annotate variant_effect_predictor/Bio/Tree/TreeFunctionsI.pm @ 3:d30fa12e4cc5 default tip

Merge heads 2:a5976b2dce6f and 1:09613ce8151e which were created as a result of a recently fixed bug.
author devteam <devteam@galaxyproject.org>
date Mon, 13 Jan 2014 10:38:30 -0500
parents 1f6dce3d34e0
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
1 # $Id: TreeFunctionsI.pm,v 1.5.2.3 2003/09/14 20:18:25 jason Exp $
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
2 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
3 # BioPerl module for Bio::Tree::TreeFunctionsI
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
4 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
5 # Cared for by Jason Stajich <jason@bioperl.org>
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
6 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
7 # Copyright Jason Stajich
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
8 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
9 # You may distribute this module under the same terms as perl itself
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
10
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
11 # POD documentation - main docs before the code
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
12
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
13 =head1 NAME
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
14
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
15 Bio::Tree::TreeFunctionsI - Decorated Interface implementing basic Tree exploration methods
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
16
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
17 =head1 SYNOPSIS
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
18
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
19 use Bio::TreeIO;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
20 my $in = new Bio::TreeIO(-format => 'newick', -file => 'tree.tre');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
21
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
22 my $tree = $in->next_tree;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
23
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
24 my @nodes = $tree->find_node('id1');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
25
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
26 if( $tree->is_monophyletic(-clade => @nodes, -outgroup => $outnode) ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
27
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
28 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
29
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
30 =head1 DESCRIPTION
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
31
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
32 Describe the interface here
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
33
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
34 =head1 FEEDBACK
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
35
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
36 =head2 Mailing Lists
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
37
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
38 User feedback is an integral part of the evolution of this and other
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
39 Bioperl modules. Send your comments and suggestions preferably to
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
40 the Bioperl mailing list. Your participation is much appreciated.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
41
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
42 bioperl-l@bioperl.org - General discussion
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
43 http://bioperl.org/MailList.shtml - About the mailing lists
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
44
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
45 =head2 Reporting Bugs
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
46
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
47 Report bugs to the Bioperl bug tracking system to help us keep track
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
48 of the bugs and their resolution. Bug reports can be submitted via
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
49 email or the web:
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
50
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
51 bioperl-bugs@bioperl.org
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
52 http://bugzilla.bioperl.org/
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
53
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
54 =head1 AUTHOR - Jason Stajich, Aaron Mackey, Justin Reese
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
55
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
56 Email jason-at-bioperl-dot-org
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
57 Email amackey-at-virginia.edu
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
58 Email jtr4v-at-virginia.edu
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
59
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
60 =head1 CONTRIBUTORS
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
61
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
62 Additional contributors names and emails here
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
63
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
64 Rerooting code was worked on by
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
65
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
66 Daniel Barker d.barker-at-reading.ac.uk
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
67 Ramiro Barrantes Ramiro.Barrantes-at-uvm.edu
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
68
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
69 =head1 APPENDIX
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
70
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
71 The rest of the documentation details each of the object methods.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
72 Internal methods are usually preceded with a _
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
73
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
74 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
75
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
76
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
77 # Let the code begin...
