| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package AlignDB::Window; |
|
2
|
6
|
|
|
6
|
|
115402
|
use Moose; |
|
|
6
|
|
|
|
|
1791386
|
|
|
|
6
|
|
|
|
|
34
|
|
|
3
|
6
|
|
|
6
|
|
28564
|
use Carp; |
|
|
6
|
|
|
|
|
10
|
|
|
|
6
|
|
|
|
|
397
|
|
|
4
|
|
|
|
|
|
|
|
|
5
|
6
|
|
|
6
|
|
23
|
use List::Util; |
|
|
6
|
|
|
|
|
15
|
|
|
|
6
|
|
|
|
|
242
|
|
|
6
|
6
|
|
|
6
|
|
3610
|
use YAML::Syck; |
|
|
6
|
|
|
|
|
8895
|
|
|
|
6
|
|
|
|
|
330
|
|
|
7
|
6
|
|
|
6
|
|
506
|
use AlignDB::IntSpan; |
|
|
6
|
|
|
|
|
5542
|
|
|
|
6
|
|
|
|
|
13124
|
|
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
our $VERSION = '1.1.0'; |
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
has 'sw_size' => ( is => 'rw', isa => 'Int', default => sub {100}, ); |
|
12
|
|
|
|
|
|
|
has 'min_interval' => ( is => 'rw', isa => 'Int', default => sub {11}, ); |
|
13
|
|
|
|
|
|
|
has 'max_out_distance' => ( is => 'rw', isa => 'Int', default => sub {10}, ); |
|
14
|
|
|
|
|
|
|
has 'max_in_distance' => ( is => 'rw', isa => 'Int', default => sub {5}, ); |
|
15
|
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
sub interval_window { |
|
17
|
8
|
|
|
8
|
1
|
5011
|
my $self = shift; |
|
18
|
8
|
|
|
|
|
9
|
my AlignDB::IntSpan $comparable_set = shift; |
|
19
|
8
|
|
|
|
|
9
|
my ( $interval_start, $interval_end, $sw_size, $min_interval ) = @_; |
|
20
|
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
# if undefined, set to default values |
|
22
|
8
|
|
33
|
|
|
224
|
$sw_size ||= $self->sw_size; |
|
23
|
8
|
|
33
|
|
|
179
|
$min_interval ||= $self->min_interval; |
|
24
|
|
|
|
|
|
|
|
|
25
|
8
|
|
|
|
|
19
|
my $interval_set = AlignDB::IntSpan->new->add_pair( $interval_start, $interval_end ); |
|
26
|
8
|
|
|
|
|
239
|
$interval_set = $interval_set->intersect($comparable_set); |
|
27
|
8
|
|
|
|
|
1133
|
my $interval_length = $interval_set->size; |
|
28
|
|
|
|
|
|
|
|
|
29
|
8
|
|
|
|
|
75
|
my $density = int( $interval_length / $sw_size ) - 1; |
|
30
|
|
|
|
|
|
|
|
|
31
|
|
|
|
|
|
|
# used for mark windows |
|
32
|
8
|
|
|
|
|
10
|
my ( $tmp_start, $tmp_end ) = ( 1, $interval_length ); |
|
33
|
|
|
|
|
|
|
|
|
34
|
8
|
|
|
|
|
7
|
my @interval_windows; |
|
35
|
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
# More windows will be submitted in the following for-loop |
|
37
|
|
|
|
|
|
|
# and if section |
|
38
|
8
|
100
|
|
|
|
22
|
if ( $interval_length < $min_interval ) { |
|
|
|
100
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
39
|
1
|
|
|
|
|
3
|
return @interval_windows; |
|
40
|
|
|
|
|
|
|
} |
|
41
|
|
|
|
|
|
|
elsif ( $interval_length < $sw_size ) { |
|
42
|
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
# $interval_length between $min_interval and 99 bp |
|
44
|
2
|
|
|
|
|
3
|
my %window_info; |
|
45
|
2
|
|
|
|
|
3
|
$window_info{set} = $interval_set; |
|
46
|
2
|
|
|
|
|
3
|
$window_info{distance} = -1; |
|
47
|
2
|
|
|
|
|
3
|
$window_info{density} = $density; |
|
48
|
2
|
|
|
|
|
2
|
$window_info{type} = 'S'; |
|
49
|
2
|
|
|
|
|
3
|
push @interval_windows, \%window_info; |
|
50
|
|
|
|
|
|
|
|
|
51
|
2
|
|
|
|
|
5
|
return @interval_windows; |
|
52
|
|
|
|
|
|
|
} |
|
53
|
|
|
|
|
|
|
elsif ( $interval_length < $sw_size * 2 ) { |
|
54
|
|
|
|
|
|
|
|
|
55
|
|
|
|
|
|
|
# $interval_length between 100 and 199 bp |
|
56
|
|
|
|
|
|
|
# LO between 50 and 100 bp |
|
57
|
|
|
|
|
|
|
# R0 between 50 and 99 bp |
|
58
|
|
|
|
|
|
|
# left windows |
|
59
|
2
|
|
|
|
|
2
|
my %l_window_info; |
|
60
|
2
|
|
|
|
|
22
|
my $l_start = $tmp_start; |
|
61
|
2
|
|
|
|
|
4
|
my $l_end = $l_start + int( $interval_length / 2 ) - 1; |
|
62
|
2
|
100
|
|
|
|
5
|
if ( $interval_length % 2 ) { |
|
63
|
1
|
|
|
|
|
2
|
$l_end++; |
|
64
|
|
|
|
|
|
|
} |
|
65
|
2
|
|
|
|
|
6
|
$l_window_info{set} = $interval_set->slice( $l_start, $l_end ); |
|
66
|
2
|
|
|
|
|
116
|
$l_window_info{distance} = 0; |
|
67
|
2
|
|
|
|
|
4
|
$l_window_info{density} = $density; |
|
68
|
2
|
|
|
|
|
2
|
$l_window_info{type} = 'L'; |
|
69
|
2
|
|
|
|
|
3
|
push @interval_windows, \%l_window_info; |
|
70
|
|
|
|
|
|
|
|
|
71
|
|
|
|
|
|
|
# right windows |
|
72
|
2
|
|
|
|
|
2
|
my %r_window_info; |
|
73
|
2
|
|
|
|
|
2
|
my $r_end = $tmp_end; |
|
74
|
2
|
|
|
|
|
4
|
my $r_start = $r_end - int( $interval_length / 2 ) + 1; |
|
75
|
2
|
|
|
|
|
4
|
$r_window_info{set} = $interval_set->slice( $r_start, $r_end ); |
|
76
|
2
|
|
|
|
|
90
|
$r_window_info{distance} = 0; |
|
77
|
2
|
|
|
|
|
3
|
$r_window_info{density} = $density; |
|
78
|
2
|
|
|
|
|
2
|
$r_window_info{type} = 'R'; |
|
79
|
2
|
|
|
|
|
3
|
push @interval_windows, \%r_window_info; |
|
80
|
|
|
|
|
|
|
|
|
81
|
2
|
|
|
|
|
7
|
return @interval_windows; |
|
82
|
|
|
|
|
|
|
} |
|
83
|
|
|
|
|
|
|
else { |
|
84
|
|
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
# LO => 50 bp, R0 => 50 bp |
|
86
|
|
|
|
|
|
|
# left windows |
|
87
|
3
|
|
|
|
|
3
|
my %l_window_info; |
|
88
|
3
|
|
|
|
|
3
|
my $l_start = $tmp_start; |
|
89
|
3
|
|
|
|
|
4
|
my $l_end = $l_start + $sw_size / 2 - 1; |
|
90
|
3
|
|
|
|
|
7
|
$l_window_info{set} = $interval_set->slice( $l_start, $l_end ); |
|
91
|
3
|
|
|
|
|
143
|
$l_window_info{distance} = 0; |
|
92
|
3
|
|
|
|
|
2
|
$l_window_info{density} = $density; |
|
93
|
3
|
|
|
|
|
4
|
$l_window_info{type} = 'L'; |
|
94
|
3
