line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package App::Sandy::Seq; |
2
|
|
|
|
|
|
|
# ABSTRACT: Base class to simulate seq entries |
3
|
|
|
|
|
|
|
|
4
|
6
|
|
|
6
|
|
4505
|
use App::Sandy::Base 'class'; |
|
6
|
|
|
|
|
17
|
|
|
6
|
|
|
|
|
46
|
|
5
|
6
|
|
|
6
|
|
3738
|
use App::Sandy::Quality; |
|
6
|
|
|
|
|
3090
|
|
|
6
|
|
|
|
|
304
|
|
6
|
|
|
|
|
|
|
|
7
|
6
|
|
|
6
|
|
73
|
use constant NUM_TRIES => 1000; |
|
6
|
|
|
|
|
12
|
|
|
6
|
|
|
|
|
8983
|
|
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
with qw{ |
10
|
|
|
|
|
|
|
App::Sandy::Role::RunTimeTemplate |
11
|
|
|
|
|
|
|
App::Sandy::Role::Template::Fastq |
12
|
|
|
|
|
|
|
App::Sandy::Role::Template::Sam |
13
|
|
|
|
|
|
|
}; |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
our $VERSION = '0.24'; # VERSION |
16
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
has 'format' => ( |
18
|
|
|
|
|
|
|
is => 'ro', |
19
|
|
|
|
|
|
|
isa => 'My:Format', |
20
|
|
|
|
|
|
|
required => 1 |
21
|
|
|
|
|
|
|
); |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
has 'template_id' => ( |
24
|
|
|
|
|
|
|
is => 'ro', |
25
|
|
|
|
|
|
|
isa => 'Str', |
26
|
|
|
|
|
|
|
required => 1 |
27
|
|
|
|
|
|
|
); |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
has 'sequencing_error' => ( |
30
|
|
|
|
|
|
|
is => 'ro', |
31
|
|
|
|
|
|
|
isa => 'My:NumHS', |
32
|
|
|
|
|
|
|
required => 1 |
33
|
|
|
|
|
|
|
); |
34
|
|
|
|
|
|
|
|
35
|
|
|
|
|
|
|
has 'quality_profile' => ( |
36
|
|
|
|
|
|
|
is => 'ro', |
37
|
|
|
|
|
|
|
isa => 'My:QualityP', |
38
|
|
|
|
|
|
|
required => 1, |
39
|
|
|
|
|
|
|
coerce => 1 |
40
|
|
|
|
|
|
|
); |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
has 'read_mean' => ( |
43
|
|
|
|
|
|
|
is => 'ro', |
44
|
|
|
|
|
|
|
isa => 'My:IntGt0', |
45
|
|
|
|
|
|
|
required => 1 |
46
|
|
|
|
|
|
|
); |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
has 'read_stdd' => ( |
49
|
|
|
|
|
|
|
is => 'ro', |
50
|
|
|
|
|
|
|
isa => 'My:IntGe0', |
51
|
|
|
|
|
|
|
required => 1 |
52
|
|
|
|
|
|
|
); |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
has '_template_id' => ( |
55
|
|
|
|
|
|
|
traits => ['Code'], |
56
|
|
|
|
|
|
|
is => 'ro', |
57
|
|
|
|
|
|
|
isa => 'CodeRef', |
58
|
|
|
|
|
|
|
builder => '_build_template_id', |
59
|
|
|
|
|
|
|
lazy_build => 1, |
60
|
|
|
|
|
|
|
handles => { |
61
|
|
|
|
|
|
|
_gen_id => 'execute' |
62
|
|
|
|
|
|
|
} |
63
|
|
|
|
|
|
|
); |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
has '_template_seq' => ( |
66
|
|
|
|
|
|
|
traits => ['Code'], |
67
|
|
|
|
|
|
|
is => 'ro', |
68
|
|
|
|
|
|
|
isa => 'CodeRef', |
69
|
|
|
|
|
|
|
builder => '_build_template_seq', |
70
|
|
|
|
|
|
|
lazy_build => 1, |
71
|
|
|
|
|
|
|
handles => { |
72
|
|
|
|
|
|
|
_gen_seq => 'execute' |
73
|
|
|
|
|
|
|
} |
74
|
|
|
|
|
|
|
); |
75
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
has '_info' => ( |
77
|
|
|
|
|
|
|
traits => ['Hash'], |
78
|
|
|
|
|
|
|
is => 'ro', |
79
|
|
|
|
|
|
|
isa => 'HashRef[Str]', |
80
|
|
|
|
|
|
|
builder => '_build_info', |
81
|
|
|
|
|
|
|
lazy_build => 1, |
82
|
|
|
|
|
|
|
