line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package App::SimulateReads::Role::IO; |
2
|
|
|
|
|
|
|
# ABSTRACT: Input and output custom wrappers. |
3
|
|
|
|
|
|
|
|
4
|
6
|
|
|
6
|
|
2840
|
use App::SimulateReads::Base 'role'; |
|
6
|
|
|
|
|
8
|
|
|
6
|
|
|
|
|
51
|
|
5
|
6
|
|
|
6
|
|
1598
|
use PerlIO::gzip; |
|
6
|
|
|
|
|
2759
|
|
|
6
|
|
|
|
|
199
|
|
6
|
6
|
|
|
6
|
|
36
|
use Scalar::Util 'looks_like_number'; |
|
6
|
|
|
|
|
12
|
|
|
6
|
|
|
|
|
4256
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
our $VERSION = '0.06'; # VERSION |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
#=== CLASS METHOD ============================================================ |
11
|
|
|
|
|
|
|
# CLASS: My::Role::IO (Role) |
12
|
|
|
|
|
|
|
# METHOD: my_open_r |
13
|
|
|
|
|
|
|
# PARAMETERS: $file File |
14
|
|
|
|
|
|
|
# RETURNS: $fh IO::File |
15
|
|
|
|
|
|
|
# DESCRIPTION: Verify if the file is gzipped compressed and open it properly |
16
|
|
|
|
|
|
|
# THROWS: If open fails, throws an error |
17
|
|
|
|
|
|
|
# COMMENTS: none |
18
|
|
|
|
|
|
|
# SEE ALSO: n/a |
19
|
|
|
|
|
|
|
#=============================================================================== |
20
|
|
|
|
|
|
|
sub my_open_r { |
21
|
20
|
|
|
20
|
0
|
35
|
my ($self, $file) = @_; |
22
|
|
|
|
|
|
|
|
23
|
20
|
|
|
|
|
25
|
my $fh; |
24
|
20
|
50
|
|
|
|
65
|
my $mode = $file =~ /\.gz$/ ? "<:gzip" : "<"; |
25
|
|
|
|
|
|
|
|
26
|
20
|
50
|
|
|
|
590
|
open $fh, $mode => $file |
27
|
|
|
|
|
|
|
or croak "Not possible to read $file: $!"; |
28
|
|
|
|
|
|
|
|
29
|
20
|
|
|
|
|
70
|
return $fh; |
30
|
|
|
|
|
|
|
} ## --- end sub my_open_r |
31
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
#=== CLASS METHOD ============================================================ |
33
|
|
|
|
|
|
|
# CLASS: My::Role::IO (Role) |
34
|
|
|
|
|
|
|
# METHOD: my_open_w |
35
|
|
|
|
|
|
|
# PARAMETERS: $file Str, $is_gzipped Bool |
36
|
|
|
|
|
|
|
# RETURNS: $fh IO::File |
37
|
|
|
|
|
|
|
# DESCRIPTION: Opens for writing a file, gzipped or not |
38
|
|
|
|
|
|
|
# THROWS: If open fails, throws an error |
39
|
|
|
|
|
|
|
# COMMENTS: none |
40
|
|
|
|
|
|
|
# SEE ALSO: n/a |
41
|
|
|
|
|
|
|
#=============================================================================== |
42
|
|
|
|
|
|
|
sub my_open_w { |
43
|
11
|
|
|
11
|
0
|
85
|
my ($self, $file, $is_gzipped) = @_; |
44
|
|
|
|
|
|
|
|
45
|
11
|
|
|
|
|
32
|
my $fh; |
46
|
|
|
|
|
|
|
my $mode; |
47
|
|
|
|
|
|
|
|
48
|
11
|
50
|
|
|
|
106
|
if ($is_gzipped) { |
49
|
0
|
|
|
|
|
0
|
$mode = ">:gzip"; |
50
|
|
|
|
|
|
|
} else { |
51
|
11
|
|
|
|
|
64
|
$mode = ">"; |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
|
54
|
11
|
50
|
|
|
|
1615
|
open $fh, $mode => $file |
55
|
|
|
|
|
|
|
or croak "Not possible to create $file: $!"; |
56
|
|
|
|
|
|
|
|
57
|
11
|
|
|
|
|
108
|
return $fh; |
58
|
|
|
|
|
|
|
} ## --- end sub my_open_w |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
#=== CLASS METHOD ============================================================ |
61
|
|
|
|
|
|
|
# CLASS: My::Role::IO (Role) |
62
|
|
|
|
|
|
|
# METHOD: index_fasta |
63
|
|
|
|
|
|
|
# PARAMETERS: $fasta My:Fasta |
64
|
|
|
|
|
|
|
# RETURNS: HashRef[Hashref] |
65
|
|
|
|
|
|
|
# DESCRIPTION: Indexes a fasta file: id => (seq, size) |
66
|
|
|
|
|
|
|
# THROWS: It tries to validate the fasta file, if fails, throws an error |
67
|
|
|
|
|
|
|
# COMMENTS: none |
68
|
|
|
|
|
|
|
# SEE ALSO: n/a |
69
|
|
|
|
|
|
|
#=============================================================================== |
70
|
|
|
|
|
|
|
sub index_fasta { |
71
|
20
|
|
|
20
|
0
|
55
|
my ($self, $fasta) = @_; |
72
|
20
|
|
|
|
|
70
|
my $fh = $self->my_open_r($fasta); |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
# indexed_genome = ID => (seq, len) |
75
|
20
|
|
|
|
|
45
|
my %indexed_fasta; |
76
|
|
|
|
|
|
|
my $id; |
77
|
20
|
|
|
|
|
365
|
while (<$fh>) { |
78
|
860
|
|
|
|
|
1015
|
chomp; |
79
|
860
|
50
|
|
|
|
1260
|
next if /^;/; |
80
|
860
|
100
|
|
|
|
1285
|
if (/^>/) { |
81
|
100
|
|
|
|
|
225
|
my @fields = split /\|/; |
