:id, :string,
|
462
491
|
:als, :pointer, # (\\0-separated string)
|
463
492
|
:allele, :pointer,
|
464
|
-
:info, :pointer, # BcfInfo.ptr,
|
465
|
-
:fmt, BcfFmt.ptr,
|
493
|
+
:info, :pointer, # array of BcfInfo.ptr,
|
494
|
+
:fmt, :pointer, # array of BcfFmt.ptr,
|
466
495
|
:var, BcfVariant.ptr,
|
467
496
|
:n_var, :int,
|
468
497
|
:var_type, :int,
|
@@ -498,5 +527,15 @@ module HTS
|
|
498
527
|
LibHTS.bcf_destroy(ptr) unless ptr.null?
|
499
528
|
end
|
500
529
|
end
|
530
|
+
|
531
|
+
CramContentType = enum(
|
532
|
+
:ct_error, -1,
|
533
|
+
:file_header, 0,
|
534
|
+
:compression_header, 1,
|
535
|
+
:mapped_slice, 2,
|
536
|
+
:unmapped_slice, 3, # cram v1.0 only
|
537
|
+
:external, 4,
|
538
|
+
:core, 5
|
539
|
+
)
|
501
540
|
end
|
502
541
|
end
|
@@ -0,0 +1,300 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module HTS
|
4
|
+
module LibHTS
|
5
|
+
typedef :pointer, :cram_fd
|
6
|
+
typedef :pointer, :cram_container
|
7
|
+
typedef :pointer, :cram_block
|
8
|
+
typedef :pointer, :cram_metrics
|
9
|
+
|
10
|
+
attach_function \
|
11
|
+
:cram_fd_get_header,
|
12
|
+
[:cram_fd],
|
13
|
+
SamHdr.by_ref
|
14
|
+
|
15
|
+
attach_function \
|
16
|
+
:cram_fd_set_header,
|
17
|
+
[:cram_fd, SamHdr.by_ref],
|
18
|
+
:void
|
19
|
+
|
20
|
+
attach_function \
|
21
|
+
:cram_fd_get_version,
|
22
|
+
[:cram_fd],
|
23
|
+
:int
|
24
|
+
|
25
|
+
attach_function \
|
26
|
+
:cram_fd_set_version,
|
27
|
+
%i[cram_fd int],
|
28
|
+
:void
|
29
|
+
|
30
|
+
attach_function \
|
31
|
+
:cram_major_vers,
|
32
|
+
[:cram_fd],
|
33
|
+
:int
|
34
|
+
|
35
|
+
attach_function \
|
36
|
+
:cram_minor_vers,
|
37
|
+
[:cram_fd],
|
38
|
+
:int
|
39
|
+
|
40
|
+
attach_function \
|
41
|
+
:cram_fd_get_fp,
|
42
|
+
[:cram_fd],
|
43
|
+
HFile.by_ref
|
44
|
+
|
45
|
+
attach_function \
|
46
|
+
:cram_fd_set_fp,
|
47
|
+
[:cram_fd, HFile],
|
48
|
+
:void
|
49
|
+
|
50
|
+
attach_function \
|
51
|
+
:cram_container_get_length,
|
52
|
+
[:cram_container],
|
53
|
+
:int32
|
54
|
+
|
55
|
+
attach_function \
|
56
|
+
:cram_container_set_length,
|
57
|
+
%i[cram_container int32],
|
58
|
+
:void
|
59
|
+
|
60
|
+
attach_function \
|
61
|
+
:cram_container_get_num_blocks,
|
62
|
+
[:cram_container],
|
63
|
+
:int32
|
64
|
+
|
65
|
+
attach_function \
|
66
|
+
:cram_container_set_num_blocks,
|
67
|
+
%i[cram_container int32],
|
68
|
+
:void
|
69
|
+
|
70
|
+
attach_function \
|
71
|
+
:cram_container_get_landmarks,
|
72
|
+
%i[cram_container int32],
|
73
|
+
:pointer
|
74
|
+
|
75
|
+
attach_function \
|
76
|
+
:cram_container_set_landmarks,
|
77
|
+
%i[cram_container int32 pointer],
|
78
|
+
:void
|
79
|
+
|
80
|
+
attach_function \
|
81
|
+
:cram_container_is_empty,
|
82
|
+
[:cram_fd],
|
83
|
+
:int
|
84
|
+
|
85
|
+
