| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package CracTools::GenomeMask; |
|
2
|
|
|
|
|
|
|
{ |
|
3
|
|
|
|
|
|
|
$CracTools::GenomeMask::DIST = 'CracTools'; |
|
4
|
|
|
|
|
|
|
} |
|
5
|
|
|
|
|
|
|
# ABSTRACT: A bit vector mask over the whole genome |
|
6
|
|
|
|
|
|
|
$CracTools::GenomeMask::VERSION = '1.25'; |
|
7
|
1
|
|
|
1
|
|
13200
|
use strict; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
23
|
|
|
8
|
1
|
|
|
1
|
|
2
|
use warnings; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
19
|
|
|
9
|
|
|
|
|
|
|
|
|
10
|
1
|
|
|
1
|
|
341
|
use CracTools::BitVector; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
36
|
|
|
11
|
1
|
|
|
1
|
|
373
|
use CracTools::Utils; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
24
|
|
|
12
|
1
|
|
|
1
|
|
4
|
use Carp; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
689
|
|
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
#You can also define is the genome should be consider as uniquely stranded or as double stranded |
|
15
|
|
|
|
|
|
|
#with the C argument. |
|
16
|
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
sub new { |
|
19
|
3
|
|
|
3
|
1
|
1309
|
my $class = shift; |
|
20
|
|
|
|
|
|
|
|
|
21
|
3
|
|
|
|
|
7
|
my %args = @_; |
|
22
|
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
# the genome mask is not stranded by default |
|
24
|
|
|
|
|
|
|
#my $is_stranded = defined $args{is_stranded}? $args{is_stranded} : 0; |
|
25
|
3
|
50
|
|
|
|
8
|
my $verbose = defined $args{verbose}? $args{verbose} : 0; |
|
26
|
|
|
|
|
|
|
|
|
27
|
3
|
|
|
|
|
3
|
my %bit_vectors; |
|
28
|
|
|
|
|
|
|
|
|
29
|
3
|
100
|
|
|
|
19
|
if(defined $args{genome}) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
|
|
31
|
1
|
|
|
|
|
2
|
foreach my $chr (keys %{$args{genome}}) { |
|
|
1
|
|
|
|
|
4
|
|
|
32
|
2
|
50
|
|
|
|
5
|
if(!defined $bit_vectors{$chr}) { |
|
33
|
2
|
|
|
|
|
8
|
$bit_vectors{$chr} = CracTools::BitVector->new($args{genome}->{$chr}); |
|
34
|
|
|
|
|
|
|
} else { |
|
35
|
0
|
|
|
|
|
0
|
croak "Multiple definition of sequence $chr inf the genome lengths"; |
|
36
|
|
|
|
|
|
|
} |
|
37
|
|
|
|
|
|
|
} |
|
38
|
|
|
|
|
|
|
} elsif(defined $args{crac_index_conf}) { |
|
39
|
1
|
50
|
|
|
|
4
|
print STDERR "Creating GenomeMask from crac index conf file : $args{crac_index_conf}\n" if $verbose; |
|
40
|
1
|
|
|
|
|
5
|
my $conf_fh = CracTools::Utils::getReadingFileHandle($args{crac_index_conf}); |
|
41
|
1
|
|
|
|
|
11
|
my $nb_chr = <$conf_fh>; |
|
42
|
1
|
|
|
|
|
2
|
my $nb_chr_found = 0; |
|
43
|
1
|
|
|
|
|
3
|
while(<$conf_fh>) { |
|
44
|
24
|
|
|
|
|
12
|
my $chr = $_; |
|
45
|
24
|
|
|
|
|
19
|
chomp $chr; |
|
46
|
24
|
|
|
|
|
18
|
my $chr_length = <$conf_fh>; |
|
47
|
24
|
|
|
|
|
13
|
chomp $chr_length; |
|
48
|
24
|
50
|
|
|
|
22
|
if(defined $chr_length) { |
|
49
|
24
|
50
|
|
|
|
25
|
print STDERR "\tCreating bitvecor for chr $chr of length $chr_length\n" if $verbose; |
|
50
|
24
|
|
|
|
|
35
|
$bit_vectors{$chr} = CracTools::BitVector->new($chr_length); |