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
78
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
79
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
80 package Bio::Tree::TreeFunctionsI;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
81 use vars qw(@ISA);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
82 use strict;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
83 use Bio::Tree::TreeI;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
84
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
85 @ISA = qw(Bio::Tree::TreeI);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
86
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
87 =head2 find_node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
88
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
89 Title : find_node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
90 Usage : my @nodes = $self->find_node(-id => 'node1');
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
91 Function: returns all nodes that match a specific field, by default this
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
92 is id, but different branch_length,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
93 Returns : List of nodes which matched search
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
94 Args : text string to search for
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
95 OR
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
96 -fieldname => $textstring
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
97
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
98 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
99
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
100 sub find_node {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
101 my ($self,$type,$field) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
102 if( ! defined $type ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
103 $self->warn("Must request a either a string or field and string when searching");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
104 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
105
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
106 # all this work for a '-' named field
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
107 # is so that we could potentially
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
108 # expand to other constraints in
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
109 # different implementations
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
110 # like 'find all nodes with boostrap < XX'
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
111
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
112 if( ! defined $field ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
113 # only 1 argument, default to searching by id
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
114 $field= $type;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
115 $type = 'id';
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
116 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
117 $type =~ s/^-//;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
118 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
119
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
120 # could actually do this by testing $rootnode->can($type) but
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
121 # it is possible that a tree is implemeted with different node types
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
122 # - although it is unlikely that the root node would be richer than the
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
123 # leaf nodes. Can't handle NHX tags right now
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
124
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
125 unless( $type eq 'id' || $type eq 'name' ||
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
126 $type eq 'bootstrap' || $type eq 'description' ||
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
127 $type eq 'internal_id') {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
128 $self->warn("unknown search type $type - will try anyways");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
129 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
130 my @nodes = grep { $_->can($type) && defined $_->$type() &&
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
131 $_->$type() eq $field } $self->get_nodes();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
132
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
133 if ( wantarray) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
134 return @nodes;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
135 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
136 if( @nodes > 1 ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
137 $self->warn("More than 1 node found but caller requested scalar, only returning first node");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
138 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
139 return shift @nodes;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
140 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
141 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
142
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
143 =head2 remove_Node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
144
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
145 Title : remove_Node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
146 Usage : $tree->remove_Node($node)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
147 Function: Removes a node from the tree
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
148 Returns : boolean represent status of success
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
149 Args : either Bio::Tree::NodeI or string of the node id
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
150
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
151
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
152 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
153
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
154 sub remove_Node {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
155 my ($self,$input) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
156 my $node = undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
157 unless( ref($input) ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
158 $node = $self->find_node($input);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
159 } elsif( ! $input->isa('Bio::Tree::NodeI') ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
160 $self->warn("Did not provide either a valid Bio::Tree::NodeI object to remove_node or the node name");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
161 return 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
162 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
163 $node = $input;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
164 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
165 if( ! $node->ancestor && $self->get_root_node->internal_id != $node->internal_id) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
166 $self->warn("Node (".$node->to_string . ") has no ancestor, can't remove!");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
167 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
168 $node->ancestor->remove_Descendent($node);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
169 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
170 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
171
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
172
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
173 # Added for Justin Reese by Jason
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
174
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
175 =head2 get_lca
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
176
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
177 Title : get_lca
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
178 Usage : get_lca(-nodes => \@nodes )
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
179 Function: given two nodes, returns the lowest common ancestor
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
180 Returns : node object
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
181 Args : -nodes => arrayref of nodes to test
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
182
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
183
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
184 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
185
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
186 sub get_lca {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
187 my ($self,@args) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
188 my ($nodes) = $self->_rearrange([qw(NODES)],@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
189 if( ! defined $nodes ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
190 $self->warn("Must supply -nodes parameter to get_lca() method");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
191 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
192 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
193 my ($node1,$node2) = $self->_check_two_nodes($nodes);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
194 return undef unless $node1 && $node2;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
195
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
196 # algorithm: Start with first node, find and save every node from it to
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
197 # root. Then start with second node; for it and each of its ancestor
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
198 # nodes, check to see if it's in the first node's ancestor list - if
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
199 # so it is the lca.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
200 #
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
201 # This is very slow and naive, but I somehow doubt the overhead
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
202 # of mapping the tree to a complete binary tree and doing the linear
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
203 # lca search would be worth the overhead, especially for small trees.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
204 # Maybe someday I'll write a linear get_lca and find out.