handles => { |
83
|
|
|
|
|
|
|
_set_info => 'set', |
84
|
|
|
|
|
|
|
_get_info => 'get' |
85
|
|
|
|
|
|
|
} |
86
|
|
|
|
|
|
|
); |
87
|
|
|
|
|
|
|
|
88
|
|
|
|
|
|
|
has '_read_size' => ( |
89
|
|
|
|
|
|
|
traits => ['Code'], |
90
|
|
|
|
|
|
|
is => 'ro', |
91
|
|
|
|
|
|
|
isa => 'CodeRef', |
92
|
|
|
|
|
|
|
builder => '_build_read_size', |
93
|
|
|
|
|
|
|
lazy_build => 1, |
94
|
|
|
|
|
|
|
handles => { |
95
|
|
|
|
|
|
|
_get_read_size => 'execute' |
96
|
|
|
|
|
|
|
} |
97
|
|
|
|
|
|
|
); |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
has '_quality' => ( |
100
|
|
|
|
|
|
|
is => 'ro', |
101
|
|
|
|
|
|
|
isa => 'App::Sandy::Quality', |
102
|
|
|
|
|
|
|
builder => '_build_quality', |
103
|
|
|
|
|
|
|
lazy_build => 1, |
104
|
|
|
|
|
|
|
handles => ['gen_quality'] |
105
|
|
|
|
|
|
|
); |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
has '_sym_table' => ( |
108
|
|
|
|
|
|
|
is => 'ro', |
109
|
|
|
|
|
|
|
isa => 'HashRef[Str]', |
110
|
|
|
|
|
|
|
builder => '_build_sym_table', |
111
|
|
|
|
|
|
|
lazy_build => 1 |
112
|
|
|
|
|
|
|
); |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
sub BUILD { |
115
|
52
|
|
|
52
|
0
|
112
|
my $self = shift; |
116
|
|
|
|
|
|
|
## Just to ensure that the lazy attributes are built before &new returns |
117
|
52
|
|
|
|
|
1529
|
$self->_quality; |
118
|
|
|
|
|
|
|
} |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
sub _build_sym_table { |
121
|
14
|
|
|
14
|
|
574
|
my $sym_table = { |
122
|
|
|
|
|
|
|
'%q' => '$info->{quality_profile}', |
123
|
|
|
|
|
|
|
'%m' => '$info->{read_mean}', |
124
|
|
|
|
|
|
|
'%d' => '$info->{read_stdd}', |
125
|
|
|
|
|
|
|
'%r' => '$info->{read_size}', |
126
|
|
|
|
|
|
|
'%e' => '$info->{sequencing_error}', |
127
|
|
|
|
|
|
|
'%c' => '$info->{seq_id}', |
128
|
|
|
|
|
|
|
'%C' => '$info->{seq_id_type}', |
129
|
|
|
|
|
|
|
'%a' => '$info->{start_ref}', |
130
|
|
|
|
|
|
|
'%b' => '$info->{end_ref}', |
131
|
|
|
|
|
|
|
'%t' => '$info->{start}', |
132
|
|
|
|
|
|
|
'%n' => '$info->{end}', |
133
|
|
|
|
|
|
|
'%i' => '$info->{instrument}', |
134
|
|
|
|
|
|
|
'%I' => '$info->{id}', |
135
|
|
|
|
|
|
|
'%R' => '$info->{read}', |
136
|
|
|
|
|
|
|
'%U' => '$info->{num}', |
137
|
|
|
|
|
|
|
'%s' => '$info->{strand}', |
138
|
|
|
|
|
|
|
'%x' => '$info->{error}', |
139
|
|
|
|
|
|
|
'%v' => '$info->{var}' |
140
|
|
|
|
|
|
|
}; |
141
|
|
|
|
|
|
|
|
142
|
14
|
|
|
|
|
259
|
return $sym_table; |
143
|
|
|
|
|
|
|
} |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
sub _build_template_id { |
146
|
14
|
|
|
14
|
|
126
|
my $self = shift; |
147
|
14
|
|
|
|
|
553
|
my $sym_table = $self->_sym_table; |
148
|
14
|
|
|
|
|
470
|
return $self->with_compile_template($self->template_id, 'info', $sym_table); |
149
|
|
|
|
|
|
|
} |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
sub _build_template_seq { |
152
|
25
|
|
|
25
|
|
75
|
my $self = shift; |
153
|
|
|
|
|
|
|
|
154
|
25
|
|
|
|
|
706
|
my $format = $self->format; |
155
|
25
|
|
|
|
|
144
|
my $gen_seq; |
156
|
|
|
|
|
|
|
|
157
|
25
|
50
|
|
|
|
212
|
if ($format =~ /fastq/) { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
