line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package BioX::Workflow::Command::run::Utils::Samples; |
2
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
4250
|
use MooseX::App::Role; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
10
|
|
4
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
with 'BioX::Workflow::Command::run::Rules::Directives::Sample'; |
6
|
|
|
|
|
|
|
with 'BioX::Workflow::Command::run::Rules::Rules'; |
7
|
|
|
|
|
|
|
|
8
|
1
|
|
|
1
|
|
11312
|
use File::Find::Rule; |
|
1
|
|
|
|
|
8444
|
|
|
1
|
|
|
|
|
9
|
|
9
|
1
|
|
|
1
|
|
100
|
use File::Basename; |
|
1
|
|
|
|
|
4
|
|
|
1
|
|
|
|
|
69
|
|
10
|
1
|
|
|
1
|
|
7
|
use File::Glob; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
62
|
|
11
|
1
|
|
|
1
|
|
7
|
use List::Util qw(uniq); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
63
|
|
12
|
|
|
|
|
|
|
|
13
|
1
|
|
|
1
|
|
9
|
use Storable qw(dclone); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
44
|
|
14
|
1
|
|
|
1
|
|
6
|
use Path::Tiny; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
983
|
|
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
=head1 BioX::Workflow::Command::run::Utils::Samples |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
=head2 Variables |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
=head2 Subroutines |
22
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
=head3 get_samples |
24
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
Get basename of the files. Can add optional rules. |
26
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
sample.vcf.gz and sample.vcf would be sample if the sample_rule is (.vcf)$|(.vcf.gz)$ |
28
|
|
|
|
|
|
|
|
29
|
|
|
|
|
|
|
Also gets the full path to infiles |
30
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
Instead of doing |
32
|
|
|
|
|
|
|
|
33
|
|
|
|
|
|
|
foreach my $sample (@$self->samples){ |
34
|
|
|
|
|
|
|
dostuff |
35
|
|
|
|
|
|
|
} |
36
|
|
|
|
|
|
|
|
37
|
|
|
|
|
|
|
Could have |
38
|
|
|
|
|
|
|
|
39
|
|
|
|
|
|
|
foreach my $infile (@$self->infiles){ |
40
|
|
|
|
|
|
|
dostuff |
41
|
|
|
|
|
|
|
} |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
=cut |
44
|
|
|
|
|
|
|
|
45
|
|
|
|
|
|
|
sub get_samples { |
46
|
0
|
|
|
0
|
|
|
my $self = shift; |
47
|
|
|
|
|
|
|
|
48
|
|
|
|
|
|
|
#Stupid resample |
49
|
0
|
|
|
|
|
|
$self->get_global_keys; |
50
|
|
|
|
|
|
|
|
51
|
0
|
|
|
|
|
|
my $exists = $self->check_sample_exist; |
52
|
0
|
0
|
|
|
|
|
return if $exists; |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
#We need to evaluate the global_dirs incase the indir has a var |
55
|
|
|
|
|
|
|
#But we don't keep it around, because that would be madness |
56
|
0
|
|
|
|
|
|
my $attr = dclone( $self->global_attr ); |
57
|
0
|
0
|
|
|
|
|
if ( $attr->indir =~ m/\{\$/ ) { |
58
|
0
|
|
|
|
|
|
$attr->walk_process_data( $self->global_keys ); |
59
|
|