attach_function \
|
86
|
+
:cram_block_get_content_id,
|
87
|
+
[:cram_block],
|
88
|
+
:int32
|
89
|
+
|
90
|
+
attach_function \
|
91
|
+
:cram_block_get_comp_size,
|
92
|
+
[:cram_block],
|
93
|
+
:int32
|
94
|
+
|
95
|
+
attach_function \
|
96
|
+
:cram_block_get_uncomp_size,
|
97
|
+
[:cram_block],
|
98
|
+
:int32
|
99
|
+
|
100
|
+
attach_function \
|
101
|
+
:cram_block_get_crc32,
|
102
|
+
[:cram_block],
|
103
|
+
:int32
|
104
|
+
|
105
|
+
attach_function \
|
106
|
+
:cram_block_get_data,
|
107
|
+
[:cram_block],
|
108
|
+
:pointer
|
109
|
+
|
110
|
+
attach_function \
|
111
|
+
:cram_block_get_content_type,
|
112
|
+
[:cram_block],
|
113
|
+
CramContentType
|
114
|
+
|
115
|
+
attach_function \
|
116
|
+
:cram_block_set_content_id,
|
117
|
+
%i[cram_block int32],
|
118
|
+
:void
|
119
|
+
|
120
|
+
attach_function \
|
121
|
+
:cram_block_set_comp_size,
|
122
|
+
%i[cram_block int32],
|
123
|
+
:void
|
124
|
+
|
125
|
+
attach_function \
|
126
|
+
:cram_block_set_uncomp_size,
|
127
|
+
%i[cram_block int32],
|
128
|
+
:void
|
129
|
+
|
130
|
+
attach_function \
|
131
|
+
:cram_block_set_crc32,
|
132
|
+
%i[cram_block int32],
|
133
|
+
:void
|
134
|
+
|
135
|
+
attach_function \
|
136
|
+
:cram_block_set_data,
|
137
|
+
%i[cram_block pointer],
|
138
|
+
:void
|
139
|
+
|
140
|
+
attach_function \
|
141
|
+
:cram_block_append,
|
142
|
+
%i[cram_block pointer int],
|
143
|
+
:int
|
144
|
+
|
145
|
+
attach_function \
|
146
|
+
:cram_block_update_size,
|
147
|
+
[:cram_block],
|
148
|
+
:void
|
149
|
+
|
150
|
+
attach_function \
|
151
|
+
:cram_block_get_offset,
|
152
|
+
[:cram_block],
|
153
|
+
:size_t
|
154
|
+
|
155
|
+
attach_function \
|
156
|
+
:cram_block_set_offset,
|
157
|
+
%i[cram_block size_t],
|
158
|
+
:void
|
159
|
+
|
160
|
+
attach_function \
|
161
|
+
:cram_block_size,
|
162
|
+
[:cram_block],
|
163
|
+
:uint32
|
164
|
+
|
165
|
+
attach_function \
|
166
|
+
:cram_transcode_rg,
|
167
|
+
%i[cram_fd cram_fd cram_container int pointer pointer],
|
168
|
+
:int
|
169
|
+
|
170
|
+
attach_function \
|
171
|
+
:cram_copy_slice,
|
172
|
+
%i[cram_fd cram_fd int32],
|
173
|
+
:int
|
174
|
+
|
175
|
+
attach_function \
|
176
|
+
:cram_new_block,
|
177
|
+
[CramContentType, :int],
|
178
|
+
:cram_block
|
179
|
+
|
180
|
+
attach_function \
|
181
|
+
:cram_read_block,
|
182
|
+
[:cram_fd],
|
183
|
+
:cram_block
|
184
|
+
|
185
|
+
attach_function \
|
186
|
+
:cram_write_block,
|
187
|
+
%i[cram_fd cram_block],
|
188
|
+
:int
|
189
|
+
|
190
|
+
attach_function \
|
191
|
+
:cram_free_block,
|
192
|
+
[:cram_block],
|
193
|
+
:void
|
194
|
+
|
195
|
+
attach_function \
|
196
|
+
:cram_uncompress_block,
|
197
|
+
[:cram_block],
|
198
|
+
:int
|
199
|
+
|
200
|
+
attach_function \
|
201
|
+
