annotate variant_effect_predictor/Bio/Assembly/Scaffold.pm @ 1:d6778b5d8382 draft default tip

Deleted selected files
author willmclaren
date Fri, 03 Aug 2012 10:05:43 -0400
parents 21066c0abaf5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
1 # $Id: Scaffold.pm,v 1.2 2002/11/11 18:16:30 lapp Exp $
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 # BioPerl module for Bio::Assembly::Scaffold
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5 # Copyright by Robson F. de Souza
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
6 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
7 # You may distribute this module under the same terms as perl itself
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
8 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
9 # POD documentation - main docs before the code
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
10
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
11 =head1 NAME
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
12
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
13 Bio::Assembly::Scaffold - Perl module to hold and manipulate sequence assembly data.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15 =head1 SYNOPSYS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 # Module loading
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18 use Bio::Assembly::IO;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20 # Assembly loading methods
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 my $aio = new Bio::Assembly::IO(-file=>"test.ace.1", -format=>'phrap');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22 my $assembly = $aio->next_assembly;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24 foreach my $contig ($assembly->all_contigs) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 # do something... (see Bio::Assembly::Contig)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28 =head1 DESCRIPTION
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30 Bio::Assembly::Scaffold was developed to store and manipulate data
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 from sequence assembly programs like Phrap. It implements the
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32 ScaffoldI interface and intends to be generic enough to be used by
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33 Bio::Assembly::IO drivers written to programs other than Phrap.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35 =head1 FEEDBACK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37 =head2 Mailing Lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39 User feedback is an integral part of the evolution of this and other
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40 Bioperl modules. Send your comments and suggestions preferably to the
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41 Bioperl mailing lists Your participation is much appreciated.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43 bioperl-l@bioperl.org - General discussion
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44 http://bio.perl.org/MailList.html - About the mailing lists
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46 =head2 Reporting Bugs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48 Report bugs to the Bioperl bug tracking system to help us keep track
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49 the bugs and their resolution. Bug reports can be submitted via email
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50 or the web:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52 bioperl-bugs@bio.perl.org
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53 http://bugzilla.bioperl.org/
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55 =head1 AUTHOR - Robson Francisco de Souza
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57 rfsouza@citri.iq.usp.br
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59 =head1 APPENDIX
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61 The rest of the documentation details each of the object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62 methods. Internal methods are usually preceded with a _
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66 package Bio::Assembly::Scaffold;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69 use vars qw(@ISA $VERSION);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71 use Bio::Root::Root;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72 use Bio::Assembly::ScaffoldI;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73 use Bio::Annotation::Collection;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75 $VERSION = '0.0.1';
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76 @ISA = qw(Bio::Root::Root Bio::Assembly::ScaffoldI);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78 =head2 new ()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 Title : new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 Usage : $assembly = new (-source=>'program_name',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82 -contigs=>\@contigs,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 -id=>"assembly 1");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84 Function: creates a new assembly object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85 Returns :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86 Args :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87 -source : [string] sequence assembly program
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88 -contigs : reference to array of
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89 Bio::Assembly::Contig objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 -id : [string] assembly name
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94 sub new {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95 my($class,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97 my $self = $class->SUPER::new(@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99 my ($src,$contigs,$id) = $self->_rearrange([qw(SOURCE CONTIGS ID)], @args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101 $self->{'_contigs'} = {};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102 $self->{'_singlets'} = {};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103 $self->{'_seqs'} = {};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104 $self->{'_annotation'} = Bio::Annotation::Collection->new();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105 $self->{'_id'} = 'NoName';
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 if (defined $contigs && ref($contigs = 'ARRAY')) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108 foreach my $contig (@{$contigs}) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109 $self->add_contig($contig);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113 $self->{'_id'} = $id if (defined $id);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 return $self;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118 =head1 Accessing general assembly data
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 =head2 id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 Title : id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 Usage : $assembly->id()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126 Function: Get/Set assembly ID
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 Returns : string or undef
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128 Args : string
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 sub id {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
133 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
134 my $id = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
135
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
136 $self->{'_id'} = $id if (defined $id);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
138 return $self->{'_id'};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
139 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
140
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
141 =head2 annotation
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
142
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
143 Title : annotation
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
144 Usage : $assembly->annotation()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
145 Function: Get/Set assembly annotation object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
146 Returns : Bio::Annotation::Collection
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
147 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
148
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
149 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
150
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
151 sub annotation {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
152 my ($self,$ref) = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
153
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
154 $self->{'_annotation'} = $ref if (defined $ref);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
155 return $self->{'_annotation'};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
156 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
157
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
158 =head2 get_nof_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
159
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
160 Title : get_nof_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
161 Usage : $assembly->get_nof_contigs()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
162 Function: Get the number of contigs included in the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
163 Returns : integer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
164 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
165
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
166 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
167
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
168 sub get_nof_contigs {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
169 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
170
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
171 return scalar( $self->get_contig_ids() );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
172 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
173
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
174 =head2 get_nof_sequences_in_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
175
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
176 Title : get_nof_sequences_in_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
177 Usage : $assembly->get_nof_sequences_in_contigs()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
178 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