|
|
|
|
|
} |
60
|
|
|
|
|
|
|
|
61
|
0
|
|
|
|
|
|
my $text = $self->get_sample_rule; |
62
|
0
|
|
|
|
|
|
$self->find_sample_file_find_rule( $attr, $text ); |
63
|
0
|
|
|
|
|
|
$self->find_sample_glob($attr, $text); |
64
|
|
|
|
|
|
|
|
65
|
0
|
0
|
|
|
|
|
if ( $self->has_no_samples ) { |
66
|
0
|
|
|
|
|
|
$self->app_log->warn('No samples were found!'); |
67
|
0
|
|
|
|
|
|
$self->app_log->warn( |
68
|
|
|
|
|
|
|
"Indir: " . $attr->indir . "\tSearch: " . $text . "\n" ); |
69
|
|
|
|
|
|
|
} |
70
|
|
|
|
|
|
|
|
71
|
0
|
|
|
|
|
|
$self->remove_excluded_samples; |
72
|
0
|
|
|
|
|
|
$self->write_sample_meta; |
73
|
|
|
|
|
|
|
} |
74
|
|
|
|
|
|
|
|
75
|
|
|
|
|
|
|
sub remove_excluded_samples { |
76
|
0
|
|
|
0
|
|
|
my $self = shift; |
77
|
|
|
|
|
|
|
|
78
|
0
|
0
|
|
|
|
|
return unless $self->has_samples; |
79
|
0
|
0
|
|
|
|
|
return unless $self->has_exclude_samples; |
80
|
|
|
|
|
|
|
|
81
|
0
|
|
|
|
|
|
my %sample_hash = (); |
82
|
0
|
|
|
|
|
|
map { $sample_hash{$_} = 1 } @{$self->samples}; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
|
84
|
0
|
|
|
|
|
|
foreach my $sample ($self->all_exclude_samples){ |
85
|
0
|
|
|
|
|
|
delete $sample_hash{$sample}; |
86
|
|
|
|
|
|
|
} |
87
|
|
|
|
|
|
|
|
88
|
0
|
|
|
|
|
|
my @new_samples = keys %sample_hash; |
89
|
0
|
|
|
|
|
|
@new_samples = sort(@new_samples); |
90
|
0
|
|
|
|
|
|
$self->samples(\@new_samples); |
91
|
|
|
|
|
|
|
} |
92
|
|
|
|
|
|
|
|
93
|
|
|
|
|
|
|
sub find_sample_glob { |
94
|
0
|
|
|
0
|
|
|
my $self = shift; |
95
|
0
|
|
|
|
|
|
my $attr = shift; |
96
|
0
|
|
|
|
|
|
my $text = shift; |
97
|
|
|
|
|
|
|
|
98
|
0
|
0
|
|
|
|
|
return unless $attr->has_sample_glob; |
99
|
|
|
|
|
|
|
|
100
|
0
|
|
|
|
|
|
my @sample_files = glob( $attr->sample_glob ); |
101
|
0
|
0
|
|
|
|
|
return unless @sample_files; |
102
|
|
|
|
|
|
|
|
103
|
0
|
|
|
|
|
|
@sample_files = sort(@sample_files); |
104
|
0
|
0
|
|
|
|
|
$self->sample_files( \@sample_files ) if @sample_files; |
105
|
|
|
|
|
|
|
|
106
|
0
|
|
|
|
|
|
my @basename = map { $self->match_samples( $_, $text ) } @sample_files; |
|
0
|
|
|
|
|
|
|
107
|
0
|
0
|
|
|
|
|
if (@basename) { |
108
|
0
|
|
|
|
|
|
@basename = uniq(@basename); |
109
|
0
|
|
|
|
|
|
@basename = sort(@basename); |
110
|
0
|
|
|
|
|
|
$self->samples( \@basename ); |
111
|
|
|
|
|
|
|
} |
112
|
|
|
|
|
|
|
|
113
|
0
|
|
|
|
|
|
$self->global_attr->samples( dclone( $self->samples ) ); |
114
|
|
|
|
|
|
|
} |
115
|
|
|
|
|
|
|
|
116
|
|
|
|
|
|
|
sub get_sample_rule { |
117
|
0
|
|
|
0
|
|
|
my $self = shift; |
118
|
0
|
|
|
|
|
|
my $text; |
119
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
#Backwards compatibility |
121
|
|
|
|
|
|
|
#For both file_rule and sample_rule |
122
|
0
|
0
|
|
0
|
|
|
if ( $self->first_index_global_keys( sub { $_ eq 'file_rule' } ) != -1 ) { |
|
0
|
0
|
|
|
|
|
|
123
|
0
|
|
|
|
|
|