:cram_compress_block,
|
202
|
+
%i[cram_fd cram_block cram_metrics int int],
|
203
|
+
:int
|
204
|
+
|
205
|
+
# attach_function \
|
206
|
+
# :cram_compress_block2,
|
207
|
+
# %i[cram_fd cram_slice cram_block cram_metrics int int],
|
208
|
+
# :int
|
209
|
+
|
210
|
+
attach_function \
|
211
|
+
:cram_new_container,
|
212
|
+
%i[int int],
|
213
|
+
:cram_container
|
214
|
+
|
215
|
+
attach_function \
|
216
|
+
:cram_free_container,
|
217
|
+
[:cram_container],
|
218
|
+
:void
|
219
|
+
|
220
|
+
attach_function \
|
221
|
+
:cram_read_container,
|
222
|
+
[:cram_fd],
|
223
|
+
:cram_container
|
224
|
+
|
225
|
+
attach_function \
|
226
|
+
:cram_write_container,
|
227
|
+
%i[cram_fd cram_container],
|
228
|
+
:int
|
229
|
+
|
230
|
+
attach_function \
|
231
|
+
:cram_store_container,
|
232
|
+
%i[cram_fd cram_container string pointer],
|
233
|
+
:int
|
234
|
+
|
235
|
+
attach_function \
|
236
|
+
:cram_container_size,
|
237
|
+
[:cram_container],
|
238
|
+
:int
|
239
|
+
|
240
|
+
attach_function \
|
241
|
+
:cram_open,
|
242
|
+
%i[string string],
|
243
|
+
:cram_fd
|
244
|
+
|
245
|
+
attach_function \
|
246
|
+
:cram_dopen,
|
247
|
+
%i[pointer string string],
|
248
|
+
:cram_fd
|
249
|
+
|
250
|
+
attach_function \
|
251
|
+
:cram_close,
|
252
|
+
[:cram_fd],
|
253
|
+
:int
|
254
|
+
|
255
|
+
attach_function \
|
256
|
+
:cram_seek,
|
257
|
+
%i[pointer off_t int],
|
258
|
+
:int # FIXME: pointer should be :cram_fd
|
259
|
+
|
260
|
+
attach_function \
|
261
|
+
:cram_flush,
|
262
|
+
[:cram_fd],
|
263
|
+
:int
|
264
|
+
|
265
|
+
attach_function \
|
266
|
+
:cram_eof,
|
267
|
+
[:cram_fd],
|
268
|
+
:int
|
269
|
+
|
270
|
+
attach_function \
|
271
|
+
:cram_set_option,
|
272
|
+
[:cram_fd, HtsFmtOption, :varargs],
|
273
|
+
:int
|
274
|
+
|
275
|
+
attach_function \
|
276
|
+
:cram_set_voption,
|
277
|
+
[:cram_fd, HtsFmtOption, :pointer], # va_list
|
278
|
+
:int
|
279
|
+
|
280
|
+
attach_function \
|
281
|
+
:cram_set_header,
|
282
|
+
[:cram_fd, SamHdr.by_ref],
|
283
|
+
:int
|
284
|
+
|
285
|
+
attach_function \
|
286
|
+
:cram_check_EOF,
|
287
|
+
[:cram_fd],
|
288
|
+
:int
|
289
|
+
|
290
|
+
attach_function \
|
291
|
+
:int32_put_blk,
|
292
|
+
%i[cram_block int32_t],
|
293
|
+
:int
|
294
|
+
|
295
|
+
attach_function \
|
296
|
+
:cram_get_refs,
|
297
|
+
[HtsFile.by_ref],
|
298
|
+
:pointer # refs_t
|
299
|
+
end
|
300
|
+
end
|
data/lib/hts/libhts/hfile.rb
CHANGED
@@ -6,13 +6,13 @@ module HTS
|
|
6
6
|
attach_function \
|
7
7
|
:hopen,
|
8
8
|
%i[string string varargs],
|
9
|
-
|
9
|
+
HFile.by_ref
|
10
10
|
|
11
11
|
# Associate a stream with an existing open file descriptor
|
12
12
|
attach_function \
|
13
13
|
:hdopen,
|
14
14
|
%i[int string],
|
15
|
-
|
15
|
+
HFile.by_ref
|
16
16
|
|