|
51
|
24
|
|
|
|
|
57
|
$nb_chr_found++; |
|
52
|
|
|
|
|
|
|
} else { |
|
53
|
0
|
|
|
|
|
0
|
croak "Missing genome length for chromosome $chr"; |
|
54
|
|
|
|
|
|
|
} |
|
55
|
|
|
|
|
|
|
} |
|
56
|
1
|
50
|
|
|
|
9
|
croak "There is less chromosome found ($nb_chr_found) in $args{crac_index_conf} than expected ($nb_chr)" if $nb_chr_found < $nb_chr; |
|
57
|
|
|
|
|
|
|
} elsif(defined $args{sam_reader}) { |
|
58
|
1
|
|
|
|
|
5
|
my $refseq_lengths = $args{sam_reader}->allRefSeqLengths(); |
|
59
|
1
|
|
|
|
|
2
|
foreach my $chr (keys %{$refseq_lengths}) { |
|
|
1
|
|
|
|
|
4
|
|
|
60
|
24
|
|
|
|
|
36
|
$bit_vectors{$chr} = CracTools::BitVector->new($refseq_lengths->{$chr}); |
|
61
|
|
|
|
|
|
|
} |
|
62
|
|
|
|
|
|
|
} else { |
|
63
|
0
|
|
|
|
|
0
|
croak "There is no valid argument to extract the chromosomes names and length"; |
|
64
|
|
|
|
|
|
|
} |
|
65
|
|
|
|
|
|
|
|
|
66
|
3
|
|
|
|
|
13
|
my $self = bless { |
|
67
|
|
|
|
|
|
|
bit_vectors => \%bit_vectors, |
|
68
|
|
|
|
|
|
|
#is_stranded => $is_stranded, |
|
69
|
|
|
|
|
|
|
}, $class; |
|
70
|
|
|
|
|
|
|
|
|
71
|
3
|
|
|
|
|
7
|
return $self; |
|
72
|
|
|
|
|
|
|
} |
|
73
|
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
#=head2 isStranded |
|
75
|
|
|
|
|
|
|
# |
|
76
|
|
|
|
|
|
|
# Description : Return true is genomeMask is double stranded |
|
77
|
|
|
|
|
|
|
# |
|
78
|
|
|
|
|
|
|
#=cut |
|
79
|
|
|
|
|
|
|
# |
|
80
|
|
|
|
|
|
|
#sub isStranded { |
|
81
|
|
|
|
|
|
|
# my $self = shift; |
|
82
|
|
|
|
|
|
|
# return $self->{is_stranded}; |
|
83
|
|
|
|
|
|
|
#} |
|
84
|
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
#Arg [2] : Integer (1,-1) - strand |
|
86
|
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
sub getBitvector { |
|
88
|
27
|
|
|
27
|
1
|
18
|
my $self = shift; |
|
89
|
27
|
|
|
|
|
16
|
my $chr = shift; |
|
90
|
|
|
|
|
|
|
#my $strand = shift; |
|
91
|
27
|
50
|
|
|
|
42
|
if(defined $self->{bit_vectors}->{$chr}) { |
|
92
|
27
|
|
|
|
|
44
|
return $self->{bit_vectors}->{$chr}; |
|
93
|
|
|
|
|
|
|
} else { |
|
94
|
0
|
|
|
|
|
0
|
carp "There is no bitvector for sequence $chr in the genome mask"; |
|
95
|
0
|
|
|
|
|
0
|
return undef; |
|
96
|
|
|
|
|
|
|
} |
|
97
|
|
|
|
|
|
|
} |
|
98
|
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
sub getChrLength { |
|
101
|
3
|
|
|
3
|
1
|
41
|
my $self = shift; |
|
102
|
3
|
|
|
|
|
4
|
my $chr = shift; |
|
103
|
3
|
|
|
|
|
4
|
my $bv = $self->getBitvector($chr); |
|
104
|
3
|
50
|
|
|
|
10
|
return defined $bv? $bv->length : undef; |
|
105
|
|
|
|
|
|
|
} |
|
106
|
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
sub setPos { |
|
109
|
6
|
|
|
6
|
1
|
7
|
my ($self,$chr,$pos) = @_; |
|
110
|
6
|
|
|
|
|
7
|
my $bv = $self->getBitvector($chr); |
|
111
|
6
|
50
|
|
|
|
15
|
$bv->set($pos) if defined $bv; |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
sub setRegion { |
|
116
|
1
|
|
|
1
|
1
|
3
|
my ($self,$chr,$start,$end) = @_; |
|
117
|
1
|
|
|
|
|
4
|