$text = $self->global_attr->sample_rule; |
124
|
|
|
|
|
|
|
} |
125
|
|
|
|
|
|
|
elsif ( |
126
|
0
|
|
|
0
|
|
|
$self->first_index_global_keys( sub { $_ eq 'sample_rule' } ) != -1 ) |
127
|
|
|
|
|
|
|
{ |
128
|
0
|
|
|
|
|
|
$text = $self->global_attr->sample_rule; |
129
|
|
|
|
|
|
|
} |
130
|
|
|
|
|
|
|
else { |
131
|
0
|
|
|
|
|
|
$text = $self->sample_rule; |
132
|
|
|
|
|
|
|
} |
133
|
|
|
|
|
|
|
} |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
sub find_sample_file_find_rule { |
136
|
0
|
|
|
0
|
|
|
my $self = shift; |
137
|
0
|
|
|
|
|
|
my $attr = shift; |
138
|
0
|
|
|
|
|
|
my $text = shift; |
139
|
|
|
|
|
|
|
|
140
|
0
|
0
|
|
|
|
|
return if $self->has_samples; |
141
|
|
|
|
|
|
|
|
142
|
0
|
|
|
|
|
|
my ( @whole, @basename, @sample_files, $find_sample_bydir ); |
143
|
|
|
|
|
|
|
|
144
|
0
|
|
|
|
|
|
$find_sample_bydir = 0; |
145
|
|
|
|
|
|
|
|
146
|
0
|
0
|
|
|
|
|
if ( $attr->find_sample_bydir ) { |
147
|
0
|
|
|
|
|
|
@whole = find( |
148
|
|
|
|
|
|
|
directory => name => qr/$text/, |
149
|
|
|
|
|
|
|
maxdepth => $attr->maxdepth, |
150
|
|
|
|
|
|
|
in => $attr->indir, |
151
|
|
|
|
|
|
|
extras => { follow => 1 }, |
152
|
|
|
|
|
|
|
); |
153
|
|
|
|
|
|
|
|
154
|
0
|
0
|
|
|
|
|
if (@whole) { |
155
|
0
|
0
|
|
|
|
|
if ( $whole[0] eq $attr->indir ) { |
156
|
0
|
|
|
|
|
|
shift(@whole); |
157
|
|
|
|
|
|
|
} |
158
|
|
|
|
|
|
|
} |
159
|
|
|
|
|
|
|
} |
160
|
|
|
|
|
|
|
else { |
161
|
0
|
|
|
|
|
|
@whole = find( |
162
|
|
|
|
|
|
|
file => name => qr/$text/, |
163
|
|
|
|
|
|
|
maxdepth => $attr->maxdepth, |
164
|
|
|
|
|
|
|
extras => { follow => 1 }, |
165
|
|
|
|
|
|
|
in => $attr->indir |
166
|
|
|
|
|
|
|
); |
167
|
|
|
|
|
|
|
} |
168
|
0
|
|
|
|
|
|
@basename = map { $self->match_samples( $_, $text ) } @whole; |
|
0
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
|
170
|
0
|
|
|
|
|
|
@sample_files = map { path($_)->absolute } @whole; |
|
0
|
|
|
|
|
|
|
171
|
0
|
|
|
|
|
|
@sample_files = sort(@sample_files); |
172
|
|
|
|
|
|
|
|
173
|
0
|
0
|
|
|
|
|
if (@basename) { |
174
|
0
|
|
|
|
|
|
@basename = uniq(@basename); |
175
|
0
|
|
|
|
|
|
@basename = sort(@basename); |
176
|
0
|
|
|
|
|
|
$self->samples( \@basename ); |
177
|
|
|
|
|
|
|
} |
178
|
0
|
0
|
|
|
|
|
$self->sample_files( \@sample_files ) if @sample_files; |
179
|
|
|
|
|
|
|
|
180
|
0
|
|
|
|
|
|
$self->global_attr->samples( dclone( $self->samples ) ); |
181
|
|
|
|
|
|
|
} |
182
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
sub check_sample_exist { |
184
|
0
|
|
|
0
|
|
|
my $self = shift; |
185
|
|
|
|
|
|
|
|
186
|
0
|
|
|
|
|
|
my $exists = 0; |
187
|
0
|
0
|
0
|
|
|
|
if ( $self->has_samples && !$self->resample ) { |
|
|
0
|
|
|
|
|
|
188
|
0
|
|
|
|
|
|
my (@samples) = $self->sorted_samples; |
189
|
0
|
|
|
|
|
|
$self->samples( \@samples ); |
190
|
|
|
|
|
|
|
## Fixes Issue #19 |
191
|
0
|
|
|
|
|
|