179
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
180 Get the number of sequences included in the
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
181 assembly. This number refers only to the sequences used
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
182 to build contigs in this assembly. It does not includes
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
183 contig consensus sequences or singlets.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
184
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
185 Returns : integer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
186 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
187
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
188 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
189
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
190 sub get_nof_sequences_in_contigs {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
191 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
192
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
193 my $nof_seqs = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
194 foreach my $contig ($self->all_contigs) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
195 $nof_seqs += scalar( $contig->get_seq_ids() );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
196 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
197
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
198 return $nof_seqs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
199 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
200
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
201 =head2 get_nof_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
202
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
203 Title : nof_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
204 Usage : $assembly->nof_singlets()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
205 Function: Get the number of singlets included in the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
206 Returns : integer
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
207 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
208
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
209 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
210
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
211 sub get_nof_singlets {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
212 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
213
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
214 return scalar( $self->get_singlet_ids() );
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
215 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
216
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
217 =head2 get_seq_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
218
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
219 Title : get_seq_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
220 Usage : $assembly->get_seq_ids()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
221 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
222
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
223 Get the ID of sequences from all contigs. This list
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
224 refers only to the aligned sequences in contigs. It does
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
225 not includes contig consensus sequences or singlets.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
226
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
227 Returns : array of strings
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
228 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
229
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
230 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
231
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
232 sub get_seq_ids {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
233 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
234
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
235 return keys %{ $self->{'_seqs'} };
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
236 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
237
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
238 =head2 get_contig_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
239
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
240 Title : get_contig_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
241 Usage : $assembly->get_contig_ids()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
242 Function: Access list of contig IDs from assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
243 Returns : an array, if there are any contigs in the
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
244 assembly. An empty array otherwise
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
245 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
246
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
247 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
248
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
249 sub get_contig_ids {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
250 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
251
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
252 return sort keys %{$self->{'_contigs'}};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
253 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
254
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
255 =head2 get_singlet_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
256
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
257 Title : get_singlet_ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
258 Usage : $assembly->get_singlet_ids()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
259 Function: Access list of singlet IDs from assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
260 Returns : array of strings if there are any singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
261 otherwise an empty array
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
262 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
263
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
264 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
265
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
266 sub get_singlet_ids {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
267 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
268
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
269 return sort keys %{$self->{'_singlets'}};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
270 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
271
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
272 =head2 get_seq_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
273
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
274 Title : get_seq_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
275 Usage : $assembly->get_seq_by_id($id)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
276 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
277
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
278 Get a reference for an aligned sequence
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
279 This sequence must be part of a contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
280 in the assembly.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
281
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
282 Returns : a Bio::LocatableSeq object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
283 undef if sequence $id is not found
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
284 in any contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
285 Args : [string] sequence identifier (id)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
286
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
287 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
288
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
289 sub get_seq_by_id {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
290 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
291 my $seqID = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
292
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
293 return undef unless (exists $self->{'_seqs'}{$seqID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
294
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
295 return $self->{'_seqs'}{$seqID}->get_seq_by_name($seqID);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
296 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
297
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
298 =head2 get_contig_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
299
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
300 Title : get_contig_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
301 Usage : $assembly->get_contig_by_id($id)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
302 Function: Get a reference for a contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
303 Returns : a Bio::Assembly::Contig object or undef
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
304 Args : [string] contig unique identifier (ID)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
305
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
306 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
307
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
308 sub get_contig_by_id {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
309 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
310 my $contigID = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
311
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
312 return undef unless (exists $self->{'_contigs'}{$contigID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
313
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
314 return $self->{'_contigs'}{$contigID};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
315 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
316
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
317 =head2 get_singlet_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
318
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
319 Title : get_singlet_by_id
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
320 Usage : $assembly->get_singlet_by_id()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
321 Function: Get a reference for a singlet
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
322 Returns : Bio::PrimarySeqI object or undef
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
323 Args : [string] a singlet ID
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
324
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
325 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
326