|
|
|
|
|
4
|
push @interval_windows, \%l_window_info; |
|
95
|
3
|
|
|
|
|
2
|
$tmp_start = $l_end + 1; |
|
96
|
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
# right windows |
|
98
|
3
|
|
|
|
|
2
|
my %r_window_info; |
|
99
|
3
|
|
|
|
|
4
|
my $r_end = $tmp_end; |
|
100
|
3
|
|
|
|
|
3
|
my $r_start = $r_end - $sw_size / 2 + 1; |
|
101
|
3
|
|
|
|
|
5
|
$r_window_info{set} = $interval_set->slice( $r_start, $r_end ); |
|
102
|
3
|
|
|
|
|
131
|
$r_window_info{distance} = 0; |
|
103
|
3
|
|
|
|
|
4
|
$r_window_info{density} = $density; |
|
104
|
3
|
|
|
|
|
2
|
$r_window_info{type} = 'R'; |
|
105
|
3
|
|
|
|
|
3
|
push @interval_windows, \%r_window_info; |
|
106
|
3
|
|
|
|
|
4
|
$tmp_end = $r_start - 1; |
|
107
|
|
|
|
|
|
|
} |
|
108
|
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
# regular 100 bp windows |
|
110
|
|
|
|
|
|
|
# When density IN (1, 2), this loop will be bypassed. |
|
111
|
3
|
|
|
|
|
6
|
my $half_windows_number = int( ( $density - 1 ) / 2 ); |
|
112
|
3
|
|
|
|
|
11
|
for ( my $i = 1; $i <= $half_windows_number; $i++ ) { |
|
113
|
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
# left windows |
|
115
|
3
|
|
|
|
|
2
|
my %l_window_info; |
|
116
|
3
|
|
|
|
|
4
|
my $l_start = $tmp_start; |
|
117
|
3
|
|
|
|
|
3
|
my $l_end = $l_start + $sw_size - 1; |
|
118
|
3
|
|
|
|
|
5
|
$l_window_info{set} = $interval_set->slice( $l_start, $l_end ); |
|
119
|
3
|
|
|
|
|
129
|
$l_window_info{distance} = $i; |
|
120
|
3
|
|
|
|
|
4
|
$l_window_info{density} = $density; |
|
121
|
3
|
|
|
|
|
3
|
$l_window_info{type} = 'L'; |
|
122
|
3
|
|
|
|
|
4
|
push @interval_windows, \%l_window_info; |
|
123
|
3
|
|
|
|
|
2
|
$tmp_start = $l_end + 1; |
|
124
|
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
# right windows |
|
126
|
3
|
|
|
|
|
10
|
my %r_window_info; |
|
127
|
3
|
|
|
|
|
3
|
my $r_end = $tmp_end; |
|
128
|
3
|
|
|
|
|
2
|
my $r_start = $r_end - $sw_size + 1; |
|
129
|
3
|
|
|
|
|
7
|
$r_window_info{set} = $interval_set->slice( $r_start, $r_end ); |
|
130
|
3
|
|
|
|
|
131
|
$r_window_info{distance} = $i; |
|
131
|
3
|
|
|
|
|
2
|
$r_window_info{density} = $density; |
|
132
|
3
|
|
|
|
|
4
|
$r_window_info{type} = 'R'; |
|
133
|
3
|
|
|
|
|
3
|
push @interval_windows, \%r_window_info; |
|
134
|
3
|
|
|
|
|
7
|
$tmp_end = $r_start - 1; |
|
135
|
|
|
|
|
|
|
} |
|
136
|
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
# Three special windwos: |
|
138
|
|
|
|
|
|
|
# 1. Ln, range from 100 to 150 bp when even density number |
|
139
|
|
|
|
|
|
|
# 2. Rn, range from 100 to 149 bp when even density number |
|
140
|
|
|
|
|
|
|
# 3. Ln or Rn, range from 100 to 199 bp when odd density number |
|
141
|
3
|
100
|
|
|
|
6
|
if ( $density % 2 ) { |
|
142
|
|
|
|
|
|
|
|
|
143
|
|
|
|
|
|
|
# middle windows |
|
144
|
2
|
|
|
|
|
2
|
my %window_info; |
|
145
|
2
|
|
|
|
|
6
|
$window_info{set} = $interval_set->slice( $tmp_start, $tmp_end ); |
|
146
|
2
|
|
|
|
|
91
|
$window_info{distance} = $half_windows_number + 1; |
|
147
|
2
|
|
|
|
|
1
|
$window_info{density} = $density; |
|
148
|
2
|
|
|
|
|
3
|
$window_info{type} = "L"; # rand > 0.5 ? "L" : "R"; |
|
149
|
2
|
|
|
|
|
3
|
push @interval_windows, \%window_info; |
|
150
|
|
|
|
|
|
|
} |
|
151
|
|
|
|
|
|
|
else { |
|
152
|
1
|
|
|
|
|
2
|
my $remain_length = $tmp_end - $tmp_start + 1; |
|
153
|
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
# left windows |
|
155
|
1
|
|
|
|
|
1
|
my %l_window_info; |
|
156
|
1
|
|
|
|
|
2
|
my $l_start = $tmp_start; |
|
157
|
1
|
|
|
|
|
2
|
my $l_end = $l_start + int( $remain_length / 2 ) - 1; |
|
158
|
1
|
50
|
|
|
|
3
|
if ( $remain_length % 2 ) { |
|
159
|
1
|
|
|
|
|
1
|
$l_end++; |
|
160
|
|
|
|
|
|
|
} |
|
161
|
1
|
|
|
|
|
2
|
$l_window_info{set} = $interval_set->slice( $l_start, $l_end ); |
|
162
|
1
|
|
|
|
|
52
|
$l_window_info{distance} = $half_windows_number + 1; |
|
163
|
1
|
|
|
|
|
1
|
$l_window_info{density} = $density; |
|
164
|
1
|
|
|
|
|
2
|
$l_window_info{type} = 'L'; |
|
165
|
1
|
|
|
|
|
2
|
push @interval_windows, \%l_window_info; |
|
166
|
|
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
# right windows |
|
168
|
1
|
|
|
|
|
1
|
my %r_window_info; |
|
169
|
1
|
|
|
|
|
1
|
my $r_end = $tmp_end; |
|
170
|
1
|
|
|
|
|
3
|
my $r_start = $r_end - int( $remain_length / 2 ) + 1; |
|
171
|
1
|
|
|
|
|
2
|
$r_window_info{set} = $interval_set->slice( $r_start, $r_end ); |
|
172
|
1
|
|
|
|
|
45
|
$r_window_info{distance} = $half_windows_number + 1; |
|
173
|
1
|
|
|
|
|
2
|
$r_window_info{density} = $density; |
|
174
|
1
|
|
|
|
|
1
|
$r_window_info{type} = 'R'; |
|
175
|
1
|
|
|
|
|
2
|
push @interval_windows, \%r_window_info; |
|
176
|
|
|
|
|
|
|
} |
|
177
|
|
|
|
|
|
|
|
|
178
|
3
|
|
|
|
|
9
|
return @interval_windows; |
|
179
|
|
|
|
|
|
|
} |
|
180
|
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
sub outside_window { |
|
182
|
3
|
|
|
3
|
1
|
2887
|
my $self = shift; |
|
183
|
3
|
|
|
|
|
4
|
my AlignDB::IntSpan $comparable_set = shift; |
|
184
|
3
|
|
|
|
|
4
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
185
|
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
# if undefined, set to default values |
|
187
|
3
|
|
33
|
|
|
90
|
$sw_size ||= $self->sw_size; |
|
188
|
3
|
|
33
|
|
|
73
|
$maximal_distance ||= $self->max_out_distance; |
