| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Proch::Seqfu; |
|
2
|
|
|
|
|
|
|
#ABSTRACT: Helper module to support Seqfu tools |
|
3
|
|
|
|
|
|
|
|
|
4
|
3
|
|
|
3
|
|
2144
|
use 5.014; |
|
|
3
|
|
|
|
|
20
|
|
|
5
|
3
|
|
|
3
|
|
15
|
use warnings; |
|
|
3
|
|
|
|
|
6
|
|
|
|
3
|
|
|
|
|
96
|
|
|
6
|
3
|
|
|
3
|
|
34
|
use Carp qw(confess); |
|
|
3
|
|
|
|
|
15
|
|
|
|
3
|
|
|
|
|
214
|
|
|
7
|
3
|
|
|
3
|
|
1273
|
use Data::Dumper; |
|
|
3
|
|
|
|
|
14463
|
|
|
|
3
|
|
|
|
|
192
|
|
|
8
|
3
|
|
|
3
|
|
1982
|
use Term::ANSIColor qw(:constants); |
|
|
3
|
|
|
|
|
28522
|
|
|
|
3
|
|
|
|
|
5758
|
|
|
9
|
|
|
|
|
|
|
require Exporter; |
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
$Proch::Seqfu::VERSION = '1.6.1'; |
|
12
|
|
|
|
|
|
|
$Proch::Seqfu::fu_linesize = 0; |
|
13
|
|
|
|
|
|
|
$Proch::Seqfu::fu_verbose = 0; |
|
14
|
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
our @ISA = qw(Exporter); |
|
16
|
|
|
|
|
|
|
our @EXPORT = qw(rc fu_printfasta fu_printfastq verbose has_seqfu seqfu_version); |
|
17
|
|
|
|
|
|
|
our @EXPORT_OK = qw($fu_linesize $fu_verbose); # symbols to export on request |
|
18
|
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
sub fu_printfasta { |
|
21
|
0
|
|
|
0
|
1
|
0
|
my ($name, $comment, $seq) = @_; |
|
22
|
0
|
|
|
|
|
0
|
my $print_comment = ''; |
|
23
|
0
|
0
|
|
|
|
0
|
if (defined $comment) { |
|
24
|
0
|
|
|
|
|
0
|
$print_comment = ' ' . $comment; |
|
25
|
|
|
|
|
|
|
} |
|
26
|
|
|
|
|
|
|
|
|
27
|
0
|
|
|
|
|
0
|
say '>', $name, $print_comment; |
|
28
|
0
|
|
|
|
|
0
|
print split_string($seq); |
|
29
|
|
|
|
|
|
|
} |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
sub fu_printfastq { |
|
32
|
0
|
|
|
0
|
1
|
0
|
my ($name, $comment, $seq, $qual) = @_; |
|
33
|
0
|
|
|
|
|
0
|
my $print_comment = ''; |
|
34
|
0
|
0
|
|
|
|
0
|
if (defined $comment) { |
|
35
|
0
|
|
|
|
|
0
|
$print_comment = ' ' . $comment; |
|
36
|
|
|
|
|
|
|
} |
|
37
|
|
|
|
|
|
|
|
|
38
|
0
|
|
|
|
|
0
|
say '@', $name, $print_comment; |
|
39
|
0
|
|
|
|
|
0
|
print split_string($seq) , "+\n", split_string($qual); |
|
40
|
|
|
|
|
|
|
} |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
# Print verbose info |
|
44
|
|
|
|
|
|
|
sub verbose { |
|
45
|
0
|
0
|
|
0
|
1
|
0
|
if ($Proch::Seqfu::fu_verbose) { |
|
46
|
0
|
|
|
|
|
0
|
say STDERR " - ", $_[0]; |
|
47
|
|
|
|
|
|
|
} |
|
48
|
|
|
|
|
|
|
} |
|
49
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
sub rc { |
|
52
|
1
|
|
|
1
|
1
|
84
|
my $sequence = reverse($_[0]); |
|
53
|
1
|
50
|
|
|
|
3
|
if (is_seq($sequence)) { |
|
54
|
1
|
50
|
|
|
|
4
|
if ($sequence =~ /U/i) { |
|
55
|
0
|
|
|
|
|
0
|
$sequence =~ tr/ACGURYSWKMBDHVacguryswkmbdhv/UGCAYRSWMKVHDBugcayrswmkvhdb/; |
|
56
|
|
|
|
|
|
|
} else { |
|
57
|
1
|
|
|
|
|
3
|
$sequence =~ tr/ACGTRYSWKMBDHVacgtryswkmbdhv/TGCAYRSWMKVHDBtgcayrswmkvhdb/; |
|
58
|
|
|
|
|
|
|
} |
|
59
|
1
|
|
|
|
|
3
|
return $sequence; |
|
60
|
|
|
|
|
|
|
} |
|
61
|
|
|
|
|
|
|
} |
|
62
|
|
|
|
|
|
|
|
|
63
|
|
|
|
|
|
|
|
|
64
|
|
|
|
|
|
|
sub is_seq { |
|
65
|
3
|
|
|
3
|
1
|
537
|
my $string = $_[0]; |
|
66
|
3
|
100
|
|
|
|
15
|
if ($string =~/[^ACGTRYSWKMBDHVNU]/i) { |
|
67
|
1
|
|
|
|
|
7
|
return 0; |
|
68
|
|
|
|
|
|
|
} else { |
|
69
|
2
|
|
|
|
|
10
|
return 1; |
|
70
|
|
|
|
|
|
|
} |
|
71
|
|
|
|
|
|
|
} |
|
72
|
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
sub split_string { |
|
75
|
0
|
|
|
0
|
1
|
0
|
my $input_string = $_[0]; |
|
76
|
0
|
|
|
|
|
0
|
my $formatted = ''; |
|
77
|
0
|
|
|
|
|
0
|
my $line_width = $Proch::Seqfu::fu_linesize; # change here |
|
78
|
0
|
0
|
|
|
|
0
|
return $input_string. "\n" unless ($line_width); |
|
79
|
0
|
|
|
|
|
0
|
for (my $i = 0; $i < length($input_string); $i += $line_width) { |
|
80
|
0
|
|
|
|
|
0
|
my $frag = substr($input_string, $i, $line_width); |
|
81
|
0
|
|
|
|
|
0
|
$formatted .= $frag."\n"; |
|
82
|
|
|
|
|
|
|
} |
|
83
|
0
|
|
|
|
|
0
|
return $formatted; |
|
84
|
|
|
|
|
|
|
} |
|
85
|
|
|
|
|
|
|
|
|
86
|
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
