annotate variant_effect_predictor/Bio/Taxonomy/Tree.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: Tree.pm,v 1.1 2002/11/18 22:08:33 kortsch Exp $
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
2 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
3 # BioPerl module for Bio::Taxonomy::Tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
4 #
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
5 # Cared for by Dan Kortschak but pilfered extensively from Bio::Tree::Tree by Jason Stajich
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::Taxonomy::Tree - An Organism Level Implementation of TreeI interface.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
14
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
15 =head1 SYNOPSIS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
16
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
17 # like from a TreeIO
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
18 my $treeio = new Bio::TreeIO(-format => 'newick', -file => 'treefile.dnd');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
19 my $tree = $treeio->next_tree;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
20 my @nodes = $tree->get_nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
21 my $root = $tree->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
22 my @leaves = $tree->get_leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
23
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
24
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
25 =head1 DESCRIPTION
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
26
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
27 This object holds handles to Taxonomic Nodes which make up a tree.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
28
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
29 =head1 EXAMPLES
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
30
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
31 use Bio::Species;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
32 use Bio::Taxonomy::Tree;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
33
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
34 my $human=new Bio::Species;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
35 my $chimp=new Bio::Species;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
36 my $bonobo=new Bio::Species;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
37
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
38 $human->classification(qw( sapiens Homo Hominidae
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
39 Catarrhini Primates Eutheria
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
40 Mammalia Euteleostomi Vertebrata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
41 Craniata Chordata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
42 Metazoa Eukaryota ));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
43 $chimp->classification(qw( troglodytes Pan Hominidae
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
44 Catarrhini Primates Eutheria
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
45 Mammalia Euteleostomi Vertebrata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
46 Craniata Chordata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
47 Metazoa Eukaryota ));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
48 $bonobo->classification(qw( paniscus Pan Hominidae
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
49 Catarrhini Primates Eutheria
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
50 Mammalia Euteleostomi Vertebrata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
51 Craniata Chordata
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
52 Metazoa Eukaryota ));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
53
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
54 # ranks passed to $taxonomy match ranks of species
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
55 my @ranks = ('superkingdom','kingdom','phylum','subphylum',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
56 'no rank 1','no rank 2','class','no rank 3','order',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
57 'suborder','family','genus','species');
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
58
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
59 my $taxonomy=new Bio::Taxonomy(-ranks => \@ranks,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
60 -method => 'trust',
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
61 -order => -1);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
62
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
63 my @nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
64
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
65 my $tree1=new Bio::Taxonomy::Tree;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
66 my $tree2=new Bio::Taxonomy::Tree;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
67
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
68 push @nodes,$tree1->make_species_branch($human,$taxonomy);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
69 push @nodes,$tree2->make_species_branch($chimp,$taxonomy);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
70
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
71 my ($homo_sapiens)=$tree1->get_leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
72
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
73 $tree1->splice($tree2);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
74
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
75 push @nodes,$tree1->add_species($bonobo,$taxonomy);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
76
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
77 my @taxa;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
78 foreach my $leaf ($tree1->get_leaves) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
79 push @taxa,$leaf->taxon;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
80 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
81 print join(", ",@taxa)."\n";
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
82
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
83 @taxa=();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
84 $tree1->remove_branch($homo_sapiens);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
85 foreach my $leaf ($tree1->get_leaves) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
86 push @taxa,$leaf->taxon;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
87 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
88 print join(", ",@taxa)."\n";
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
89
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
90 =head1 FEEDBACK
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
91
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
92 See AUTHOR
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
93
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
94 =head1 AUTHOR - Dan Kortschak
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
95
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
96 Email kortschak@rsbs.anu.edu.au
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
97
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
98 =head1 CONTRIBUTORS
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
99
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
100 Mainly Jason Stajich
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
101
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
102 =head1 APPENDIX
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
103
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
104 The rest of the documentation details each of the object methods.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
105 Internal methods are usually preceded with a _
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
106
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
107 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
108
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
109
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
110 # Code begins...