|
189
|
|
|
|
|
|
|
|
|
190
|
3
|
|
|
|
|
9
|
my $comparable_number = $comparable_set->size; |
|
191
|
|
|
|
|
|
|
|
|
192
|
3
|
|
|
|
|
37
|
my @outside_windows; |
|
193
|
|
|
|
|
|
|
|
|
194
|
3
|
|
|
|
|
4
|
for my $sw_type (qw{L R}) { |
|
195
|
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
197
|
6
|
|
|
|
|
4
|
my ( $sw_start, $sw_end ); |
|
198
|
|
|
|
|
|
|
|
|
199
|
6
|
100
|
|
|
|
14
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
200
|
3
|
|
|
|
|
5
|
$sw_start = $comparable_set->index($internal_end) + 1; |
|
201
|
3
|
|
|
|
|
28
|
$sw_end = $sw_start + $sw_size - 1; |
|
202
|
|
|
|
|
|
|
} |
|
203
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
204
|
3
|
|
|
|
|
6
|
$sw_end = $comparable_set->index($internal_start) - 1; |
|
205
|
3
|
|
|
|
|
37
|
$sw_start = $sw_end - $sw_size + 1; |
|
206
|
|
|
|
|
|
|
} |
|
207
|
|
|
|
|
|
|
|
|
208
|
|
|
|
|
|
|
# $sw_distance is from 1 to $sw_max_distance |
|
209
|
6
|
|
|
|
|
10
|
OSW: for my $sw_distance ( 1 .. $maximal_distance ) { |
|
210
|
12
|
100
|
|
|
|
19
|
last if $sw_start < 1; |
|
211
|
11
|
50
|
|
|
|
14
|
last if $sw_end > $comparable_number; |
|
212
|
11
|
|
|
|
|
16
|
my $sw_set = $comparable_set->slice( $sw_start, $sw_end ); |
|
213
|
11
|
|
|
|
|
536
|
my $sw_set_member_number = $sw_set->size; |
|
214
|
11
|
50
|
|
|
|
92
|
if ( $sw_set_member_number < $sw_size ) { |
|
215
|
0
|
|
|
|
|
0
|
last OSW; |
|
216
|
|
|
|
|
|
|
} |
|
217
|
|
|
|
|
|
|
|
|
218
|
11
|
|
|
|
|
5
|
my %window_info; |
|
219
|
11
|
|
|
|
|
15
|
$window_info{type} = $sw_type; |
|
220
|
11
|
|
|
|
|
9
|
$window_info{set} = $sw_set; |
|
221
|
11
|
|
|
|
|
10
|
$window_info{distance} = $sw_distance; |
|
222
|
|
|
|
|
|
|
|
|
223
|
11
|
|
|
|
|
27
|
push @outside_windows, \%window_info; |
|
224
|
|
|
|
|
|
|
|
|
225
|
11
|
100
|
|
|
|
21
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
226
|
6
|
|
|
|
|
3
|
$sw_start = $sw_end + 1; |
|
227
|
6
|
|
|
|
|
8
|
$sw_end = $sw_start + $sw_size - 1; |
|
228
|
|
|
|
|
|
|
} |
|
229
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
230
|
5
|
|
|
|
|
4
|
$sw_end = $sw_start - 1; |
|
231
|
5
|
|
|
|
|
8
|
$sw_start = $sw_end - $sw_size + 1; |
|
232
|
|
|
|
|
|
|
} |
|
233
|
|
|
|
|
|
|
} |
|
234
|
|
|
|
|
|
|
} |
|
235
|
|
|
|
|
|
|
|
|
236
|
3
|
|
|
|
|
7
|
return @outside_windows; |
|
237
|
|
|
|
|
|
|
} |
|
238
|
|
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
sub outside_window_2 { |
|
240
|
2
|
|
|
2
|
1
|
1343
|
my $self = shift; |
|
241
|
2
|
|
|
|
|
3
|
my AlignDB::IntSpan $comparable_set = shift; |
|
242
|
2
|
|
|
|
|
2
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
243
|
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
# if undefined, set to default values |
|
245
|
2
|
|
33
|
|
|
70
|
$sw_size ||= $self->sw_size; |
|
246
|
2
|
|
33
|
|
|
49
|
$maximal_distance ||= $self->max_out_distance; |
|
247
|
|
|
|
|
|
|
|
|
248
|
2
|
|
|
|
|
4
|
my $window0_size = int( $sw_size / 2 ); |
|
249
|
|
|
|
|
|
|
|
|
250
|
2
|
|
|
|
|
7
|
my $comparable_number = $comparable_set->size; |
|
251
|
|
|
|
|
|
|
|
|
252
|
2
|
|
|
|
|
22
|
my @outside_windows; |
|
253
|
|
|
|
|
|
|
|
|
254
|
2
|
|
|
|
|
3
|
for my $sw_type (qw{L R}) { |
|
255
|
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
257
|
4
|
|
|
|
|
3
|
my ( $sw_start, $sw_end ); |
|
258
|
|
|
|
|
|
|
|
|
259
|
4
|
100
|
|
|
|
9
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
260
|
2
|
|
|
|
|
5
|
$sw_start = $comparable_set->index($internal_end) + 1; |
|
261
|
2
|
|
|
|
|
19
|
$sw_end = $sw_start + $window0_size - 1; |
|
262
|
|
|
|
|
|
|
} |
|
263
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
264
|
2
|
|
|
|
|
4
|
$sw_end = $comparable_set->index($internal_start) - 1; |
|
265
|
2
|
|
|
|
|
23
|
$sw_start = $sw_end - $window0_size + 1; |
|
266
|
|
|
|
|
|
|
} |
|
267
|
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
# distance is from 0 to $maximal_distance |
|
269
|
4
|
|
|
|
|
8
|
OSW2: for my $sw_distance ( 0 .. $maximal_distance ) { |
|
270
|
8
|
100
|
|
|
|
12
|
last if $sw_start < 1; |
|
271
|
7
|
50
|
|
|
|
16
|
last if $sw_end > $comparable_number; |
|
272
|
7
|
|
|
|
|
12
|
my $sw_set = $comparable_set->slice( $sw_start, $sw_end ); |
|
273
|
7
|
|
|
|
|
342
|
my $sw_set_member_number = $sw_set->size; |
|
274
|
7
|
50
|
|
|
|
65
|
if ( $sw_set_member_number < $window0_size ) { |
|
275
|
|
|
|
|
|
|
|
|
276
|
|
|
|
|
|
|
# warn YAML::Syck::Dump { |
|
277
|
|
|
|
|
|
|
# sw_type => $sw_type, |
|
278
|
|
|
|
|
|
|
# sw_start => $sw_start, |
|
279
|
|
|
|
|
|
|
# sw_end => $sw_end, |
|
280
|
|
|
|
|
|
|
# }; |
|
281
|
0
|
|
|
|
|
0
|
last OSW2; |
|
282
|
|
|
|
|
|
|
} |
|
283
|
|
|
|
|
|
|
|
|
284
|
7
|
|
|
|
|
6
|
my %window_info; |
|
285
|
7
|
|
|
|
|
9
|
$window_info{type} = $sw_type; |
|
286
|
7
|
|
|
|
|
7
|
$window_info{set} = $sw_set; |
|
287
|
7
|
|
|
|
|
6
|
$window_info{distance} = $sw_distance; |
|
288
|
|
|
|
|
|
|
|
|
289
|
7
|
|
|
|
|
8
|
push @outside_windows, \%window_info; |
|
290
|
|
|
|
|
|
|
|
|
291
|
7
|
100
|
|
|
|
18
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
292
|
4
|
|
|
|
|
4
|
$sw_start = $sw_end + 1; |
|
293
|
4
|
|
|
|
|
6
|
$sw_end = $sw_start + $sw_size - 1; |
|
294
|
|
|
|
|
|
|
} |
|
295
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
296
|
3
|
|