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
205
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
206 # find and save every ancestor of node1 (including itself)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
207
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
208 my %node1_ancestors; # keys are internal ids, values are objects
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
209 my $place = $node1; # start at node1
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
210
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
211 while ( $place ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
212 $node1_ancestors{$place->internal_id} = $place;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
213 $place = $place->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
214 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
215
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
216 # now climb up node2, for each node checking whether
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
217 # it's in node1_ancestors
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
218 $place = $node2; # start at node2
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
219 while ( $place ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
220 foreach my $key ( keys %node1_ancestors ){ # ugh
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
221 if ( $place->internal_id == $key){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
222 return $node1_ancestors{$key};
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
223 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
224 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
225 $place = $place->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
226 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
227 $self->warn("Could not find lca!"); # should never execute,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
228 # if so, there's a problem
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
229 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
230 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
231
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
232 # Added for Justin Reese by Jason
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
233
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
234 =head2 distance
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
235
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
236 Title : distance
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
237 Usage : distance(-nodes => \@nodes )
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
238 Function: returns the distance between two given nodes
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
239 Returns : numerical distance
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
240 Args : -nodes => arrayref of nodes to test
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
241
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
242
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
243 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
244
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
245 sub distance {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
246 my ($self,@args) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
247 my ($nodes) = $self->_rearrange([qw(NODES)],@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
248 if( ! defined $nodes ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
249 $self->warn("Must supply -nodes parameter to distance() method");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
250 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
251 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
252 my ($node1,$node2) = $self->_check_two_nodes($nodes);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
253 # algorithm:
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
254
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
255 # Find lca: Start with first node, find and save every node from it
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
256 # to root, saving cumulative distance. Then start with second node;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
257 # for it and each of its ancestor nodes, check to see if it's in
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
258 # the first node's ancestor list - if so it is the lca. Return sum
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
259 # of (cumul. distance from node1 to lca) and (cumul. distance from
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
260 # node2 to lca)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
261
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
262 # find and save every ancestor of node1 (including itself)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
263
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
264 my %node1_ancestors; # keys are internal ids, values are objects
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
265 my %node1_cumul_dist; # keys are internal ids, values
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
266 # are cumulative distance from node1 to given node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
267 my $place = $node1; # start at node1
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
268 my $cumul_dist = 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
269
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
270 while ( $place ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
271 $node1_ancestors{$place->internal_id} = $place;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
272 $node1_cumul_dist{$place->internal_id} = $cumul_dist;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
273 if ($place->branch_length) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
274 $cumul_dist += $place->branch_length; # include current branch
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
275 # length in next iteration
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
276 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
277 $place = $place->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
278 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
279
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
280 # now climb up node2, for each node checking whether
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
281 # it's in node1_ancestors
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
282 $place = $node2; # start at node2
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
283 $cumul_dist = 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
284 while ( $place ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
285 foreach my $key ( keys %node1_ancestors ){ # ugh
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
286 if ( $place->internal_id == $key){ # we're at lca
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
287 return $node1_cumul_dist{$key} + $cumul_dist;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
288 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
289 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
290 # include current branch length in next iteration
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
291 $cumul_dist += $place->branch_length;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
292 $place = $place->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
293 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
294 $self->warn("Could not find distance!"); # should never execute,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
295 # if so, there's a problem
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
296 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
297 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
298
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
299 # helper function to check lca and distance arguments
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
300
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
301 sub _check_two_nodes {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
302 my ($self, $nodes) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
303
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
304 if( ref($nodes) !~ /ARRAY/i ||
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
305 !ref($nodes->[0]) ||
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
306 !ref($nodes->[1])
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
307 ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
308 $self->warn("Must provide a valid array reference for -nodes");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
309 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
310 } elsif( scalar(@$nodes) > 2 ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
311 $self->warn("More than two nodes given, using first two");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
