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
|
|
4459
|
use App::Sandy::Base 'class'; |
|
6
|
|
|
|
|
17
|
|
|
6
|
|
|
|
|
47
|
|
5
|
6
|
|
|
6
|
|
3408
|
use App::Sandy::Quality; |
|
6
|
|
|
|
|
2683
|
|
|
6
|
|
|
|
|
292
|
|
6
|
|
|
|
|
|
|
|
7
|
6
|
|
|
6
|
|
55
|
use constant NUM_TRIES => 1000; |
|
6
|
|
|
|
|
66
|
|
|
6
|
|
|
|
|
7890
|
|
8
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
with qw{ |
10
|
|
|
|
|
|
|
App::Sandy::Role::RunTimeTemplate |
11
|
|
|
|
|
|
|
App::Sandy::Role::Template::Fastq |
12
|
|
|
|
|
|
|
App::Sandy::Role::Template::Sam |
13
|
|
|
|
|
|
|
App::Sandy::Role::RNorm |
14
|
|
|
|
|
|
|
}; |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
our $VERSION = '0.22'; # VERSION |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
has 'format' => ( |
19
|
|
|
|
|
|
|
is => 'ro', |
20
|
|
|
|
|
|
|
isa => 'My:Format', |
21
|
|
|
|
|
|
|
required => 1 |
22
|
|
|
|
|
|
|
); |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
has 'template_id' => ( |
25
|
|
|
|
|
|
|
is => 'ro', |
26
|
|
|
|
|
|
|
isa => 'Str', |
27
|
|
|
|
|
|
|
required => 1 |
28
|
|
|
|
|
|
|
); |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
has 'sequencing_error' => ( |
31
|
|
|
|
|
|
|
is => 'ro', |
32
|
|
|
|
|
|
|
isa => 'My:NumHS', |
33
|
|
|
|
|
|
|
required => 1 |
34
|
|
|
|
|
|
|
); |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
has 'quality_profile' => ( |
37
|
|
|
|
|
|
|
is => 'ro', |
38
|
|
|
|
|
|
|
isa => 'My:QualityP', |
39
|
|
|
|
|
|
|
required => 1, |
40
|
|
|
|
|
|
|
coerce => 1 |
41
|
|
|
|
|
|
|
); |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
has 'read_mean' => ( |
44
|
|
|
|
|
|
|
is => 'ro', |
45
|
|
|
|
|
|
|
isa => 'My:IntGt0', |
46
|
|
|
|
|
|
|
required => 1 |
47
|
|
|
|
|
|
|
); |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
has 'read_stdd' => ( |
50
|
|
|
|
|
|
|
is => 'ro', |
51
|
|
|
|
|
|
|
isa => 'My:IntGe0', |
52
|
|
|
|
|
|
|
required => 1 |
53
|
|
|
|
|
|
|
); |
54
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
has '_template_id' => ( |
56
|
|
|
|
|
|
|
traits => ['Code'], |
57
|
|
|
|
|
|
|
is => 'ro', |
58
|
|
|
|
|
|
|
isa => 'CodeRef', |
59
|
|
|
|
|
|
|
builder => '_build_template_id', |
60
|
|
|
|
|
|
|
lazy_build => 1, |
61
|
|
|
|
|
|
|
handles => { |
62
|
|
|
|
|
|
|
_gen_id => 'execute' |
63
|
|
|
|
|
|
|
} |
64
|
|
|
|
|
|
|
); |
65
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
has '_template_seq' => ( |
67
|
|
|
|
|
|
|
traits => ['Code'], |
68
|
|
|
|
|
|
|
is => 'ro', |
69
|
|
|
|
|
|
|
isa => 'CodeRef', |
70
|
|
|
|
|
|
|
builder => '_build_template_seq', |
71
|
|
|
|
|
|
|
lazy_build => 1, |
72
|
|
|
|
|
|
|
handles => { |
73
|
|
|
|
|
|
|
_gen_seq => 'execute' |
74
|
|
|
|
|
|
|
} |
75
|
|
|
|
|
|
|
); |
76
|
|
|
|
|
|
|
|
77
|
|
|
|
|
|
|
has '_info' => ( |
78
|
|
|
|
|
|
|
traits => ['Hash'], |
79
|
|
|
|
|
|
|
is => 'ro', |
80
|
|
|
|
|
|