$self->global_attr->samples( \@samples ); |
192
|
0
|
|
|
|
|
|
$self->app_log->info('Samples passed in on command line.'); |
193
|
0
|
|
|
|
|
|
$exists = 1; |
194
|
|
|
|
|
|
|
} |
195
|
|
|
|
|
|
|
elsif ( $self->global_attr->has_samples ) { |
196
|
0
|
|
|
|
|
|
my (@samples) = @{ $self->global_attr->samples }; |
|
0
|
|
|
|
|
|
|
197
|
0
|
|
|
|
|
|
@samples = sort(@samples); |
198
|
0
|
|
|
|
|
|
$self->samples( \@samples ); |
199
|
0
|
|
|
|
|
|
$self->app_log->info('Samples were defined in the global key.'); |
200
|
0
|
|
|
|
|
|
$exists = 1; |
201
|
|
|
|
|
|
|
} |
202
|
|
|
|
|
|
|
|
203
|
0
|
0
|
|
|
|
|
$self->write_sample_meta if $exists; |
204
|
0
|
|
|
|
|
|
return $exists; |
205
|
|
|
|
|
|
|
} |
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
=head2 match_samples |
208
|
|
|
|
|
|
|
|
209
|
|
|
|
|
|
|
Match samples based on regex written in sample_rule |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
=cut |
212
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
sub match_samples { |
214
|
0
|
|
|
0
|
|
|
my $self = shift; |
215
|
0
|
|
|
|
|
|
my $file = shift; |
216
|
0
|
|
|
|
|
|
my $text = shift; |
217
|
|
|
|
|
|
|
|
218
|
0
|
0
|
|
|
|
|
if ( $text =~ m/\(/ ) { |
219
|
0
|
|
|
|
|
|
my @tmp = fileparse($file); |
220
|
0
|
|
|
|
|
|
my ($m) = $tmp[0] =~ qr/$text/; |
221
|
|
|
|
|
|
|
|
222
|
0
|
|
|
|
|
|
return $m; |
223
|
|
|
|
|
|
|
} |
224
|
|
|
|
|
|
|
else { |
225
|
0
|
|
|
|
|
|
my @tmp = fileparse($file); |
226
|
0
|
|
|
|
|
|
return $tmp[0]; |
227
|
|
|
|
|
|
|
} |
228
|
|
|
|
|
|
|
} |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
=head3 process_by_sample_outdir |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
Make sure indir/outdirs are named appropriated for samples when using by |
233
|
|
|
|
|
|
|
|
234
|
|
|
|
|
|
|
=cut |
235
|
|
|
|
|
|
|
|
236
|
|
|
|
|
|
|
sub process_by_sample_outdir { |
237
|
0
|
|
|
0
|
|
|
my $self = shift; |
238
|
0
|
|
|
|
|
|
my $sample = shift; |
239
|
|
|
|
|
|
|
|
240
|
0
|
|
|
|
|
|
my ( $tt, $key ); |
241
|
0
|
|
|
|
|
|
$tt = $self->outdir; |
242
|
0
|
|
|
|
|
|
$key = $self->key; |
243
|
0
|
|
|
|
|
|
$tt =~ s/$key/$sample\/$key/; |
244
|
0
|
|
|
|
|
|
$self->outdir($tt); |
245
|
0
|
|
|
|
|
|
$self->make_outdir; |
246
|
|
|
|
|
|
|
|
247
|
0
|
|
|
|
|
|
$tt = $self->indir; |
248
|
0
|
0
|
|
|
|
|
if ( $tt =~ m/\{\$self/ ) { |
|
|
0
|
|
|
|
|
|
249
|
0
|
|
|
|
|
|
$tt = "$tt/{\$sample}"; |
250
|
0
|
|
|
|
|
|
$self->indir($tt); |
251
|
|
|
|
|
|
|
} |
252
|
|
|
|
|
|
|
elsif ( $self->has_pkey ) { |
253
|
0
|
|
|
|
|
|
$key = $self->pkey; |
254
|
0
|
|
|
|
|
|
$tt =~ s/$key/$sample\/$key/; |
255
|
0
|
|
|
|
|
|
$self->indir($tt); |
256
|
|
|
|
|
|
|
} |
257
|
|
|
|
|
|
|
else { |
258
|
0
|
|
|
|
|
|
$tt = "$tt/$sample"; |
259
|
0
|
|
|
|
|
|
$self->indir($tt); |
260
|
|
|
|
|
|
|
} |
261
|
|
|
|
|
|
|
} |
262
|
|
|
|
|
|
|
|
263
|
|
|
|
|
|
|
1; |