17
17
|
# Report whether the file name or URL denotes remote storage
|
18
18
|
attach_function \
|
@@ -29,55 +29,81 @@ module HTS
|
|
29
29
|
# Flush (for output streams) and close the stream
|
30
30
|
attach_function \
|
31
31
|
:hclose,
|
32
|
-
[
|
32
|
+
[HFile],
|
33
33
|
:int
|
34
34
|
|
35
35
|
# Close the stream, without flushing or propagating errors
|
36
36
|
attach_function \
|
37
37
|
:hclose_abruptly,
|
38
|
-
[
|
38
|
+
[HFile],
|
39
39
|
:void
|
40
40
|
|
41
41
|
# Reposition the read/write stream offset
|
42
42
|
attach_function \
|
43
43
|
:hseek,
|
44
|
-
|
44
|
+
[HFile, :off_t, :int],
|
45
45
|
:off_t
|
46
46
|
|
47
|
+
# Report the current stream offset
|
48
|
+
def self.htell(fp)
|
49
|
+
# TODO: This is a hack. Is this OK?
|
50
|
+
bg = FFI::Pointer.new(:int, fp.pointer.address + fp.offset_of(:begin)).read_int
|
51
|
+
bf = FFI::Pointer.new(:int, fp.pointer.address + fp.offset_of(:buffer)).read_int
|
52
|
+
fp[:offset] + (bg - bf)
|
53
|
+
end
|
54
|
+
|
47
55
|
# Read from the stream until the delimiter, up to a maximum length
|
48
56
|
attach_function \
|
49
57
|
:hgetdelim,
|
50
|
-
|
58
|
+
[:string, :size_t, :int, HFile],
|
51
59
|
:ssize_t
|
52
60
|
|
53
61
|
# Read a line from the stream, up to a maximum length
|
54
62
|
attach_function \
|
55
63
|
:hgets,
|
56
|
-
|
64
|
+
[:string, :int, HFile],
|
57
65
|
:string
|
58
66
|
|
59
67
|
# Peek at characters to be read without removing them from buffers
|
60
68
|
attach_function \
|
61
69
|
:hpeek,
|
62
|
-
|
70
|
+
[HFile, :pointer, :size_t],
|
63
71
|
:ssize_t
|
64
72
|
|
65
73
|
# For writing streams, flush buffered output to the underlying stream
|
66
74
|
attach_function \
|
67
75
|
:hflush,
|
68
|
-
[
|
76
|
+
[HFile],
|
69
77
|
:int
|
70
78
|
|
71
79
|
# For hfile_mem: get the internal buffer and it's size from a hfile
|
72
80
|
attach_function \
|
73
81
|
:hfile_mem_get_buffer,
|
74
|
-
|
82
|
+
[HFile, :pointer],
|
75
83
|
:string
|
76
84
|
|
77
85
|
# For hfile_mem: get the internal buffer and it's size from a hfile.
|
78
86
|
attach_function \
|
79
87
|
:hfile_mem_steal_buffer,
|
80
|
-
|
88
|
+
[HFile, :pointer],
|
81
89
|
:string
|
90
|
+
|
91
|
+
# Fills out sc_list[] with the list of known URL schemes.
|
92
|
+
attach_function \
|
93
|
+
:hfile_list_schemes,
|
94
|
+
%i[string pointer pointer], # mutable string
|
95
|
+
:int
|
96
|
+
|
97
|
+
# Fills out plist[] with the list of known hFILE plugins.
|
98
|
+
attach_function \
|
99
|
+
:hfile_list_plugins,
|
100
|
+
%i[pointer pointer], # mutable string
|
101
|
+
:int
|
102
|
+
|
103
|
+
# Tests for the presence of a specific hFILE plugin.
|
104
|
+
attach_function \
|
105
|
+
:hfile_has_plugin,
|
106
|
+
[:string],
|
107
|
+
:int
|
82
108
|
end
|
83
109
|
end
|