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
327 sub get_singlet_by_id {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
328 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
329
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
330 my $singletID = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
331
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
332 return undef unless (exists $self->{'_singlets'}{$singletID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
333
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
334 return $self->{'_singlets'}{$singletID};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
335 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
336
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
337 =head1 Modifier methods
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
338
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
339 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
340
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
341 =head2 add_contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
342
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
343 Title : add_contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
344 Usage : $assembly->add_contig($contig)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
345 Function: Add a contig to the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
346 Returns : 1 on success
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
347 Args : a Bio::Assembly::Contig object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
348 order (optional)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
349
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
350 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
351
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
352 sub add_contig {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
353 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
354 my $contig = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
355
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
356 if( !ref $contig || ! $contig->isa('Bio::Assembly::Contig') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
357 $self->throw("Unable to process non Bio::Assembly::Contig object [", ref($contig), "]");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
358 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
359 my $contigID = $contig->id();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
360 if( !defined $contigID ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
361 $contigID = 'Unknown_' . ($self->get_nof_contigs() + 1);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
362 $contig->id($contigID);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
363 $self->warn("Attributing ID $contigID to unidentified Bio::Assembly::Contig object.");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
364 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
365
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
366 $self->warn("Replacing contig $contigID with a new contig object")
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
367 if (exists $self->{'_contigs'}{$contigID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
368 $self->{'_contigs'}{$contigID} = $contig;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
369
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
370 foreach my $seqID ($contig->get_seq_ids()) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
371 if (exists $self->{'_seqs'}{$seqID}) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
372 $self->warn( "Sequence $seqID already assigned to contig ".
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
373 $self->{'_seqs'}{$seqID}->id().". Moving to contig $contigID")
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
374 unless ($self->{'_seqs'}{$seqID} eq $contig);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
375 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
376 $self->{'_seqs'}{$seqID} = $contig;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
377 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
378
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
379 return 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
380 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
381
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
382 =head2 add_singlet
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
383
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
384 Title : add_singlet
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
385 Usage : $assembly->add_singlet($seq)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
386 Function: Add a singlet to the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
387 Returns : 1 on success, 0 otherwise
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
388 Args : a Bio::PrimarySeqI object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
389 order (optional)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
390
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
391 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
392
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
393 sub add_singlet {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
394 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
395 my $singlet = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
396
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
397 if( !ref $singlet || ! $singlet->isa('Bio::PrimarySeqI') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
398 $self->warn("Unable to process non Bio::SeqI object [", ref($singlet), "]");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
399 return 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
400 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
401
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
402 my $singletID = $singlet->id();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
403 $self->warn("Replacing singlet $singletID wih a new sequence object")
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
404 if (exists $self->{'_contigs'}{$singletID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
405 $self->{'_singlets'}{$singletID} = $singlet;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
406
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
407 return 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
408 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
409
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
410 =head2 update_seq_list
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
411
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
412 Title : update_seq_list
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
413 Usage : $assembly->update_seq_list()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
414 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
415
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
416 Synchronizes the assembly registry for sequences in
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
417 contigs and contig actual aligned sequences content. You
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
418 probably want to run this after you remove/add a
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
419 sequence from/to a contig in the assembly.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
420
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
421 Returns : nothing
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
422 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
423
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
424 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
425
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
426 sub update_seq_list {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
427 my $self = shift;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
428
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
429 $self->{'_seqs'} = {};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
430 foreach my $contig ($self->all_contigs) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
431 foreach my $seqID ($contig->get_seq_ids) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
432 $self->{'_seqs'}{$seqID} = $contig;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
433 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
434 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
435
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
436 return 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
437 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
438
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
439 =head2 remove_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
440
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
441 Title : remove_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
442 Usage : $assembly->remove_contigs(1..4)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
443 Function: Remove contig from assembly object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
444 Returns : an array of removed Bio::Assembly::Contig
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
445 objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
446 Args : an array of contig IDs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
447
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
448 See function get_contig_ids() above
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
449
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
450 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
451
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
452 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
453 sub remove_contigs {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
454 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
455 my ($self,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
456
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
457 my @ret = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
458 foreach my $contigID (@args) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
459 foreach my $seqID ($self->get_contig_by_id($contigID)->get_seq_ids()) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
460 delete $self->{'_seqs'}{$seqID};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
461 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
462 push(@ret,$self->{'_contigs'}{$contigID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
463 delete $self->{'_contigs'}{$contigID};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
464 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
465
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
466 return @ret;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