|
isa => 'HashRef[Str]', |
81
|
|
|
|
|
|
|
builder => '_build_info', |
82
|
|
|
|
|
|
|
lazy_build => 1, |
83
|
|
|
|
|
|
|
handles => { |
84
|
|
|
|
|
|
|
_set_info => 'set', |
85
|
|
|
|
|
|
|
_get_info => 'get' |
86
|
|
|
|
|
|
|
} |
87
|
|
|
|
|
|
|
); |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
has '_read_size' => ( |
90
|
|
|
|
|
|
|
traits => ['Code'], |
91
|
|
|
|
|
|
|
is => 'ro', |
92
|
|
|
|
|
|
|
isa => 'CodeRef', |
93
|
|
|
|
|
|
|
builder => '_build_read_size', |
94
|
|
|
|
|
|
|
lazy_build => 1, |
95
|
|
|
|
|
|
|
handles => { |
96
|
|
|
|
|
|
|
_get_read_size => 'execute' |
97
|
|
|
|
|
|
|
} |
98
|
|
|
|
|
|
|
); |
99
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
has '_quality' => ( |
101
|
|
|
|
|
|
|
is => 'ro', |
102
|
|
|
|
|
|
|
isa => 'App::Sandy::Quality', |
103
|
|
|
|
|
|
|
builder => '_build_quality', |
104
|
|
|
|
|
|
|
lazy_build => 1, |
105
|
|
|
|
|
|
|
handles => ['gen_quality'] |
106
|
|
|
|
|
|
|
); |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
has '_sym_table' => ( |
109
|
|
|
|
|
|
|
is => 'ro', |
110
|
|
|
|
|
|
|
isa => 'HashRef[Str]', |
111
|
|
|
|
|
|
|
builder => '_build_sym_table', |
112
|
|
|
|
|
|
|
lazy_build => 1 |
113
|
|
|
|
|
|
|
); |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
sub BUILD { |
116
|
60
|
|
|
60
|
0
|
125
|
my $self = shift; |
117
|
|
|
|
|
|
|
## Just to ensure that the lazy attributes are built before &new returns |
118
|
60
|
|
|
|
|
1740
|
$self->_quality; |
119
|
|
|
|
|
|
|
} |
120
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
sub _build_sym_table { |
122
|
14
|
|
|
14
|
|
660
|
my $sym_table = { |
123
|
|
|
|
|
|
|
'%q' => '$info->{quality_profile}', |
124
|
|
|
|
|
|
|
'%m' => '$info->{read_mean}', |
125
|
|
|
|
|
|
|
'%d' => '$info->{read_stdd}', |
126
|
|
|
|
|
|
|
'%r' => '$info->{read_size}', |
127
|
|
|
|
|
|
|
'%e' => '$info->{sequencing_error}', |
128
|
|
|
|
|
|
|
'%c' => '$info->{seq_id}', |
129
|
|
|
|
|
|
|
'%C' => '$info->{seq_id_type}', |
130
|
|
|
|
|
|
|
'%a' => '$info->{start_ref}', |
131
|
|
|
|
|
|
|
'%b' => '$info->{end_ref}', |
132
|
|
|
|
|
|
|
'%t' => '$info->{start}', |
133
|
|
|
|
|
|
|
'%n' => '$info->{end}', |
134
|
|
|
|
|
|
|
'%i' => '$info->{instrument}', |
135
|
|
|
|
|
|
|
'%I' => '$info->{id}', |
136
|
|
|
|
|
|
|
'%R' => '$info->{read}', |
137
|
|
|
|
|
|
|
'%U' => '$info->{num}', |
138
|
|
|
|
|
|
|
'%s' => '$info->{strand}', |
139
|
|
|
|
|
|
|
'%x' => '$info->{error}', |
140
|
|
|
|
|
|
|
'%v' => '$info->{var}' |
141
|
|
|
|
|
|
|
}; |
142
|
|
|
|
|
|
|
|
143
|
14
|
|
|
|
|
238
|
return $sym_table; |
144
|
|
|
|
|
|
|
} |
145
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
sub _build_template_id { |
147
|
14
|
|
|
14
|
|
35
|
my $self = shift; |
148
|
14
|
|
|
|
|
501
|
my $sym_table = $self->_sym_table; |
149
|
14
|
|
|
|
|
617
|
return $self->with_compile_template($self->template_id, 'info', $sym_table); |