|
|
|
3
|
$sw_end = $sw_start - 1; |
|
297
|
3
|
|
|
|
|
4
|
$sw_start = $sw_end - $sw_size + 1; |
|
298
|
|
|
|
|
|
|
} |
|
299
|
|
|
|
|
|
|
} |
|
300
|
|
|
|
|
|
|
} |
|
301
|
|
|
|
|
|
|
|
|
302
|
2
|
|
|
|
|
5
|
return @outside_windows; |
|
303
|
|
|
|
|
|
|
} |
|
304
|
|
|
|
|
|
|
|
|
305
|
|
|
|
|
|
|
sub inside_window { |
|
306
|
3
|
|
|
3
|
1
|
1888
|
my $self = shift; |
|
307
|
3
|
|
|
|
|
4
|
my AlignDB::IntSpan $comparable_set = shift; |
|
308
|
3
|
|
|
|
|
4
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
309
|
|
|
|
|
|
|
|
|
310
|
|
|
|
|
|
|
# if undefined, set to default values |
|
311
|
3
|
|
33
|
|
|
89
|
$sw_size ||= $self->sw_size; |
|
312
|
3
|
|
33
|
|
|
73
|
$maximal_distance ||= $self->max_in_distance; |
|
313
|
|
|
|
|
|
|
|
|
314
|
3
|
|
|
|
|
3
|
my @inside_windows; |
|
315
|
|
|
|
|
|
|
|
|
316
|
3
|
|
|
|
|
5
|
for my $sw_type (qw{l r}) { |
|
317
|
|
|
|
|
|
|
|
|
318
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
319
|
6
|
|
|
|
|
4
|
my ( $sw_start, $sw_end ); |
|
320
|
|
|
|
|
|
|
|
|
321
|
6
|
|
|
|
|
18
|
my $working_set = $comparable_set->intersect("$internal_start-$internal_end"); |
|
322
|
6
|
|
|
|
|
1212
|
my $working_length = $working_set->size; |
|
323
|
6
|
50
|
|
|
|
63
|
last if $working_length < $sw_size; |
|
324
|
|
|
|
|
|
|
|
|
325
|
|
|
|
|
|
|
# the 'r' and 'l' windows may overlap each other |
|
326
|
6
|
100
|
|
|
|
15
|
if ( $sw_type eq 'r' ) { |
|
|
|
50
|
|
|
|
|
|
|
327
|
3
|
|
|
|
|
5
|
$sw_end = $working_set->size; |
|
328
|
3
|
|
|
|
|
23
|
$sw_start = $sw_end - $sw_size + 1; |
|
329
|
|
|
|
|
|
|
} |
|
330
|
|
|
|
|
|
|
elsif ( $sw_type eq 'l' ) { |
|
331
|
3
|
|
|
|
|
2
|
$sw_start = 1; |
|
332
|
3
|
|
|
|
|
8
|
$sw_end = $sw_start + $sw_size - 1; |
|
333
|
|
|
|
|
|
|
} |
|
334
|
|
|
|
|
|
|
|
|
335
|
6
|
|
|
|
|
12
|
my $available_distance = int( $working_length / ( $sw_size * 2 ) ); |
|
336
|
6
|
|
|
|
|
14
|
my $max_distance = List::Util::min( $available_distance, $maximal_distance ); |
|
337
|
|
|
|
|
|
|
|
|
338
|
|
|
|
|
|
|
# sw_distance is from -1 to -max_distance |
|
339
|
6
|
|
|
|
|
12
|
ISW: for my $i ( 1 .. $max_distance ) { |
|
340
|
6
|
|
|
|
|
13
|
my $sw_set = $working_set->slice( $sw_start, $sw_end ); |
|
341
|
6
|
|
|
|
|
304
|
my $sw_set_member_number = $sw_set->size; |
|
342
|
6
|
50
|
|
|
|
73
|
if ( $sw_set_member_number < $sw_size ) { |
|
343
|
0
|
|
|
|
|
0
|
last ISW; |
|
344
|
|
|
|
|
|
|
} |
|
345
|
6
|
|
|
|
|
4
|
my $sw_distance = -$i; |
|
346
|
|
|
|
|
|
|
|
|
347
|
6
|
|
|
|
|
5
|
my %window_info; |
|
348
|
6
|
|
|
|
|
8
|
$window_info{type} = $sw_type; |
|
349
|
6
|
|
|
|
|
6
|
$window_info{set} = $sw_set; |
|
350
|
6
|
|
|
|
|
4
|
$window_info{distance} = $sw_distance; |
|
351
|
|
|
|
|
|
|
|
|
352
|
6
|
|
|
|
|
7
|
push @inside_windows, \%window_info; |
|
353
|
|
|
|
|
|
|
|
|
354
|
6
|
100
|
|
|
|
12
|
if ( $sw_type eq 'r' ) { |
|
|
|
50
|
|
|
|
|
|
|
355
|
3
|
|
|
|
|
3
|
$sw_end = $sw_start - 1; |
|
356
|
3
|
|
|
|
|
5
|
$sw_start = $sw_end - $sw_size + 1; |
|
357
|
|
|
|
|
|
|
} |
|
358
|
|
|
|
|
|
|
elsif ( $sw_type eq 'l' ) { |
|
359
|
3
|
|
|
|
|
3
|
$sw_start = $sw_end + 1; |
|
360
|
3
|
|
|
|
|
7
|
$sw_end = $sw_start + $sw_size - 1; |
|
361
|
|
|
|
|
|
|
} |
|
362
|
|
|
|
|
|
|
} |
|
363
|
|
|
|
|
|
|
} |
|
364
|
|
|
|
|
|
|
|
|
365
|
3
|
|
|
|
|
8
|
return @inside_windows; |
|
366
|
|
|
|
|
|
|
} |
|
367
|
|
|
|
|
|
|
|
|
368
|
|
|
|
|
|
|
sub inside_window_2 { |
|
369
|
3
|
|
|
3
|
1
|
1480
|
my $self = shift; |
|
370
|
3
|
|
|
|
|
4
|
my AlignDB::IntSpan $comparable_set = shift; |
|
371
|
3
|
|
|
|
|
5
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
372
|
|
|
|
|
|
|
|
|
373
|
|
|
|
|
|
|
# if undefined, set to default values |
|
374
|
3
|
|
33
|
|
|
85
|
$sw_size ||= $self->sw_size; |
|
375
|
3
|
|
33
|
|
|
70
|
$maximal_distance ||= $self->max_in_distance; |
|
376
|
|
|
|
|
|
|
|
|
377
|
3
|
|
|
|
|
3
|
my @inside_windows; |
|
378
|
|
|
|
|
|
|
|
|
379
|
3
|
|
|
|
|
4
|
for my $sw_type (qw{l r}) { |
|
380
|
|
|
|
|
|
|
|
|
381
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
382
|
6
|
|
|
|
|
7
|
my ( $sw_start, $sw_end ); |
|
383
|
|
|
|
|
|
|
|
|
384
|
6
|
|
|
|
|
29
|
my $working_set = $comparable_set->intersect("$internal_start-$internal_end"); |
|
385
|
6
|
|
|
|
|
1090
|
my $working_length = $working_set->size; |
|
386
|
6
|
50
|
|
|
|
56
|
last if $working_length < $sw_size; |
|
387
|
|
|
|
|
|
|
|
|
388
|
|
|
|
|
|
|
# the windows start from the center |
|
389
|
6
|
100
|
|
|
|
14
|
if ( $sw_type eq 'r' ) { |
|
|
|
50
|
|
|
|
|
|
|
390
|
3
|
|
|
|
|
5
|
$sw_start = int( $working_length / 2 ) + 1; |
|
391
|
3
|
|
|
|
|
4
|
$sw_end = $sw_start + $sw_size - 1; |
|
392
|
|
|
|
|
|
|
} |
|
393
|
|
|
|
|
|
|
elsif ( $sw_type eq 'l' ) { |
|
394
|
3
|
|
|
|
|
4
|
$sw_end = int( $working_length / 2 ); |
|
395
|
3
|
|
|
|
|
4
|
$sw_start = $sw_end - $sw_size + 1; |
|
396
|
|
|
|
|
|
|
} |
|
397
|
|
|
|
|
|
|
|
|
398
|
6
|
|
|
|
|
7
|
my $available_distance = int( $working_length / ( $sw_size * 2 ) ); |
|
399
|
6
|
|
|
|
|
14
|
my $max_distance = List::Util::min( $available_distance, $maximal_distance ); |
|
400
|
|
|
|
|
|
|
|
|
401
|
|
|
|
|
|
|
# sw_distance is from -90 to -90 + max_distance - 1 |
|
402
|
6
|
|
|
|
|
12
|
ISW2: for my $i ( -90 .. ( -90 + $max_distance - 1 ) ) { |