312 } elsif( scalar(@$nodes) < 2 ){
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
313 $self->warn("-nodes parameter does not contain reference to two nodes");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
314 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
315 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
316 unless( $nodes->[0]->isa('Bio::Tree::NodeI') &&
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
317 $nodes->[1]->isa('Bio::Tree::NodeI') ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
318 $self->warn("Did not provide valid Bio::Tree::NodeI objects as nodes\n");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
319 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
320 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
321 return @$nodes;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
322 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
323
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
324
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
325 =head2 is_monophyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
326
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
327 Title : is_monophyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
328 Usage : if( $tree->is_monophyletic(-nodes => \@nodes,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
329 -outgroup => $outgroup)
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
330 Function: Will do a test of monophyly for the nodes specified
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
331 in comparison to a chosen outgroup
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
332 Returns : boolean
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
333 Args : -nodes => arrayref of nodes to test
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
334 -outgroup => outgroup to serve as a reference
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
335
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
336
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
337 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
338
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
339 sub is_monophyletic{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
340 my ($self,@args) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
341 my ($nodes,$outgroup) = $self->_rearrange([qw(NODES OUTGROUP)],@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
342
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
343 if( ! defined $nodes || ! defined $outgroup ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
344 $self->warn("Must supply -nodes and -outgroup parameters to the method
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
345 is_monophyletic");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
346 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
347 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
348 if( ref($nodes) !~ /ARRAY/i ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
349 $self->warn("Must provide a valid array reference for -nodes");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
350 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
351 my $clade_root;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
352 # this is to combine multiple tests into a single node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
353 # order doesn't really matter as long as get_lca does its job right
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
354 while( @$nodes > 2 ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
355 my ($a,$b) = ( shift @$nodes, shift @$nodes);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
356 $clade_root = $self->get_lca(-nodes => [$a,$b] );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
357 unshift @$nodes, $clade_root;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
358 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
359 $clade_root = $self->get_lca(-nodes => $nodes );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
360 my $og_ancestor = $outgroup->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
361 while( defined ($og_ancestor ) ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
362 if( $og_ancestor->internal_id == $clade_root->internal_id ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
363 # monophyly is violated
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
364 return 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
365 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
366 $og_ancestor = $og_ancestor->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
367 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
368 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
369 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
370
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
371 =head2 is_paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
372
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
373 Title : is_paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
374 Usage : if( $tree->is_paraphyletic(-nodes =>\@nodes,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
375 -outgroup => $node) ){ }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
376 Function: Tests whether or not a given set of nodes are paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
377 (representing the full clade) given an outgroup
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
378 Returns : [-1,0,1] , -1 if the group is not monophyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
379 0 if the group is not paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
380 1 if the group is paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
381 Args : -nodes => Array of Bio::Tree::NodeI objects which are in the tree
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
382 -outgroup => a Bio::Tree::NodeI to compare the nodes to
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
383
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
384
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
385 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
386
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
387 sub is_paraphyletic{
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
388 my ($self,@args) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
389 my ($nodes,$outgroup) = $self->_rearrange([qw(NODES OUTGROUP)],@args);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
390
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
391 if( ! defined $nodes || ! defined $outgroup ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
392 $self->warn("Must suply -nodes and -outgroup parameters to the method is_paraphyletic");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
393 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
394 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
395 if( ref($nodes) !~ /ARRAY/i ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
396 $self->warn("Must provide a valid array reference for -nodes");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
397 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
398 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
399
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
400 # Algorithm
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
401 # Find the lca
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
402 # Find all the nodes beneath the lca
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
403 # Test to see that none are missing from the nodes list
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
404 my %nodehash;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
405 foreach my $n ( @$nodes ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
406 $nodehash{$n->internal_id} = $n;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
407 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
408 while( @$nodes > 2 ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
409 unshift @$nodes, $self->get_lca(-nodes => [( shift @$nodes,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
410 shift @$nodes)] );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
411 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
412 my $clade_root = $self->get_lca(-nodes => $nodes );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
413 unless( defined $clade_root ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
414 $self->warn("could not find clade root via lca");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
415 return undef;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
416 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
417 my $og_ancestor = $outgroup->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
418
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
419 # Is this necessary/correct for paraphyly test?