150
|
|
|
|
|
|
|
} |
151
|
|
|
|
|
|
|
|
152
|
|
|
|
|
|
|
sub _build_template_seq { |
153
|
29
|
|
|
29
|
|
54
|
my $self = shift; |
154
|
|
|
|
|
|
|
|
155
|
29
|
|
|
|
|
912
|
my $format = $self->format; |
156
|
29
|
|
|
|
|
54
|
my $gen_seq; |
157
|
|
|
|
|
|
|
|
158
|
29
|
50
|
|
|
|
239
|
if ($format =~ /fastq/) { |
|
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
159
|
29
|
|
|
2325
|
|
207
|
$gen_seq = sub { $self->with_fastq_template(@_) }; |
|
2325
|
|
|
|
|
12381
|
|
160
|
|
|
|
|
|
|
} elsif ($format eq 'sam') { |
161
|
0
|
|
|
0
|
|
0
|
$gen_seq = sub { $self->with_sam_align_template(@_) }; |
|
0
|
|
|
|
|
0
|
|
162
|
|
|
|
|
|
|
} elsif ($format eq 'bam') { |
163
|
0
|
|
|
0
|
|
0
|
$gen_seq = sub { $self->with_bam_align_template(@_) }; |
|
0
|
|
|
|
|
0
|
|
164
|
|
|
|
|
|
|
} else { |
165
|
0
|
|
|
|
|
0
|
croak "No valid format: '$format'"; |
166
|
|
|
|
|
|
|
} |
167
|
|
|
|
|
|
|
|
168
|
29
|
|
|
|
|
960
|
return $gen_seq; |
169
|
|
|
|
|
|
|
} |
170
|
|
|
|
|
|
|
|
171
|
|
|
|
|
|
|
sub _build_info { |
172
|
14
|
|
|
14
|
|
685
|
my $self = shift; |
173
|
|
|
|
|
|
|
|
174
|
14
|
|
|
|
|
557
|
my $info = { |
175
|
|
|
|
|
|
|
instrument => 'SR', |
176
|
|
|
|
|
|
|
quality_profile => $self->quality_profile, |
177
|
|
|
|
|
|
|
sequencing_error => $self->sequencing_error, |
178
|
|
|
|
|
|
|
read_mean => $self->read_mean, |
179
|
|
|
|
|
|
|
read_stdd => $self->read_stdd |
180
|
|
|
|
|
|
|
}; |
181
|
|
|
|
|
|
|
|
182
|
14
|
|
|
|
|
276
|
return $info; |
183
|
|
|
|
|
|
|
} |
184
|
|
|
|
|
|
|
|
185
|
|
|
|
|
|
|
sub _build_read_size { |
186
|
14
|
|
|
14
|
|
98
|
my $self = shift; |
187
|
14
|
|
|
|
|
58
|
my $fun; |
188
|
|
|
|
|
|
|
|
189
|
14
|
50
|
|
|
|
640
|
if ($self->read_stdd == 0) { |
190
|
14
|
|
|
1730
|
|
181
|
$fun = sub { $self->read_mean }; |
|
1730
|
|
|
|
|
41963
|
|
191
|
|
|
|
|
|
|
} else { |
192
|
|
|
|
|
|
|
$fun = sub { |
193
|
0
|
|
|
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 = $self->with_random_half_normal($self->read_mean, |
200
|
|
|
|
|
|
|
$self->read_stdd) |
201
|
|
|
|
|
|
|
} |
202
|
0
|
|
|
|
|
0
|
return $size; |
203
|
0
|
|
|
|
|
0
|
}; |
204
|
|
|
|
|
|
|
} |
205
|
|
|
|
|
|
|
|
206
|
14
|
|
|
|
|
548
|
return $fun; |
207
|
|
|
|
|
|
|
} |
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
sub _build_quality { |
210
|
60
|
|
|
60
|
|
100
|
my $self = shift; |
211
|
60
|
|
|
|
|
1640
|
App::Sandy::Quality->new( |
212
|
|
|
|
|
|
|
quality_profile => $self->quality_profile |
213
|
|
|
|
|
|
|
); |
214
|
|
|
|
|
|
|
} |
215
|
|
|
|
|
|
|
|
216
|
|
|
|
|
|
|
sub gen_sam_header { |
217
|
0
|
|
|
0
|
0
|
|
my ($self, $argv) = @_; |
218
|
0
|
|
|
|
|
|
my $format = $self->format; |
219
|
0
|
|
|
|
|
|
my $header; |
220
|
|
|
|
|
|
|
|
221
|
0
|
0
|
|
|
|
|
if ($format eq 'sam') { |
|
|
0
|
|
|
|
|
|
222
|
0
|
|
|
|
|
|
$header = $self->with_sam_header_template($argv); |