|
403
|
6
|
|
|
|
|
9
|
my $sw_set = $working_set->slice( $sw_start, $sw_end ); |
|
404
|
6
|
|
|
|
|
270
|
my $sw_set_member_number = $sw_set->size; |
|
405
|
6
|
50
|
|
|
|
50
|
if ( $sw_set_member_number < $sw_size ) { |
|
406
|
0
|
|
|
|
|
0
|
last ISW2; |
|
407
|
|
|
|
|
|
|
} |
|
408
|
6
|
|
|
|
|
6
|
my $sw_distance = $i; |
|
409
|
|
|
|
|
|
|
|
|
410
|
6
|
|
|
|
|
5
|
my %window_info; |
|
411
|
6
|
|
|
|
|
7
|
$window_info{type} = $sw_type; |
|
412
|
6
|
|
|
|
|
5
|
$window_info{set} = $sw_set; |
|
413
|
6
|
|
|
|
|
9
|
$window_info{distance} = $sw_distance; |
|
414
|
|
|
|
|
|
|
|
|
415
|
6
|
|
|
|
|
6
|
push @inside_windows, \%window_info; |
|
416
|
|
|
|
|
|
|
|
|
417
|
6
|
100
|
|
|
|
11
|
if ( $sw_type eq 'r' ) { |
|
|
|
50
|
|
|
|
|
|
|
418
|
3
|
|
|
|
|
4
|
$sw_start = $sw_end + 1; |
|
419
|
3
|
|
|
|
|
5
|
$sw_end = $sw_start + $sw_size - 1; |
|
420
|
|
|
|
|
|
|
} |
|
421
|
|
|
|
|
|
|
elsif ( $sw_type eq 'l' ) { |
|
422
|
3
|
|
|
|
|
3
|
$sw_end = $sw_start - 1; |
|
423
|
3
|
|
|
|
|
6
|
$sw_start = $sw_end - $sw_size + 1; |
|
424
|
|
|
|
|
|
|
} |
|
425
|
|
|
|
|
|
|
} |
|
426
|
|
|
|
|
|
|
} |
|
427
|
|
|
|
|
|
|
|
|
428
|
3
|
|
|
|
|
8
|
return @inside_windows; |
|
429
|
|
|
|
|
|
|
} |
|
430
|
|
|
|
|
|
|
|
|
431
|
|
|
|
|
|
|
sub center_window { |
|
432
|
3
|
|
|
3
|
1
|
2541
|
my $self = shift; |
|
433
|
3
|
|
|
|
|
3
|
my AlignDB::IntSpan $comparable_set = shift; |
|
434
|
3
|
|
|
|
|
4
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
435
|
|
|
|
|
|
|
|
|
436
|
|
|
|
|
|
|
# if undefined, set to default values |
|
437
|
3
|
|
33
|
|
|
88
|
$sw_size ||= $self->sw_size; |
|
438
|
3
|
|
33
|
|
|
70
|
$maximal_distance ||= $self->max_out_distance; |
|
439
|
|
|
|
|
|
|
|
|
440
|
3
|
|
|
|
|
9
|
my $comparable_number = $comparable_set->size; |
|
441
|
|
|
|
|
|
|
|
|
442
|
3
|
|
|
|
|
38
|
my @center_windows; |
|
443
|
|
|
|
|
|
|
|
|
444
|
|
|
|
|
|
|
my $original_set; |
|
445
|
3
|
100
|
|
|
|
8
|
if ( $internal_start < $internal_end ) { |
|
|
|
50
|
|
|
|
|
|
|
446
|
1
|
|
|
|
|
5
|
$original_set = AlignDB::IntSpan->new("$internal_start-$internal_end"); |
|
447
|
|
|
|
|
|
|
} |
|
448
|
|
|
|
|
|
|
elsif ( $internal_start == $internal_end ) { |
|
449
|
2
|
|
|
|
|
4
|
$original_set = AlignDB::IntSpan->new($internal_start); |
|
450
|
|
|
|
|
|
|
} |
|
451
|
|
|
|
|
|
|
else { |
|
452
|
0
|
|
|
|
|
0
|
return; |
|
453
|
|
|
|
|
|
|
} |
|
454
|
|
|
|
|
|
|
|
|
455
|
3
|
|
|
|
|
148
|
my AlignDB::IntSpan $window0_set = _center_resize( $original_set, $comparable_set, $sw_size ); |
|
456
|
3
|
50
|
|
|
|
8
|
return unless $window0_set; |
|
457
|
|
|
|
|
|
|
|
|
458
|
3
|
|
|
|
|
45
|
my $window0_start = $window0_set->min; |
|
459
|
3
|
|
|
|
|
30
|
my $window0_end = $window0_set->max; |
|
460
|
3
|
|
|
|
|
27
|
push @center_windows, |
|
461
|
|
|
|
|
|
|
{ |
|
462
|
|
|
|
|
|
|
type => 'M', |
|
463
|
|
|
|
|
|
|
set => $window0_set, |
|
464
|
|
|
|
|
|
|
distance => 0, |
|
465
|
|
|
|
|
|
|
}; |
|
466
|
|
|
|
|
|
|
|
|
467
|
3
|
|
|
|
|
6
|
for my $sw_type (qw{L R}) { |
|
468
|
|
|
|
|
|
|
|
|
469
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
470
|
6
|
|
|
|
|
6
|
my ( $sw_start, $sw_end ); |
|
471
|
|
|
|
|
|
|
|
|
472
|
6
|
100
|
|
|
|
12
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
473
|
3
|
|
|
|
|
3
|
$sw_start = $comparable_set->index($window0_end) + 1; |
|
474
|
3
|
|
|
|
|
30
|
$sw_end = $sw_start + $sw_size - 1; |
|
475
|
|
|
|
|
|
|
} |
|
476
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
477
|
3
|
|
|
|
|
5
|
$sw_end = $comparable_set->index($window0_start) - 1; |
|
478
|
3
|
|
|
|
|
30
|
$sw_start = $sw_end - $sw_size + 1; |
|
479
|
|
|
|
|
|
|
} |
|
480
|
|
|
|
|
|
|
|
|
481
|
|
|
|
|
|
|
# $sw_distance is from 1 to $maximal_distance |
|
482
|
6
|
|
|
|
|
8
|
CSW: for my $sw_distance ( 1 .. $maximal_distance ) { |
|
483
|
6
|
100
|
|
|
|
10
|
last if $sw_start < 1; |
|
484
|
5
|
50
|
|
|
|
6
|
last if $sw_end > $comparable_number; |
|
485
|
5
|
|
|
|
|
7
|
my $sw_set = $comparable_set->slice( $sw_start, $sw_end ); |
|
486
|
5
|
|
|
|
|
236
|
my $sw_set_member_number = $sw_set->size; |
|
487
|
5
|
50
|
|
|
|
44
|
if ( $sw_set_member_number < $sw_size ) { |
|
488
|
0
|
|
|
|
|
0
|
last CSW; |
|
489
|
|
|
|
|
|
|
} |
|
490
|
|
|
|
|
|
|
|
|
491
|
5
|
|
|
|
|
9
|
push @center_windows, |
|
492
|
|
|
|
|
|
|
{ |
|
493
|
|
|
|
|
|
|
type => $sw_type, |
|
494
|
|
|
|
|
|
|
set => $sw_set, |
|
495
|
|
|
|
|
|
|
distance => $sw_distance, |
|
496
|
|
|
|
|
|
|
}; |
|
497
|
|
|
|
|
|
|
|
|
498
|
5
|
100
|
|
|
|
8
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
499
|
3
|
|
|
|
|
3
|
$sw_start = $sw_end + 1; |
|
500
|
3
|
|
|
|
|
5
|
$sw_end = $sw_start + $sw_size - 1; |
|
501
|
|
|
|
|
|
|
} |
|
502
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
503
|
2
|
|
|
|
|
2
|
$sw_end = $sw_start - 1; |
|
504
|
2
|
|
|
|
|
3
|
$sw_start = $sw_end - $sw_size + 1; |
|
505
|
|
|
|
|
|
|
} |
|
506
|
|
|
|
|
|
|
} |
|
507
|
|
|
|
|
|
|
} |
|
508
|
|
|
|
|
|
|
|
|
509
|
3
|
|
|
|
|
14
|
return @center_windows; |
|
510
|
|
|
|
|
|
|
} |
|
511
|
|
|
|
|
|
|
|
|
512
|
|
|
|
|
|
|
sub _center_resize { |
|
513
|
5
|
|
|
5
|
|
6
|
my AlignDB::IntSpan $old_set = shift; |
|
514
|
5
|
|
|
|
|
2
|
my AlignDB::IntSpan $parent_set = shift; |
|
515
|
5
|
|
|
|