158
|
25
|
|
|
9161
|
|
174
|
$gen_seq = sub { $self->with_fastq_template(@_) }; |
|
9161
|
|
|
|
|
29695
|
|
159
|
|
|
|
|
|
|
} elsif ($format eq 'sam') { |
160
|
0
|
|
|
0
|
|
0
|
$gen_seq = sub { $self->with_sam_align_template(@_) }; |
|
0
|
|
|
|
|
0
|
|
161
|
|
|
|
|
|
|
} elsif ($format eq 'bam') { |
162
|
0
|
|
|
0
|
|
0
|
$gen_seq = sub { $self->with_bam_align_template(@_) }; |
|
0
|
|
|
|
|
0
|
|
163
|
|
|
|
|
|
|
} else { |
164
|
0
|
|
|
|
|
0
|
croak "No valid format: '$format'"; |
165
|
|
|
|
|
|
|
} |
166
|
|
|
|
|
|
|
|
167
|
25
|
|
|
|
|
875
|
return $gen_seq; |
168
|
|
|
|
|
|
|
} |
169
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
sub _build_info { |
171
|
14
|
|
|
14
|
|
762
|
my $self = shift; |
172
|
|
|
|
|
|
|
|
173
|
14
|
|
|
|
|
576
|
my $info = { |
174
|
|
|
|
|
|
|
instrument => 'SR', |
175
|
|
|
|
|
|
|
quality_profile => $self->quality_profile, |
176
|
|
|
|
|
|
|
sequencing_error => $self->sequencing_error, |
177
|
|
|
|
|
|
|
read_mean => $self->read_mean, |
178
|
|
|
|
|
|
|
read_stdd => $self->read_stdd |
179
|
|
|
|
|
|
|
}; |
180
|
|
|
|
|
|
|
|
181
|
14
|
|
|
|
|
278
|
return $info; |
182
|
|
|
|
|
|
|
} |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
sub _build_read_size { |
185
|
14
|
|
|
14
|
|
114
|
my $self = shift; |
186
|
14
|
|
|
|
|
67
|
my $fun; |
187
|
|
|
|
|
|
|
|
188
|
14
|
50
|
|
|
|
547
|
if ($self->read_stdd == 0) { |
189
|
14
|
|
|
6860
|
|
208
|
$fun = sub { $self->read_mean }; |
|
6860
|
|
|
|
|
166345
|
|
190
|
|
|
|
|
|
|
} else { |
191
|
|
|
|
|
|
|
$fun = sub { |
192
|
0
|
|
|
0
|
|
0
|
my $rng = shift; |
193
|
0
|
|
|
|
|
0
|
my $size = 0; |
194
|
0
|
|
|
|
|
0
|
my $random_tries = 0; |
195
|
0
|
|
|
|
|
0
|
until ($size > 0) { |
196
|
0
|
0
|
|
|
|
0
|
if (++$random_tries > NUM_TRIES) { |
197
|
0
|
|
|
|
|
0
|
croak "So many tries to calculate a seq size greater than zero ..."; |
198
|
|
|
|
|
|
|
} |
199
|
0
|
|
|
|
|
0
|
$size = $rng->get_norm($self->read_mean, $self->read_stdd); |
200
|
|
|
|
|
|
|
} |
201
|
0
|
|
|
|
|
0
|
return $size; |
202
|
0
|
|
|
|
|
0
|
}; |
203
|
|
|
|
|
|
|
} |
204
|
|
|
|
|
|
|
|
205
|
14
|
|
|
|
|
533
|
return $fun; |
206
|
|
|
|
|
|
|
} |
207
|
|
|
|
|
|
|
|
208
|
|
|
|
|
|
|
sub _build_quality { |
209
|
52
|
|
|
52
|
|
97
|
my $self = shift; |
210
|
52
|
|
|
|
|
1410
|
App::Sandy::Quality->new( |
211
|
|
|
|
|
|
|
quality_profile => $self->quality_profile |
212
|
|
|
|
|
|
|
); |
213
|
|
|
|
|
|
|
} |
214
|
|
|
|
|
|
|
|
215
|
|
|
|
|
|
|
sub gen_sam_header { |
216
|
0
|
|
|
0
|
0
|
|
my ($self, $argv) = @_; |
217
|
0
|
|
|
|
|
|
my $format = $self->format; |
218
|
0
|
|
|
|
|
|
my $header; |
219
|
|
|
|
|
|
|
|
220
|
0
|
0
|
|
|
|
|
if ($format eq 'sam') { |
|
|
0
|
|
|
|
|
|
221
|
0
|
|
|
|
|
|
$header = $self->with_sam_header_template($argv); |
222
|
|
|
|
|
|
|
} elsif ($format eq 'bam') { |
223
|
0
|
|
|
|
|
|
$header = $self->with_bam_header_template($argv); |
224
|
|
|
|
|
|
|
} else { |
225
|
0
|
|
|
|
|
|
croak "No valid format: '$format'"; |
226
|
|
|
|
|
|
|
} |
227
|
|
|
|
|
|
|
|
228
|
0
|
|
|
|
|
|