82
|
100
|
|
|
|
|
200
|
$id = (split / / => $fields[0])[0]; |
83
|
100
|
|
|
|
|
245
|
$id =~ s/^>//; |
84
|
100
|
|
|
|
|
340
|
$id = uc $id; |
85
|
|
|
|
|
|
|
} else { |
86
|
760
|
50
|
|
|
|
970
|
croak "Error reading fasta file '$fasta': Not defined id" |
87
|
|
|
|
|
|
|
unless defined $id; |
88
|
760
|
|
|
|
|
1810
|
$indexed_fasta{$id}{seq} .= $_; |
89
|
|
|
|
|
|
|
} |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
|
92
|
20
|
|
|
|
|
200
|
for (keys %indexed_fasta) { |
93
|
100
|
|
|
|
|
170
|
$indexed_fasta{$_}{size} = length $indexed_fasta{$_}{seq}; |
94
|
|
|
|
|
|
|
} |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
$fh->close |
97
|
20
|
50
|
|
|
|
145
|
or croak "Cannot close file $fasta: $!\n"; |
98
|
|
|
|
|
|
|
|
99
|
20
|
|
|
|
|
300
|
return \%indexed_fasta; |
100
|
|
|
|
|
|
|
} ## --- end sub index_fasta |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
#=== CLASS METHOD ============================================================ |
103
|
|
|
|
|
|
|
# CLASS: My::Role::IO |
104
|
|
|
|
|
|
|
# METHOD: index_weight_file |
105
|
|
|
|
|
|
|
# PARAMETERS: $weight_file My:File |
106
|
|
|
|
|
|
|
# RETURNS: $indexed_file Hashref[Int] |
107
|
|
|
|
|
|
|
# DESCRIPTION: It indexes a tab separated file with a seqid and its weight |
108
|
|
|
|
|
|
|
# THROWS: It tries to validate the file, if fails, the throws an exception |
109
|
|
|
|
|
|
|
# COMMENTS: none |
110
|
|
|
|
|
|
|
# SEE ALSO: n/a |
111
|
|
|
|
|
|
|
#=============================================================================== |
112
|
|
|
|
|
|
|
sub index_weight_file { |
113
|
0
|
|
|
0
|
0
|
|
my ($self, $weight_file) = @_; |
114
|
0
|
|
|
|
|
|
my $fh = $self->my_open_r($weight_file); |
115
|
0
|
|
|
|
|
|
my %indexed_file; |
116
|
0
|
|
|
|
|
|
my $line = 0; |
117
|
0
|
|
|
|
|
|
while (<$fh>) { |
118
|
0
|
|
|
|
|
|
$line++; |
119
|
0
|
|
|
|
|
|
chomp; |
120
|
0
|
0
|
|
|
|
|
next if /^\s*$/; |
121
|
0
|
|
|
|
|
|
my @fields = split /\t/; |
122
|
0
|
0
|
|
|
|
|
croak "Error parsing '$weight_file': seqid (first column) not found at line $line\n" unless defined $fields[0]; |
123
|
0
|
0
|
|
|
|
|
croak "Error parsing '$weight_file': weight (second column) not found at line $line\n" unless defined $fields[1]; |
124
|
0
|
0
|
|
|
|
|
croak "Error parsing '$weight_file': weight (second column) does not look like a number at line $line\n" if not looks_like_number($fields[1]); |
125
|
0
|
0
|
|
|
|
|
croak "Error parsing '$weight_file': weight (second column) lesser or equal to zero at line $line\n" if $fields[1] <= 0; |
126
|
0
|
|
|
|
|
|
$indexed_file{uc $fields[0]} = $fields[1]; |
127
|
|
|
|
|
|
|
} |
128
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
$fh->close |
130
|
0
|
0
|
|
|
|
|
or croak "Cannot close file $weight_file: $!\n"; |
131
|
|
|
|
|
|
|
|
132
|
0
|
|
|
|
|
|
return \%indexed_file; |
133
|
|
|
|
|
|
|
} ## --- end sub index_weight_file |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
__END__ |
136
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
=pod |
138
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=encoding UTF-8 |
140
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
=head1 NAME |
142
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
App::SimulateReads::Role::IO - Input and output custom wrappers. |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
=head1 VERSION |
146
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
version 0.06 |
148
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
=head1 AUTHOR |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
Thiago L. A. Miller <tmiller@mochsl.org.br> |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
154
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
This software is Copyright (c) 2017 by Teaching and Research Institute from SÃrio-Libanês Hospital. |
156
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
This is free software, licensed under: |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
The GNU General Public License, Version 3, June 2007 |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
=cut |