|
4
|
my $resize = shift; |
|
516
|
|
|
|
|
|
|
|
|
517
|
|
|
|
|
|
|
# find the middles of old_set |
|
518
|
5
|
|
|
|
|
5
|
my ( $midleft_parent_idx, $midright_parent_idx ); |
|
519
|
|
|
|
|
|
|
{ |
|
520
|
5
|
100
|
|
|
|
2
|
if ( $old_set->size == 1 ) { |
|
|
5
|
|
|
|
|
18
|
|
|
521
|
4
|
|
|
|
|
42
|
my $mid = $old_set->at(1); |
|
522
|
4
|
|
|
|
|
88
|
$midleft_parent_idx = $parent_set->index($mid); |
|
523
|
4
|
|
|
|
|
46
|
$midright_parent_idx = $parent_set->index($mid); |
|
524
|
|
|
|
|
|
|
} |
|
525
|
|
|
|
|
|
|
else { |
|
526
|
1
|
|
|
|
|
10
|
my $half_size = int( $old_set->size / 2 ); |
|
527
|
1
|
|
|
|
|
12
|
my $midleft = $old_set->at($half_size); |
|
528
|
1
|
|
|
|
|
21
|
my $midright = $old_set->at( $half_size + 1 ); |
|
529
|
1
|
|
|
|
|
19
|
$midleft_parent_idx = $parent_set->index($midleft); |
|
530
|
1
|
|
|
|
|
10
|
$midright_parent_idx = $parent_set->index($midright); |
|
531
|
|
|
|
|
|
|
} |
|
532
|
|
|
|
|
|
|
} |
|
533
|
5
|
50
|
33
|
|
|
60
|
return unless $midleft_parent_idx and $midright_parent_idx; |
|
534
|
|
|
|
|
|
|
|
|
535
|
|
|
|
|
|
|
# map to parent |
|
536
|
5
|
|
|
|
|
9
|
my $parent_size = $parent_set->size; |
|
537
|
5
|
|
|
|
|
43
|
my $half_resize = int( $resize / 2 ); |
|
538
|
5
|
|
|
|
|
15
|
my $new_left_idx = $midleft_parent_idx - $half_resize + 1; |
|
539
|
5
|
50
|
|
|
|
8
|
$new_left_idx = 1 if $new_left_idx < 1; |
|
540
|
5
|
|
|
|
|
5
|
my $new_right_idx = $midright_parent_idx + $half_resize - 1; |
|
541
|
5
|
50
|
|
|
|
12
|
$new_right_idx = $parent_size if $new_right_idx > $parent_size; |
|
542
|
|
|
|
|
|
|
|
|
543
|
5
|
|
|
|
|
8
|
my $new_set = $parent_set->slice( $new_left_idx, $new_right_idx ); |
|
544
|
|
|
|
|
|
|
|
|
545
|
5
|
|
|
|
|
267
|
return $new_set; |
|
546
|
|
|
|
|
|
|
} |
|
547
|
|
|
|
|
|
|
|
|
548
|
|
|
|
|
|
|
sub center_intact_window { |
|
549
|
3
|
|
|
3
|
1
|
2022
|
my $self = shift; |
|
550
|
3
|
|
|
|
|
4
|
my AlignDB::IntSpan $comparable_set = shift; |
|
551
|
3
|
|
|
|
|
4
|
my ( $internal_start, $internal_end, $sw_size, $maximal_distance ) = @_; |
|
552
|
|
|
|
|
|
|
|
|
553
|
|
|
|
|
|
|
# if undefined, set to default values |
|
554
|
3
|
|
33
|
|
|
98
|
$sw_size ||= $self->sw_size; |
|
555
|
3
|
|
33
|
|
|
73
|
$maximal_distance ||= $self->max_out_distance; |
|
556
|
|
|
|
|
|
|
|
|
557
|
3
|
|
|
|
|
7
|
my $comparable_number = $comparable_set->size; |
|
558
|
|
|
|
|
|
|
|
|
559
|
3
|
|
|
|
|
34
|
my @center_windows; |
|
560
|
|
|
|
|
|
|
|
|
561
|
|
|
|
|
|
|
my $original_set; |
|
562
|
3
|
100
|
|
|
|
9
|
if ( $internal_start < $internal_end ) { |
|
|
|
50
|
|
|
|
|
|
|
563
|
1
|
|
|
|
|
5
|
$original_set = AlignDB::IntSpan->new("$internal_start-$internal_end"); |
|
564
|
|
|
|
|
|
|
} |
|
565
|
|
|
|
|
|
|
elsif ( $internal_start == $internal_end ) { |
|
566
|
2
|
|
|
|
|
5
|
$original_set = AlignDB::IntSpan->new($internal_start); |
|
567
|
|
|
|
|
|
|
} |
|
568
|
|
|
|
|
|
|
else { |
|
569
|
0
|
|
|
|
|
0
|
return; |
|
570
|
|
|
|
|
|
|
} |
|
571
|
|
|
|
|
|
|
|
|
572
|
3
|
|
|
|
|
119
|
my $window0_set; |
|
573
|
3
|
100
|
|
|
|
5
|
if ( $original_set->size < $sw_size ) { |
|
574
|
2
|
|
|
|
|
20
|
$window0_set = _center_resize( $original_set, $comparable_set, $sw_size ); |
|
575
|
|
|
|
|
|
|
} |
|
576
|
|
|
|
|
|
|
else { |
|
577
|
1
|
|
|
|
|
12
|
$window0_set = $original_set->intersect($comparable_set); |
|
578
|
|
|
|
|
|
|
} |
|
579
|
|
|
|
|
|
|
|
|
580
|
3
|
50
|
|
|
|
176
|
return unless $window0_set; |
|
581
|
3
|
50
|
|
|
|
41
|
return unless $window0_set->size; |
|
582
|
|
|
|
|
|
|
|
|
583
|
3
|
|
|
|
|
31
|
my $window0_start = $window0_set->min; |
|
584
|
3
|
|
|
|
|
25
|
my $window0_end = $window0_set->max; |
|
585
|
3
|
|
|
|
|
29
|
push @center_windows, |
|
586
|
|
|
|
|
|
|
{ |
|
587
|
|
|
|
|
|
|
type => 'M', |
|
588
|
|
|
|
|
|
|
set => $window0_set, |
|
589
|
|
|
|
|
|
|
distance => 0, |
|
590
|
|
|
|
|
|
|
}; |
|
591
|
|
|
|
|
|
|
|
|
592
|
3
|
|
|
|
|
3
|
for my $sw_type (qw{L R}) { |
|
593
|
|
|
|
|
|
|
|
|
594
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
595
|
6
|
|
|
|
|
6
|
my ( $sw_start, $sw_end ); |
|
596
|
|
|
|
|
|
|
|
|
597
|
6
|
100
|
|
|
|
13
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
598
|
3
|
|
|
|
|
5
|
$sw_start = $comparable_set->index($window0_end) + 1; |
|
599
|
3
|
|
|
|
|
29
|
$sw_end = $sw_start + $sw_size - 1; |
|
600
|
|
|
|
|
|
|
} |
|
601
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
602
|
3
|
|
|
|
|
5
|
$sw_end = $comparable_set->index($window0_start) - 1; |
|
603
|
3
|
|
|
|
|
32
|
$sw_start = $sw_end - $sw_size + 1; |
|
604
|
|
|
|
|
|
|
} |
|
605
|
|
|
|
|
|
|
|
|
606
|
|
|
|
|
|
|
# $sw_distance is from 1 to $maximal_distance |
|
607
|
6
|
|
|
|
|
12
|
CISW: for my $sw_distance ( 1 .. $maximal_distance ) { |
|
608
|
6
|
100
|
|
|
|
9
|
last if $sw_start < 1; |
|
609
|
5
|
50
|
|
|
|
6
|
last if $sw_end > $comparable_number; |
|
610
|
5
|
|
|
|
|
11
|
my $sw_set = $comparable_set->slice( $sw_start, $sw_end ); |
|
611
|
5
|
|
|
|
|
241
|
my $sw_set_member_number = $sw_set->size; |
|
612
|
5
|
50
|
|
|
|
43
|
if ( $sw_set_member_number < $sw_size ) { |
|
613
|
0
|
|
|
|
|
0
|
last CISW; |
|
614
|
|
|
|
|
|
|
} |
|
615
|
|
|
|
|
|
|
|
|
616
|
5
|
|
|
|
|
9
|
push @center_windows, |
|
617
|
|
|
|
|
|
|
{ |
|
618
|
|
|
|
|
|