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
420 while( defined ($og_ancestor ) ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
421 if( $og_ancestor->internal_id == $clade_root->internal_id ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
422 # monophyly is violated, could be paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
423 return -1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
424 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
425 $og_ancestor = $og_ancestor->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
426 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
427 my $tree = new Bio::Tree::Tree(-root => $clade_root,
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
428 -nodelete => 1);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
429
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
430 foreach my $n ( $tree->get_nodes() ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
431 next unless $n->is_Leaf();
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
432 # if any leaf node is not in the list
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
433 # then it is part of the clade and so the list
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
434 # must be paraphyletic
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
435 return 1 unless ( $nodehash{$n->internal_id} );
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
436 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
437 return 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
438 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
439
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
440
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
441 =head2 reroot
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
442
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
443 Title : reroot_tree
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
444 Usage : $tree->reroot($node);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
445 Function: Reroots a tree either making a new node the root
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
446 Returns : 1 on success, 0 on failure
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
447 Args : Bio::Tree::NodeI that is in the tree, but is not the current root
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
448
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
449 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
450
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
451 sub reroot {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
452 my ($self,$new_root) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
453 unless (defined $new_root && $new_root->isa("Bio::Tree::NodeI")) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
454 $self->warn("Must provide a valid Bio::Tree::NodeI when rerooting");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
455 return 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
456 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
457 if( $new_root->is_Leaf() ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
458 $self->warn("Asking to root with a leaf, will use the leaf's ancestor");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
459 $new_root = $new_root->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
460 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
461
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
462 my $old_root = $self->get_root_node;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
463 if( $new_root == $old_root ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
464 $self->warn("Node requested for reroot is already the root node!");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
465 return 0;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
466 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
467
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
468 my @path = (); # along tree, from newroot to oldroot
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
469 my $node = $new_root;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
470 while ($node) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
471 push @path, $node;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
472 $node = $node->ancestor;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
473 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
474
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
475 my @path_from_oldroot = reverse @path;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
476 for (my $i = 0; $i < @path_from_oldroot - 1; $i++) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
477 my $current = $path_from_oldroot[$i];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
478 my $next = $path_from_oldroot[$i + 1];
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
479 $current->remove_Descendent($next);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
480 $current->branch_length($next->branch_length);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
481 $next->add_Descendent($current);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
482
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
483 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
484 $new_root->branch_length(undef);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
485 $self->set_root_node($new_root);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
486
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
487 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
488 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
489
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
490 =head2 reverse_edge
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
491
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
492 Title : reverse_edge
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
493 Usage : $node->reverse_edge(child);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
494 Function: makes child be a parent of node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
495 Requires: child must be a direct descendent of node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
496 Returns : nothing
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
497 Args : Bio::Tree::NodeI that is in the tree
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
498
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
499 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
500
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
501 sub reverse_edge {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
502 my ($self,$node) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
503 delete_edge($self, $node);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
504 $node->add_Descendent($self);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
505 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
506 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
507
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
508 =head2 delete_edge
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
509
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
510 Title : delete_edge
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
511 Usage : $node->reverse_edge(child);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
512 Function: makes child be a parent of node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
513 Requires: child must be a direct descendent of node
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
514 Returns : nothing
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
515 Args : Bio::Tree::NodeI that is in the tree
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
516
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
517 =cut
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
518
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
519 sub delete_edge {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
520 my ($self,$node) = @_;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
521 unless (defined $self && $self->isa("Bio::Tree::NodeI")) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
522 $self->warn("Must provide a valid Bio::Tree::NodeI when rerooting");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
523 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
524 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
525 unless (defined $node && $node->isa("Bio::Tree::NodeI")) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
526 $self->warn("Must provide a valid Bio::Tree::NodeI when rerooting");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
527 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
528 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
529 if( $self->{'_desc'}->{$node->internal_id} ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
530 $node->ancestor(undef);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
531 $self->{'_desc'}->{$node->internal_id}->ancestor(undef);
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
532 delete $self->{'_desc'}->{$node->internal_id};
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
533 } else {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
534 $self->warn("First argument must be direct parent of node");
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
535 return 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
536 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
537 1;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
538 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
539
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
540 sub findnode_by_id {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
541 my $tree = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
542 my $id = shift;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
543 my $rootnode = $tree->get_root_node;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
544 if ( ($rootnode->id) and ($rootnode->id eq $id) ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
545 return $rootnode;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
546 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
547 # process all the children
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
548 foreach my $node ( $rootnode->get_Descendents ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
549 if ( ($node->id) and ($node->id eq $id ) ) {
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
550 return $node;
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
551 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
552 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
553 }
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
554
1f6dce3d34e0 Uploaded
mahtabm
parents:
diff changeset
555 1;