return $header; |
229
|
|
|
|
|
|
|
} |
230
|
|
|
|
|
|
|
|
231
|
|
|
|
|
|
|
sub gen_eof_marker { |
232
|
0
|
|
|
0
|
0
|
|
my ($self, $file) = @_; |
233
|
|
|
|
|
|
|
|
234
|
0
|
0
|
|
|
|
|
open my $fh, ">>" => $file |
235
|
|
|
|
|
|
|
or croak "Cannot open '$file': $!"; |
236
|
|
|
|
|
|
|
|
237
|
0
|
|
|
|
|
|
binmode $fh; |
238
|
|
|
|
|
|
|
|
239
|
0
|
|
|
|
|
|
my $eof_ref = $self->with_eof_marker; |
240
|
0
|
|
|
|
|
|
print {$fh} $$eof_ref; |
|
0
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
|
242
|
0
|
0
|
|
|
|
|
close $fh |
243
|
|
|
|
|
|
|
or croak "Cannot write to '$file': $!"; |
244
|
|
|
|
|
|
|
} |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
__END__ |
247
|
|
|
|
|
|
|
|
248
|
|
|
|
|
|
|
=pod |
249
|
|
|
|
|
|
|
|
250
|
|
|
|
|
|
|
=encoding UTF-8 |
251
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
=head1 NAME |
253
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
App::Sandy::Seq - Base class to simulate seq entries |
255
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
=head1 VERSION |
257
|
|
|
|
|
|
|
|
258
|
|
|
|
|
|
|
version 0.24 |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
=head1 AUTHORS |
261
|
|
|
|
|
|
|
|
262
|
|
|
|
|
|
|
=over 4 |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
=item * |
265
|
|
|
|
|
|
|
|
266
|
|
|
|
|
|
|
Thiago L. A. Miller <tmiller@mochsl.org.br> |
267
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
=item * |
269
|
|
|
|
|
|
|
|
270
|
|
|
|
|
|
|
J. Leonel Buzzo <lbuzzo@mochsl.org.br> |
271
|
|
|
|
|
|
|
|
272
|
|
|
|
|
|
|
=item * |
273
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
Felipe R. C. dos Santos <fsantos@mochsl.org.br> |
275
|
|
|
|
|
|
|
|
276
|
|
|
|
|
|
|
=item * |
277
|
|
|
|
|
|
|
|
278
|
|
|
|
|
|
|
Helena B. Conceição <hconceicao@mochsl.org.br> |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
=item * |
281
|
|
|
|
|
|
|
|
282
|
|
|
|
|
|
|
Rodrigo Barreiro <rbarreiro@mochsl.org.br> |
283
|
|
|
|
|
|
|
|
284
|
|
|
|
|
|
|
=item * |
285
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
Gabriela Guardia <gguardia@mochsl.org.br> |
287
|
|
|
|
|
|
|
|
288
|
|
|
|
|
|
|
=item * |
289
|
|
|
|
|
|
|
|
290
|
|
|
|
|
|
|
Fernanda Orpinelli <forpinelli@mochsl.org.br> |
291
|
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
=item * |
293
|
|
|
|
|
|
|
|
294
|
|
|
|
|
|
|
Rafael Mercuri <rmercuri@mochsl.org.br> |
295
|
|
|
|
|
|
|
|
296
|
|
|
|
|
|
|
=item * |
297
|
|
|
|
|
|
|
|
298
|
|
|
|
|
|
|
Rodrigo Barreiro <rbarreiro@mochsl.org.br> |
299
|
|
|
|
|
|
|
|
300
|
|
|
|
|
|
|
=item * |
301
|
|
|
|
|
|
|
|
302
|
|
|
|
|
|
|
Pedro A. F. Galante <pgalante@mochsl.org.br> |
303
|
|
|
|
|
|
|
|
304
|
|
|
|
|
|
|
=back |
305
|
|
|
|
|
|
|
|
306
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
307
|
|
|
|
|
|
|
|
308
|
|
|
|
|
|
|
This software is Copyright (c) 2023 by Teaching and Research Institute from Sírio-Libanês Hospital. |
309
|
|
|
|
|
|
|
|
310
|
|
|
|
|
|
|
This is free software, licensed under: |
311
|
|
|
|
|
|
|
|
312
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
313
|
|
|
|
|
|
|
|
314
|
|
|
|
|
|
|
=cut |