|
type => $sw_type, |
|
619
|
|
|
|
|
|
|
set => $sw_set, |
|
620
|
|
|
|
|
|
|
distance => $sw_distance, |
|
621
|
|
|
|
|
|
|
}; |
|
622
|
|
|
|
|
|
|
|
|
623
|
5
|
100
|
|
|
|
10
|
if ( $sw_type eq 'R' ) { |
|
|
|
50
|
|
|
|
|
|
|
624
|
3
|
|
|
|
|
6
|
$sw_start = $sw_end + 1; |
|
625
|
3
|
|
|
|
|
5
|
$sw_end = $sw_start + $sw_size - 1; |
|
626
|
|
|
|
|
|
|
} |
|
627
|
|
|
|
|
|
|
elsif ( $sw_type eq 'L' ) { |
|
628
|
2
|
|
|
|
|
2
|
$sw_end = $sw_start - 1; |
|
629
|
2
|
|
|
|
|
4
|
$sw_start = $sw_end - $sw_size + 1; |
|
630
|
|
|
|
|
|
|
} |
|
631
|
|
|
|
|
|
|
} |
|
632
|
|
|
|
|
|
|
} |
|
633
|
|
|
|
|
|
|
|
|
634
|
3
|
|
|
|
|
10
|
return @center_windows; |
|
635
|
|
|
|
|
|
|
} |
|
636
|
|
|
|
|
|
|
|
|
637
|
|
|
|
|
|
|
sub strand_window { |
|
638
|
3
|
|
|
3
|
1
|
1451
|
my $self = shift; |
|
639
|
3
|
|
|
|
|
4
|
my AlignDB::IntSpan $comparable_set = shift; |
|
640
|
3
|
|
|
|
|
4
|
my ( $internal_start, $internal_end, $sw_size, $strand ) = @_; |
|
641
|
|
|
|
|
|
|
|
|
642
|
|
|
|
|
|
|
# if undefined, set to default values |
|
643
|
3
|
|
66
|
|
|
61
|
$sw_size ||= $self->sw_size; |
|
644
|
3
|
|
100
|
|
|
11
|
$strand ||= '+'; |
|
645
|
|
|
|
|
|
|
|
|
646
|
3
|
|
|
|
|
2
|
my @windows; |
|
647
|
|
|
|
|
|
|
|
|
648
|
|
|
|
|
|
|
my $working_set; |
|
649
|
3
|
100
|
|
|
|
6
|
if ( $internal_start < $internal_end ) { |
|
650
|
2
|
|
|
|
|
10
|
$working_set = $comparable_set->intersect("$internal_start-$internal_end"); |
|
651
|
|
|
|
|
|
|
} |
|
652
|
|
|
|
|
|
|
else { |
|
653
|
1
|
|
|
|
|
3
|
return @windows; |
|
654
|
|
|
|
|
|
|
} |
|
655
|
|
|
|
|
|
|
|
|
656
|
2
|
|
|
|
|
463
|
my $working_length = $working_set->size; |
|
657
|
2
|
50
|
|
|
|
26
|
return @windows if $working_length < $sw_size; |
|
658
|
|
|
|
|
|
|
|
|
659
|
|
|
|
|
|
|
# $sw_start and $sw_end are both index of $comparable_set |
|
660
|
2
|
|
|
|
|
3
|
my ( $sw_start, $sw_end ); |
|
661
|
|
|
|
|
|
|
|
|
662
|
2
|
100
|
|
|
|
7
|
if ( $strand eq '+' ) { |
|
|
|
50
|
|
|
|
|
|
|
663
|
1
|
|
|
|
|
1
|
$sw_start = 1; |
|
664
|
1
|
|
|
|
|
2
|
$sw_end = $sw_start + $sw_size - 1; |
|
665
|
|
|
|
|
|
|
} |
|
666
|
|
|
|
|
|
|
elsif ( $strand eq '-' ) { |
|
667
|
1
|
|
|
|
|
2
|
$sw_end = $working_set->size; |
|
668
|
1
|
|
|
|
|
10
|
$sw_start = $sw_end - $sw_size + 1; |
|
669
|
|
|
|
|
|
|
} |
|
670
|
|
|
|
|
|
|
else { |
|
671
|
0
|
|
|
|
|
0
|
return @windows; |
|
672
|
|
|
|
|
|
|
} |
|
673
|
|
|
|
|
|
|
|
|
674
|
2
|
|
|
|
|
4
|
my $available_distance = int( $working_length / ($sw_size) ); |
|
675
|
|
|
|
|
|
|
|
|
676
|
|
|
|
|
|
|
# sw_distance is from 1 to max_distance |
|
677
|
2
|
|
|
|
|
3
|
SSW: for my $i ( 1 .. $available_distance ) { |
|
678
|
6
|
|
|
|
|
11
|
my $sw_set = $working_set->slice( $sw_start, $sw_end ); |
|
679
|
6
|
|
|
|
|
366
|
my $sw_set_member_number = $sw_set->size; |
|
680
|
6
|
50
|
|
|
|
59
|
if ( $sw_set_member_number < $sw_size ) { |
|
681
|
0
|
|
|
|
|
0
|
last SSW; |
|
682
|
|
|
|
|
|
|
} |
|
683
|
6
|
|
|
|
|
3
|
my $sw_distance = $i; |
|
684
|
|
|
|
|
|
|
|
|
685
|
6
|
|
|
|
|
6
|
my %window_info; |
|
686
|
6
|
|
|
|
|
9
|
$window_info{type} = $strand; |
|
687
|
6
|
|
|
|
|
5
|
$window_info{set} = $sw_set; |
|
688
|
6
|
|
|
|
|
20
|
$window_info{distance} = $sw_distance; |
|
689
|
|
|
|
|
|
|
|
|
690
|
6
|
|
|
|
|
7
|
push @windows, \%window_info; |
|
691
|
|
|
|
|
|
|
|
|
692
|
6
|
100
|
|
|
|
13
|
if ( $strand eq '+' ) { |
|
|
|
50
|
|
|
|
|
|
|
693
|
5
|
|
|
|
|
5
|
$sw_start = $sw_end + 1; |
|
694
|
5
|
|
|
|
|
7
|
$sw_end = $sw_start + $sw_size - 1; |
|
695
|
|
|
|
|
|
|
} |
|
696
|
|
|
|
|
|
|
elsif ( $strand eq '-' ) { |
|
697
|
1
|
|
|
|
|
2
|
$sw_end = $sw_start - 1; |
|
698
|
1
|
|
|
|
|
2
|
$sw_start = $sw_end - $sw_size + 1; |
|
699
|
|
|
|
|
|
|
} |
|
700
|
|
|
|
|
|
|
} |
|
701
|
|
|
|
|
|
|
|
|
702
|
2
|
|
|
|
|
6
|
return @windows; |
|
703
|
|
|
|
|
|
|
} |
|
704
|
|
|
|
|
|
|
|
|
705
|
|
|
|
|
|
|
1; # Magic true value required at end of module |
|
706
|
|
|
|
|
|
|
|
|
707
|
|
|
|
|
|
|
__END__ |
|
708
|
|
|
|
|
|
|
|
|
709
|
|
|
|
|
|
|
=pod |
|
710
|
|
|
|
|
|
|
|
|
711
|
|
|
|
|
|
|
=encoding UTF-8 |
|
712
|
|
|
|
|
|
|
|
|
713
|
|
|
|
|
|
|
=head1 NAME |
|
714
|
|
|
|
|
|
|
|
|
715
|
|
|
|
|
|
|
AlignDB::Window - Split integer spans into a series of windows |
|
716
|
|
|
|
|
|
|
|
|
717
|
|
|
|
|
|
|
=head1 DESCRIPTION |
|
718
|
|
|
|
|
|
|
|
|
719
|
|
|
|
|
|
|
AlignDB::Window provides methods to split integer spans, including interval, inside area and |
|
720
|
|
|
|
|
|
|
outside area, into a series of windows. |
|
721
|
|
|
|
|
|
|
|
|
722
|
|
|
|
|
|
|
=head1 ATTRIBUTES |
|
723
|
|
|
|
|
|
|
|
|
724
|
|
|
|
|
|
|
C<sw_size> - sliding windows' size, default is 100 |
|
725
|
|
|
|
|
|
|
|
|
726
|
|
|
|
|
|
|
C<min_interval> - mininal indel interval length, default is 11 |
|
727
|
|
|
|
|
|
|
|
|
728
|
|
|
|
|
|
|
C<max_out_distance> - maximal outside distance, default is 10 |
|
729
|
|
|
|
|
|
|
|
|
730
|
|
|
|
|
|
|
C<max_in_distance> - maximal inside distance, default is 5 |
|
731
|
|
|
|
|
|
|
|
|
732
|
|
|
|
|
|
|
=head1 METHODS |
|
733
|
|
|
|
|
|
|
|
|
734
|
|
|
|
|
|
|
=head2 common parameters |
|
735
|
|
|
|
|
|
|
|
|
736
|
|
|
|
|
|
|
C<$comparable_set> - AlignDB::IntSpan object |
|
737
|
|
|
|
|
|
|
|
|
738
|
|
|
|
|
|
|