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
111
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
112
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
113 package Bio::Taxonomy::Tree;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
114 use vars qw(@ISA);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
115 use strict;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
116
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
117 # Object preamble - inherits from Bio::Root::Root
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
118
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
119 use Bio::Root::Root;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
120 use Bio::Tree::TreeFunctionsI;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
121 use Bio::Tree::TreeI;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
122 use Bio::Taxonomy::Taxon;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
123
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
124 # Import rank information from Bio::Taxonomy.pm
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
125 use vars qw(@RANK %RANK);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
126
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
127 @ISA = qw(Bio::Root::Root Bio::Tree::TreeI Bio::Tree::TreeFunctionsI);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
128
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
129 =head2 new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
130
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
131 Title : new
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
132 Usage : my $obj = new Bio::Taxonomy::Tree();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
133 Function: Builds a new Bio::Taxonomy::Tree object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
134 Returns : Bio::Taxonomy::Tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
135 Args :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
136
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
137
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
138 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
139
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
140 sub new {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
141 my($class,@args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
142
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
143 my $self = $class->SUPER::new(@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
144 $self->{'_rootnode'} = undef;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
145 $self->{'_maxbranchlen'} = 0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
146
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
147 my ($root)= $self->_rearrange([qw(ROOT)], @args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
148 if( $root ) { $self->set_root_node($root); }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
149 return $self;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
150 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
151
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
152
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
153 =head2 get_nodes
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
154
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
155 Title : get_nodes
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
156 Usage : my @nodes = $tree->get_nodes()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
157 Function: Return list of Bio::Taxonomy::Taxon objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
158 Returns : array of Bio::Taxonomy::Taxon objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
159 Args : (named values) hash with one value
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
160 order => 'b|breadth' first order or 'd|depth' first order
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
161
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
162 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
163
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
164 sub get_nodes{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
165 my ($self, @args) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
166
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
167 my ($order, $sortby) = $self->_rearrange([qw(ORDER SORTBY)],@args);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
168 $order ||= 'depth';
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
169 $sortby ||= 'height';
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
170
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
171 if ($order =~ m/^b|(breadth)$/oi) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
172 my $node = $self->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
173 my @children = ($node);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
174 for (@children) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
175 push @children, $_->each_Descendent($sortby);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
176 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
177 return @children;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
178 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
179
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
180 if ($order =~ m/^d|(depth)$/oi) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
181 # this is depth-first search I believe
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
182 my $node = $self->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
183 my @children = ($node,$node->get_Descendents($sortby));
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
184 return @children;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
185 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
186 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
187
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
188 =head2 get_root_node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
189
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
190 Title : get_root_node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
191 Usage : my $node = $tree->get_root_node();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
192 Function: Get the Top Node in the tree, in this implementation
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
193 Trees only have one top node.
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
194 Returns : Bio::Taxonomy::Taxon object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
195 Args : none
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
196
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
197 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
198
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
199
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
200 sub get_root_node{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
201 my ($self) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
202 return $self->{'_rootnode'};
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
203 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
204
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
205 =head2 set_root_node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
206
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
207 Title : set_root_node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
208 Usage : $tree->set_root_node($node)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
209 Function: Set the Root Node for the Tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
210 Returns : Bio::Taxonomy::Taxon
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
211 Args : Bio::Taxonomy::Taxon
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
212
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
213 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
214
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
215
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
216 sub set_root_node{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
217 my ($self,$value) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
218 if( defined $value ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
219 if( ! $value->isa('Bio::Taxonomy::Taxon') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
220 $self->warn("Trying to set the root node to $value which is not a Bio::Taxonomy::Taxon");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
221 return $self->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
222 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
223 $self->{'_rootnode'} = $value;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
224 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
225 return $self->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
226 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
227
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
228
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
229 =head2 get_leaves
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
230
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
231 Title : get_leaves
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
232 Usage : my @nodes = $tree->get_leaves()
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
233 Function: Return list of Bio::Taxonomy::Taxon objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
234 Returns : array of Bio::Taxonomy::Taxon objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
235 Args :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
236
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
237 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
238
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
239
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
240 sub get_leaves{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
241 my ($self) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
242
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
243 my $node = $self->get_root_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
244 my @leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
245 my @children = ($node);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
246 for (@children) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
247 push @children, $_->each_Descendent();
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
248 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
249 for (@children) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
250 push @leaves, $_ if $_->is_Leaf;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
251 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
252 return @leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
253 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
254
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
255 =head2 make_species_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
256
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
257 Title : make_species_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
258 Usage : @nodes = $tree->make_species_branch($species,$taxonomy)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
259 Function: Return list of Bio::Taxonomy::Taxon objects based on a Bio::Species object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
260 Returns : array of Bio::Taxonomy::Taxon objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
261 Args : Bio::Species and Bio::Taxonomy objects
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
262
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
263 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
264
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
265 # I'm not happy that make_species_branch and make_branch are seperate routines
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
266 # should be able to just make_branch and have it sort things out
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
267
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
268 sub make_species_branch{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
269 my ($self,$species,$taxonomy) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
270
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
271 if (! $species->isa('Bio::Species') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
272 $self->throw("Trying to classify $species which is not a Bio::Species object");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
273 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
274 if (! $taxonomy->isa('Bio::Taxonomy') ) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
275 $self->throw("Trying to classify with $taxonomy which is not a Bio::Taxonomy object");
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
276 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
277
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
278 # this is done to make sure we aren't duplicating a path (let God sort them out)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
279 if (defined $self->get_root_node) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
280 $self->get_root_node->remove_all_Descendents;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