223
|
|
|
|
|
|
|
} elsif ($format eq 'bam') { |
224
|
0
|
|
|
|
|
|
$header = $self->with_bam_header_template($argv); |
225
|
|
|
|
|
|
|
} else { |
226
|
0
|
|
|
|
|
|
croak "No valid format: '$format'"; |
227
|
|
|
|
|
|
|
} |
228
|
|
|
|
|
|
|
|
229
|
0
|
|
|
|
|
|
return $header; |
230
|
|
|
|
|
|
|
} |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
sub gen_eof_marker { |
233
|
0
|
|
|
0
|
0
|
|
my ($self, $file) = @_; |
234
|
|
|
|
|
|
|
|
235
|
0
|
0
|
|
|
|
|
open my $fh, ">>" => $file |
236
|
|
|
|
|
|
|
or croak "Cannot open '$file': $!"; |
237
|
|
|
|
|
|
|
|
238
|
0
|
|
|
|
|
|
binmode $fh; |
239
|
|
|
|
|
|
|
|
240
|
0
|
|
|
|
|
|
my $eof_ref = $self->with_eof_marker; |
241
|
0
|
|
|
|
|
|
print {$fh} $$eof_ref; |
|
0
|
|
|
|
|
|
|
242
|
|
|
|
|
|
|
|
243
|
0
|
0
|
|
|
|
|
close $fh |
244
|
|
|
|
|
|
|
or croak "Cannot write to '$file': $!"; |
245
|
|
|
|
|
|
|
} |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
__END__ |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
=pod |
250
|
|
|
|
|
|
|
|
251
|
|
|
|
|
|
|
=encoding UTF-8 |
252
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
=head1 NAME |
254
|
|
|
|
|
|
|
|
255
|
|
|
|
|
|
|
App::Sandy::Seq - Base class to simulate seq entries |
256
|
|
|
|
|
|
|
|
257
|
|
|
|
|
|
|
=head1 VERSION |
258
|
|
|
|
|
|
|
|
259
|
|
|
|
|
|
|
version 0.22 |
260
|
|
|
|
|
|
|
|
261
|
|
|
|
|
|
|
=head1 AUTHORS |
262
|
|
|
|
|
|
|
|
263
|
|
|
|
|
|
|
=over 4 |
264
|
|
|
|
|
|
|
|
265
|
|
|
|
|
|
|
=item * |
266
|
|
|
|
|
|
|
|
267
|
|
|
|
|
|
|
Thiago L. A. Miller <tmiller@mochsl.org.br> |
268
|
|
|
|
|
|
|
|
269
|
|
|
|
|
|
|
=item * |
270
|
|
|
|
|
|
|
|
271
|
|
|
|
|
|
|
J. Leonel Buzzo <lbuzzo@mochsl.org.br> |
272
|
|
|
|
|
|
|
|
273
|
|
|
|
|
|
|
=item * |
274
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
Felipe R. C. dos Santos <fsantos@mochsl.org.br> |
276
|
|
|
|
|
|
|
|
277
|
|
|
|
|
|
|
=item * |
278
|
|
|
|
|
|
|
|
279
|
|
|
|
|
|
|
Helena B. Conceição <hconceicao@mochsl.org.br> |
280
|
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
=item * |
282
|
|
|
|
|
|
|
|
283
|
|
|
|
|
|
|
Gabriela Guardia <gguardia@mochsl.org.br> |
284
|
|
|
|
|
|
|
|
285
|
|
|
|
|
|
|
=item * |
286
|
|
|
|
|
|
|
|
287
|
|
|
|
|
|
|
Fernanda Orpinelli <forpinelli@mochsl.org.br> |
288
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
=item * |
290
|
|
|
|
|
|
|
|
291
|
|
|
|
|
|
|
Pedro A. F. Galante <pgalante@mochsl.org.br> |
292
|
|
|
|
|
|
|
|
293
|
|
|
|
|
|
|
=back |
294
|
|
|
|
|
|
|
|
295
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
296
|
|
|
|
|
|
|
|
297
|
|
|
|
|
|
|
This software is Copyright (c) 2018 by Teaching and Research Institute from SÃrio-Libanês Hospital. |
298
|
|
|
|
|
|
|
|
299
|
|
|
|
|
|
|
This is free software, licensed under: |
300
|
|
|
|
|
|
|
|
301
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
302
|
|
|
|
|
|
|
|
303
|
|
|
|
|
|
|
=cut |