467 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
468
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
469 =head2 remove_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
470
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
471 Title : remove_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
472 Usage : $assembly->remove_singlets(@singlet_ids)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
473 Function: Remove singlet from assembly object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
474 Returns : the Bio::SeqI objects removed
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
475 Args : a list of singlet IDs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
476
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
477 See function get_singlet_ids() above
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
478
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
479 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
480
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
481 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
482 sub remove_singlets {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
483 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
484 my ($self,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
485
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
486 my @ret = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
487 foreach my $singletID (@args) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
488 push(@ret,$self->{'_singlets'}{$singletID});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
489 delete $self->{'_singlets'}{$singletID};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
490 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
491
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
492 return @ret;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
493 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
494
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
495 =head1 Contig and singlet selection methos
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
496
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
497 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
498
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
499 =head2 select_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
500
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
501 Title : select_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
502 Usage : $assembly->select_contigs(@list)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
503 Function: Select an array of contigs from the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
504 Returns : an array of Bio::Assembly::Contig objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
505 Args : an array of contig ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
506
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
507 See function get_contig_ids() above
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
508
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
509 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
510
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
511 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
512 sub select_contigs {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
513 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
514 my ($self,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
515
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
516 my @contigs = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
517 foreach my $contig (@args) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
518 unless (exists $self->{'_contigs'}{$contig}) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
519 $self->warn("$contig contig not found. Ignoring...");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
520 next;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
521 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
522 push(@contigs, $self->{'_contigs'}{$contig});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
523 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
524
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
525 return @contigs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
526 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
527
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
528 =head2 select_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
529
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
530 Title : select_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
531 Usage : $assembly->select_singlets(@list)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
532 Function: Selects an array of singlets from the assembly
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
533 Returns : an array of Bio::SeqI objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
534 Args : an array of singlet ids
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
535
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
536 See function get_singlet_ids() above
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
537
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
538 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
539
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
540 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
541 sub select_singlets {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
542 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
543 my ($self,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
544
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
545 my @singlets = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
546 foreach my $singlet (@args) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
547 unless (exists $self->{'_singlets'}{$singlet}) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
548 $self->warn("$singlet singlet not found. Ignoring...");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
549 next;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
550 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
551 push(@singlets, $self->{'_singlets'}{$singlet});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
552 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
553
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
554 return @singlets;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
555 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
556
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
557 =head2 all_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
558
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
559 Title : all_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
560 Usage : my @contigs = $assembly->all_contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
561 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
562
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
563 Returns a list of all contigs in this assembly. Contigs
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
564 are both clusters and alignments of one or more reads,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
565 with an associated consensus sequence.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
566
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
567 Returns : array of Bio::Assembly::Contig (in lexical id order)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
568 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
569
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
570 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
571
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
572 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
573 sub all_contigs {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
574 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
575 my ($self) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
576
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
577 my @contigs = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
578 foreach my $contig (sort { $a cmp $b } keys %{ $self->{'_contigs'} }) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
579 push(@contigs, $self->{'_contigs'}{$contig});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
580 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
581
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
582 return @contigs;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
583 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
584
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
585 =head2 all_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
586
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
587 Title : all_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
588 Usage : my @singlets = $assembly->all_singlets
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
589 Function:
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
590
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
591 Returns a list of all singlets in this assembly.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
592 Singlets are isolated reads, without non-vector
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
593 matches to any other read in the assembly.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
594
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
595 Returns : array of Bio::SeqI (in lexical order by id)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
596 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
597
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
598 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
599
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
600 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
601 sub all_singlets {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
602 #---------------------
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
603 my ($self) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
604
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
605 my @singlets = ();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
606 foreach my $singlet (sort { $a cmp $b } keys %{ $self->{'_singlets'} }) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
607 push(@singlets, $self->{'_singlets'}{$singlet});
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
608 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
609
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
610 return @singlets;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
611 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
612
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
613 # =head1 Internal Methods
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
614
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
615 1;