281 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
282
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
283 my @nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
284
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
285 # nb taxa in [i][0] and ranks in [i][1]
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
286 my @taxa=$taxonomy->classify($species);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
287
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
288 for (my $i = 0; $i < @taxa; $i++) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
289 $nodes[$i]=Bio::Taxonomy::Taxon->new(-taxon => $taxa[$i][0],
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
290 -rank => $taxa[$i][1]);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
291 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
292
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
293 for (my $i = 0; $i < @taxa-1; $i++) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
294 $nodes[$i]->add_Descendent($nodes[$i+1]);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
295 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
296
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
297 $self->set_root_node($nodes[0]);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
298
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
299 return @nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
300 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
301
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
302
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
303 =head2 make_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
304
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
305 Title : make_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
306 Usage : $tree->make_branch($node)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
307 Function: Make a linear Bio::Taxonomy::Tree object from a leafish node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
308 Returns :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
309 Args : Bio::Taxonomy::Taxon object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
310
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
311 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
312
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
313
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
314 sub make_branch{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
315 my ($self,$node) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
316
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
317 # this is done to make sure we aren't duplicating a path (let God sort them out)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
318 # note that if you are using a linked set of node which include node
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
319 # already in the tree, this will break
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
320 $self->get_root_node->remove_all_Descendents;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
321
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
322 while (defined $node->ancestor) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
323 $self->set_root_node($node);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
324 $node=$node->ancestor;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
325 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
326 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
327
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
328
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
329 =head2 splice
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
330
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
331 Title : splice
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
332 Usage : @nodes = $tree->splice($tree)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
333 Function: Return a of Bio::Taxonomy::Tree object that is a fusion of two
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
334 Returns : array of Bio::Taxonomy::Taxon added to tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
335 Args : Bio::Taxonomy::Tree object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
336
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
337 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
338
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
339
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
340 sub splice{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
341 my ($self,$tree) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
342
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
343 my @nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
344
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
345 my @newleaves = $tree->get_leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
346 foreach my $leaf (@newleaves) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
347 push @nodes,$self->add_branch($leaf);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
348 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
349
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
350 return @nodes;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
351 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
352
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
353 =head2 add_species
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
354
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
355 Title : add_species
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
356 Usage : @nodes = $tree->add_species($species,$taxonomy)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
357 Function: Return a of Bio::Taxonomy::Tree object with a new species added
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
358 Returns : array of Bio::Taxonomy::Taxon added to tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
359 Args : Bio::Species object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
360
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
361 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
362
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
363
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
364 sub add_species{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
365 my ($self,$species,$taxonomy) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
366
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
367 my $branch=Bio::Taxonomy::Tree->new;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
368 my @nodes=$branch->make_species_branch($species,$taxonomy);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
369
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
370 my ($newleaf)=$branch->get_leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
371
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
372 return $self->add_branch($newleaf);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
373 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
374
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
375 =head2 add_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
376
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
377 Title : add_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
378 Usage : $tree->add_branch($node,boolean)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
379 Function: Return a of Bio::Taxonomy::Tree object with a new branch added
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
380 Returns : array of Bio::Taxonomy::Taxon objects of the resulting tree
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
381 Args : Bio::Taxonomy::Taxon object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
382 boolean flag to force overwrite of descendent
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
383 (see Bio::Node->add_Descendent)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
384
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
385 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
386
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
387
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
388 sub add_branch {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
389 my ($self,$node,$force) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
390
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
391 my $best_node_level=0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
392 my ($best_node,@nodes,$common);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
393
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
394 my @leaves=$self->get_leaves;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
395 foreach my $leaf (@leaves) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
396 $common=$node->recent_common_ancestor($leaf); # the root of the part to add
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
397 if (defined $common && ($common->distance_to_root > $best_node_level)) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
398 $best_node_level = $common->distance_to_root;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
399 $best_node = $common;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
400 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
401 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
402
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
403 return unless defined $best_node;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
404
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
405 push @nodes,($self->get_root_node,$self->get_root_node->get_Descendents);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
406 foreach my $node (@nodes) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
407 if ((defined $best_node->id && $best_node->id == $node->id) ||
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
408 ($best_node->rank eq $node->rank && $best_node->taxon eq $node->taxon) &&
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
409 ($best_node->rank ne 'no rank')) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
410 foreach my $descendent ($common->each_Descendent) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
411 $node->add_Descendent($descendent,$force);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
412 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
413 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
414
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
415 $self->set_root_node($node) if $node->distance_to_root==0;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
416 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
417
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
418 return ($common->get_Descendents);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
419 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
420
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
421 =head2 remove_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
422
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
423 Title : remove_branch
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
424 Usage : $tree->remove_branch($node)
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
425 Function: remove a branch up to the next multifurcation
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
426 Returns :
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
427 Args : Bio::Taxonomy::Taxon object
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
428
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
429 =cut
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
430
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
431
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
432 sub remove_branch{
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
433 my ($self,$node) = @_;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
434
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
435 # we can define a branch at any point along it
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
436
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
437 while (defined $node->ancestor) {
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
438 last if $node->ancestor->each_Descendent > 1;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
439 $node=$node->ancestor;
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
440 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
441 $node->remove_all_Descendents; # I'm not sure if this is necessary,
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
442 # but I don't see that remove_Descendent
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
443 # has the side effect of deleting
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
444 # descendent nodes of the deletee
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
445 $node->ancestor->remove_Descendent($node);
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
446 }
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
447
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
448
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
449
21066c0abaf5 Uploaded
willmclaren
parents:
diff changeset
450 1;