sub seqfu_version { |
|
88
|
2
|
|
|
2
|
1
|
88
|
my $cmd = ''; |
|
89
|
2
|
|
|
|
|
7
|
eval { |
|
90
|
2
|
|
|
|
|
4237
|
$cmd = `seqfu version`; |
|
91
|
|
|
|
|
|
|
}; |
|
92
|
2
|
|
|
|
|
190
|
chomp($cmd); |
|
93
|
2
|
50
|
|
|
|
78
|
if (length($@) > 0) { |
|
|
|
50
|
|
|
|
|
|
|
94
|
0
|
|
|
|
|
0
|
return -2; |
|
95
|
|
|
|
|
|
|
} elsif ($cmd =~/^(\d+)\.(\d+)\.?(\d+)?$/) { |
|
96
|
0
|
|
|
|
|
0
|
return $cmd; |
|
97
|
|
|
|
|
|
|
} else { |
|
98
|
2
|
|
|
|
|
72
|
return "-" . $cmd; |
|
99
|
|
|
|
|
|
|
} |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
|
|
103
|
|
|
|
|
|
|
sub has_seqfu { |
|
104
|
1
|
|
|
1
|
1
|
25
|
my $ver = seqfu_version(); |
|
105
|
1
|
50
|
|
|
|
29
|
if (substr($ver, 0, 1) eq '-') { |
|
|
|
0
|
|
|
|
|
|
|
106
|
1
|
|
|
|
|
20
|
return 0 |
|
107
|
|
|
|
|
|
|
} elsif (length($ver) > 0) { |
|
108
|
0
|
|
|
|
|
|
return 1 |
|
109
|
|
|
|
|
|
|
} else { |
|
110
|
0
|
|
|
|
|
|
return undef; |
|
111
|
|
|
|
|
|
|
} |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
1; |
|
114
|
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
=pod |
|
116
|
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
=encoding UTF-8 |
|
118
|
|
|
|
|
|
|
|
|
119
|
|
|
|
|
|
|
=head1 NAME |
|
120
|
|
|
|
|
|
|
|
|
121
|
|
|
|
|
|
|
Proch::Seqfu - Helper module to support Seqfu tools |
|
122
|
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
=head1 VERSION |
|
124
|
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
version 1.5.8 |
|
126
|
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
=head1 Proch::Seqfu |
|
128
|
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
a legacy module for Seqfu utilities |
|
130
|
|
|
|
|
|
|
|
|
131
|
|
|
|
|
|
|
=head2 fu_printfasta(name, comment, seq) |
|
132
|
|
|
|
|
|
|
|
|
133
|
|
|
|
|
|
|
This function prints a sequence in fasta format |
|
134
|
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
=head2 fu_printfastq(name, comment, seq, qual) |
|
136
|
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
This function prints a sequence in FASTQ format |
|
138
|
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
=head2 verbose(msg) |
|
140
|
|
|
|
|
|
|
|
|
141
|
|
|
|
|
|
|
Print a text if $fu_verbose is set to 1 |
|
142
|
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
=head2 rc(dna) |
|
144
|
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
Return the reverse complement of a string [degenerate base not supported] |
|
146
|
|
|
|
|
|
|
|
|
147
|
|
|
|
|
|
|
=head2 is_seq(seq) |
|
148
|
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
Check if a string is a DNA sequence, including degenerate chars. |
|
150
|
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
=head2 split_string(dna) |
|
152
|
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
Add newlines using $Proch::Seqfu::fu_linesize as line width |
|
154
|
|
|
|
|
|
|
|
|
155
|
|
|
|
|
|
|
=head2 seqfu_version() |
|
156
|
|
|
|
|
|
|
|
|
157
|
|
|
|
|
|
|
Check if a `seqfu` binary is present and returns its version if found. |
|
158
|
|
|
|
|
|
|
Note this will require SeqFu > 1.13 |
|
159
|
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
=head2 has_seqfu() |
|
161
|
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
If SeqFu is detected returns 1, 0 otherwise, I when detection of SeqFu version fails. |
|
163
|
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
=head1 AUTHOR |
|
165
|
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
Andrea Telatin |
|
167
|
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
|
169
|
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
This software is Copyright (c) 2018-2023 by Andrea Telatin. |
|
171
|
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
This is free software, licensed under: |
|
173
|
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
The MIT (X11) License |
|
175
|
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
=cut |
|
177
|
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
__END__ |