for(my $i = $start; $i <= $end; $i++) { |
|
118
|
4
|
|
|
|
|
5
|
$self->setPos($chr,$i); |
|
119
|
|
|
|
|
|
|
} |
|
120
|
|
|
|
|
|
|
} |
|
121
|
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
sub getPos { |
|
124
|
5
|
|
|
5
|
1
|
7
|
my ($self,$chr,$pos) = @_; |
|
125
|
5
|
|
|
|
|
9
|
my $bv = $self->getBitvector($chr); |
|
126
|
5
|
50
|
|
|
|
16
|
return $bv->get($pos) if defined $bv; |
|
127
|
|
|
|
|
|
|
} |
|
128
|
|
|
|
|
|
|
|
|
129
|
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
sub getPosSetInRegion { |
|
131
|
2
|
|
|
2
|
1
|
3
|
my ($self,$chr,$start,$end) = @_; |
|
132
|
2
|
|
|
|
|
3
|
my $bv = $self->getBitvector($chr); |
|
133
|
2
|
|
|
|
|
2
|
my @pos; |
|
134
|
2
|
50
|
|
|
|
4
|
if(defined $bv) { |
|
135
|
2
|
|
|
|
|
6
|
for(my $i = $start; $i <= $end; $i++) { |
|
136
|
13
|
100
|
|
|
|
19
|
push(@pos,$i) if $bv->get($i) == 1; |
|
137
|
|
|
|
|
|
|
} |
|
138
|
|
|
|
|
|
|
} |
|
139
|
2
|
|
|
|
|
6
|
return \@pos; |
|
140
|
|
|
|
|
|
|
} |
|
141
|
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
sub getNbBitsSetInRegion { |
|
144
|
1
|
|
|
1
|
1
|
2
|
my $self = shift; |
|
145
|
1
|
|
|
|
|
1
|
return scalar @{$self->getPosSetInRegion(@_)}; |
|
|
1
|
|
|
|
|
3
|
|
|
146
|
|
|
|
|
|
|
} |
|
147
|
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
|
|
149
|
|
|
|
|
|
|
sub rank { |
|
150
|
2
|
|
|
2
|
1
|
957
|
my ($self,$chr,$pos) = @_; |
|
151
|
2
|
|
|
|
|
3
|
my $cumulated_bits = 0; |
|
152
|
2
|
|
|
|
|
2
|
my $i = 0; |
|
153
|
2
|
|
|
|
|
2
|
my @chr_sorted = sort keys %{$self->{bit_vectors}}; |
|
|
2
|
|
|
|
|
9
|
|
|
154
|
2
|
|
|
|
|
7
|
while($chr_sorted[$i] ne $chr) { |
|
155
|
2
|
|
|
|
|
7
|
$cumulated_bits += $self->getBitvector($chr_sorted[$i])->nb_set; |
|
156
|
2
|
|
|
|
|
5
|
$i++; |
|
157
|
|
|
|
|
|
|
} |
|
158
|
2
|
|
|
|
|
3
|
return $cumulated_bits + $self->getBitvector($chr_sorted[$i])->rank($pos); |
|
159
|
|
|
|
|
|
|
} |
|
160
|
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
|
|
162
|
|
|
|
|
|
|
sub select { |
|
163
|
1
|
|
|
1
|
1
|
2
|
my $self = shift; |
|
164
|
1
|
|
|
|
|
1
|
my $i = shift; |
|
165
|
1
|
|
|
|
|
1
|
my $cumulated_bits = 0; |
|
166
|
1
|
|
|
|
|
1
|
my @chr_sorted = sort keys %{$self->{bit_vectors}}; |
|
|
1
|
|
|
|
|
6
|
|
|
167
|
1
|
|
|
|
|
1
|
my $j = 0; |
|
168
|
1
|
|
66
|
|
|
6
|
while($j < @chr_sorted && $cumulated_bits + $self->getBitvector($chr_sorted[$j])->nb_set < $i) { |
|
169
|
2
|
|
|
|
|
3
|
my $chr = $chr_sorted[$j]; |
|
170
|
2
|
|
|
|
|
2
|
my $bv = $self->getBitvector($chr); |
|
171
|
2
|
|
|
|
|
3
|
$cumulated_bits += $self->getBitvector($chr)->nb_set; |
|
172
|
2
|
|
|
|
|
6
|
$j++; |
|
173
|
|
|
|
|
|
|
} |
|
174
|
1
|
|
|
|
|
1
|
my $chr = $chr_sorted[$j-1]; |
|
175
|
1
|
|
|
|
|
2
|
my $pos = $self->getBitvector($chr_sorted[$j-1])->select($i - $cumulated_bits + 1); |
|
176
|
1
|
|
|
|
|
3
|
return ($chr,$pos); |
|
177
|
|
|
|
|
|
|
} |
|
178
|
|
|
|
|
|
|
|
|
179
|
|
|
|
|
|
|
1; |
|
180
|
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
__END__ |