C<$interval_start> - start point of the interval |
|
739
|
|
|
|
|
|
|
|
|
740
|
|
|
|
|
|
|
C<$interval_end> - end point in the interval |
|
741
|
|
|
|
|
|
|
|
|
742
|
|
|
|
|
|
|
C<$internal_start> - start position in the internal region |
|
743
|
|
|
|
|
|
|
|
|
744
|
|
|
|
|
|
|
C<$internal_end> - end position in the internal region |
|
745
|
|
|
|
|
|
|
|
|
746
|
|
|
|
|
|
|
C<$sw_size> - size of windows |
|
747
|
|
|
|
|
|
|
|
|
748
|
|
|
|
|
|
|
C<$min_interval> - minimal size of intervals |
|
749
|
|
|
|
|
|
|
|
|
750
|
|
|
|
|
|
|
C<$maximal_distance> - maximal distance |
|
751
|
|
|
|
|
|
|
|
|
752
|
|
|
|
|
|
|
C<$strand> - '+' or '-' |
|
753
|
|
|
|
|
|
|
|
|
754
|
|
|
|
|
|
|
=head2 interval_window |
|
755
|
|
|
|
|
|
|
|
|
756
|
|
|
|
|
|
|
my @interval_windows = $self->interval_window( |
|
757
|
|
|
|
|
|
|
$comparable_set, $interval_start, $interval_end, |
|
758
|
|
|
|
|
|
|
$sw_size, $min_interval, |
|
759
|
|
|
|
|
|
|
); |
|
760
|
|
|
|
|
|
|
|
|
761
|
|
|
|
|
|
|
Split an interval to windows. |
|
762
|
|
|
|
|
|
|
|
|
763
|
|
|
|
|
|
|
Length of windows are variable, but all positions of the interval are counted. |
|
764
|
|
|
|
|
|
|
|
|
765
|
|
|
|
|
|
|
=head2 outside_window |
|
766
|
|
|
|
|
|
|
|
|
767
|
|
|
|
|
|
|
my @outside_windows = $self->outside_window( |
|
768
|
|
|
|
|
|
|
$comparable_set, $internal_start, $internal_end, |
|
769
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
770
|
|
|
|
|
|
|
); |
|
771
|
|
|
|
|
|
|
|
|
772
|
|
|
|
|
|
|
Draw outside windows from a internal region. |
|
773
|
|
|
|
|
|
|
|
|
774
|
|
|
|
|
|
|
All windows are 100 bp length. Start from 1 and end to $maximal_distance. |
|
775
|
|
|
|
|
|
|
|
|
776
|
|
|
|
|
|
|
=head2 outside_window_2 |
|
777
|
|
|
|
|
|
|
|
|
778
|
|
|
|
|
|
|
my @outside_windows = $self->outside_window_2( |
|
779
|
|
|
|
|
|
|
$comparable_set, $internal_start, $internal_end, |
|
780
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
781
|
|
|
|
|
|
|
); |
|
782
|
|
|
|
|
|
|
|
|
783
|
|
|
|
|
|
|
Draw outside windows from a internal region. |
|
784
|
|
|
|
|
|
|
|
|
785
|
|
|
|
|
|
|
The first window is 50 bp and all others are 100 bp length. |
|
786
|
|
|
|
|
|
|
Start from 0 and end to $maximal_distance. |
|
787
|
|
|
|
|
|
|
|
|
788
|
|
|
|
|
|
|
=head2 inside_window |
|
789
|
|
|
|
|
|
|
|
|
790
|
|
|
|
|
|
|
my @inside_windows = $self->inside_window( |
|
791
|
|
|
|
|
|
|
$comparable_set, $interval_start, $interval_end, |
|
792
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
793
|
|
|
|
|
|
|
); |
|
794
|
|
|
|
|
|
|
|
|
795
|
|
|
|
|
|
|
Draw inside windows from a internal region. |
|
796
|
|
|
|
|
|
|
|
|
797
|
|
|
|
|
|
|
All windows are 100 bp length. Start counting from the edges. |
|
798
|
|
|
|
|
|
|
|
|
799
|
|
|
|
|
|
|
=head2 inside_window_2 |
|
800
|
|
|
|
|
|
|
|
|
801
|
|
|
|
|
|
|
my @inside_windows = $self->inside_window_2( |
|
802
|
|
|
|
|
|
|
$comparable_set, $interval_start, $interval_end, |
|
803
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
804
|
|
|
|
|
|
|
); |
|
805
|
|
|
|
|
|
|
|
|
806
|
|
|
|
|
|
|
Draw inside windows from a internal region. |
|
807
|
|
|
|
|
|
|
|
|
808
|
|
|
|
|
|
|
All windows are 100 bp length. Start counting from the center. |
|
809
|
|
|
|
|
|
|
|
|
810
|
|
|
|
|
|
|
=head2 center_window |
|
811
|
|
|
|
|
|
|
|
|
812
|
|
|
|
|
|
|
my @center_windows = $self->center_window( |
|
813
|
|
|
|
|
|
|
$comparable_set, $internal_start, $internal_end, |
|
814
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
815
|
|
|
|
|
|
|
); |
|
816
|
|
|
|
|
|
|
|
|
817
|
|
|
|
|
|
|
Draw windows for a certain region. |
|
818
|
|
|
|
|
|
|
|
|
819
|
|
|
|
|
|
|
Center is 0, and the first window is 50 bp and all others are 100 bp length. |
|
820
|
|
|
|
|
|
|
Start from 0 and end to $maximal_distance. |
|
821
|
|
|
|
|
|
|
|
|
822
|
|
|
|
|
|
|
=head2 center_intact_window |
|
823
|
|
|
|
|
|
|
|
|
824
|
|
|
|
|
|
|
my @center_intact_windows = $self->center_intact_window( |
|
825
|
|
|
|
|
|
|
$comparable_set, $internal_start, $internal_end, |
|
826
|
|
|
|
|
|
|
$sw_size, $maximal_distance, |
|
827
|
|
|
|
|
|
|
); |
|
828
|
|
|
|
|
|
|
|
|
829
|
|
|
|
|
|
|
Draw windows for a certain region. |
|
830
|
|
|
|
|
|
|
|
|
831
|
|
|
|
|
|
|
Center is 0, and the first window is 50 bp and all others are 100 bp length. |
|
832
|
|
|
|
|
|
|
Start from 0 and end to $maximal_distance. |
|
833
|
|
|
|
|
|
|
|
|
834
|
|
|
|
|
|
|
=head2 strand_window |
|
835
|
|
|
|
|
|
|
|
|
836
|
|
|
|
|
|
|
my @windows = $self->strand_window( |
|
837
|
|
|
|
|
|
|
$comparable_set, $interval_start, $interval_end, |
|
838
|
|
|
|
|
|
|
$sw_size, $strand, |
|
839
|
|
|
|
|
|
|
); |
|
840
|
|
|
|
|
|
|
|
|
841
|
|
|
|
|
|
|
Draw windows for a certain region |
|
842
|
|
|
|
|
|
|
|
|
843
|
|
|
|
|
|
|
=head1 AUTHOR |
|
844
|
|
|
|
|
|
|
|
|
845
|
|
|
|
|
|
|
Qiang Wang <wang-q@outlook.com> |
|
846
|
|
|
|
|
|
|
|
|
847
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
|
848
|
|
|
|
|
|
|
|
|
849
|
|
|
|
|
|
|
This software is copyright (c) 2008 by Qiang Wang. |
|
850
|
|
|
|
|
|
|
|
|
851
|
|
|
|
|
|
|
This is free software; you can redistribute it and/or modify it under |
|
852
|
|
|
|
|
|
|
the same terms as the Perl 5 programming language system itself. |
|
853
|
|
|
|
|
|
|
|
|
854
|
|
|
|
|
|
|
=cut |