2 // FORESTER -- software libraries and applications
3 // for evolutionary biology research and applications.
5 // Copyright (C) 2014 Christian M. Zmasek
8 // This library is free software; you can redistribute it and/or
9 // modify it under the terms of the GNU Lesser General Public
10 // License as published by the Free Software Foundation; either
11 // version 2.1 of the License, or (at your option) any later version.
13 // This library is distributed in the hope that it will be useful,
14 // but WITHOUT ANY WARRANTY; without even the implied warranty of
15 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 // Lesser General Public License for more details.
18 // You should have received a copy of the GNU Lesser General Public
19 // License along with this library; if not, write to the Free Software
20 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
22 // Contact: phylosoft @ gmail . com
23 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
25 package org.forester.evoinference.distance;
27 import java.math.RoundingMode;
28 import java.text.DecimalFormat;
29 import java.util.ArrayList;
30 import java.util.List;
32 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
33 import org.forester.phylogeny.Phylogeny;
34 import org.forester.phylogeny.PhylogenyNode;
35 import org.forester.util.ForesterUtil;
37 public final class NeighborJoining {
39 private BasicSymmetricalDistanceMatrix _d;
40 private double[][] _d_values;
41 private final DecimalFormat _df;
42 private PhylogenyNode[] _external_nodes;
43 private double[][] _m_values;
44 private int[] _mappings;
47 private final boolean _verbose;
49 private NeighborJoining() {
54 private NeighborJoining( final boolean verbose, final int maximum_fraction_digits_for_distances ) {
55 if ( ( maximum_fraction_digits_for_distances < 1 ) || ( maximum_fraction_digits_for_distances > 9 ) ) {
56 throw new IllegalArgumentException( "maximum fraction digits for distances is out of range: "
57 + maximum_fraction_digits_for_distances );
60 _df = new DecimalFormat();
61 _df.setMaximumFractionDigits( maximum_fraction_digits_for_distances );
62 _df.setRoundingMode( RoundingMode.HALF_UP );
65 public final Phylogeny execute( final BasicSymmetricalDistanceMatrix distance ) {
67 final Phylogeny phylogeny = new Phylogeny();
70 // Calculates the minimal distance.
71 // If more than one minimal distances, always the first found is used
72 // could randomize this, so that any would be returned in a randomized fashion...
73 double minimum = _m_values[ 0 ][ 1 ];
76 for( int j = 1; j < _n; ++j ) {
77 for( int i = 0; i < j; ++i ) {
78 if ( _m_values[ i ][ j ] < minimum ) {
79 minimum = _m_values[ i ][ j ];
85 // It is a condition that otu1 < otu2.
86 final PhylogenyNode node = new PhylogenyNode();
87 final double d = _d_values[ _mappings[ otu1 ] ][ _mappings[ otu2 ] ];
88 final double d1 = ( d / 2 ) + ( ( _r[ otu1 ] - _r[ otu2 ] ) / ( 2 * ( _n - 2 ) ) );
89 final double d2 = d - d1;
91 getExternalPhylogenyNode( otu1 ).setDistanceToParent( d1 );
92 getExternalPhylogenyNode( otu2 ).setDistanceToParent( d2 );
95 // yes, yes, slow but only grows with n (and not n^2 or worse)...
96 getExternalPhylogenyNode( otu1 ).setDistanceToParent( Double.parseDouble( _df.format( d1 ) ) );
97 getExternalPhylogenyNode( otu2 ).setDistanceToParent( Double.parseDouble( _df.format( d2 ) ) );
99 node.addAsChild( getExternalPhylogenyNode( otu1 ) );
100 node.addAsChild( getExternalPhylogenyNode( otu2 ) );
102 printProgress( otu1, otu2 );
104 calculateDistancesFromNewNode( otu1, otu2, d );
105 _external_nodes[ _mappings[ otu1 ] ] = node;
106 updateMappings( otu2 );
109 final double d = _d_values[ _mappings[ 0 ] ][ _mappings[ 1 ] ] / 2;
111 getExternalPhylogenyNode( 0 ).setDistanceToParent( d );
112 getExternalPhylogenyNode( 1 ).setDistanceToParent( d );
115 final double dd = Double.parseDouble( _df.format( d ) );
116 getExternalPhylogenyNode( 0 ).setDistanceToParent( dd );
117 getExternalPhylogenyNode( 1 ).setDistanceToParent( dd );
119 final PhylogenyNode root = new PhylogenyNode();
120 root.addAsChild( getExternalPhylogenyNode( 0 ) );
121 root.addAsChild( getExternalPhylogenyNode( 1 ) );
123 printProgress( 0, 1 );
125 phylogeny.setRoot( root );
126 phylogeny.setRooted( false );
130 public final List<Phylogeny> execute( final List<BasicSymmetricalDistanceMatrix> distances_list ) {
131 final List<Phylogeny> pl = new ArrayList<Phylogeny>();
132 for( final BasicSymmetricalDistanceMatrix distances : distances_list ) {
133 pl.add( execute( distances ) );
138 private final void calculateDistancesFromNewNode( final int otu1, final int otu2, final double d ) {
139 final int m_otu1 = _mappings[ otu1 ];
140 final int m_otu2 = _mappings[ otu2 ];
141 for( int i = 0; i < _n; ++i ) {
142 if ( ( i == otu1 ) || ( i == otu2 ) ) {
145 final int m_i = _mappings[ i ];
148 _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
151 _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
156 _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
159 _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
165 private final void calculateNetDivergences() {
167 for( int i = 0; i < _n; ++i ) {
169 final int m_i = _mappings[ i ];
170 for( int n = 0; n < _n; ++n ) {
173 d += _d_values[ _mappings[ n ] ][ m_i ];
176 d += _d_values[ m_i ][ _mappings[ n ] ];
184 private final PhylogenyNode getExternalPhylogenyNode( final int i ) {
185 return _external_nodes[ _mappings[ i ] ];
188 private final void initExternalNodes() {
189 _external_nodes = new PhylogenyNode[ _n ];
191 for( int i = 0; i < _n; ++i ) {
192 _external_nodes[ i ] = new PhylogenyNode();
193 id = _d.getIdentifier( i );
195 _external_nodes[ i ].setName( id );
198 _external_nodes[ i ].setName( Integer.toString( i ) );
204 private final void printProgress( final int otu1, final int otu2 ) {
205 final PhylogenyNode n1 = getExternalPhylogenyNode( otu1 );
206 final PhylogenyNode n2 = getExternalPhylogenyNode( otu2 );
207 System.out.println( "Node " + ( ForesterUtil.isEmpty( n1.getName() ) ? n1.getId() : n1.getName() ) + " joins "
208 + ( ForesterUtil.isEmpty( n2.getName() ) ? n2.getId() : n2.getName() ) );
211 // only the values in the lower triangle are used.
212 // !matrix values will be changed!
213 private final void reset( final BasicSymmetricalDistanceMatrix distances ) {
214 _n = distances.getSize();
216 _r = new double[ _n ];
217 _mappings = new int[ _n ];
218 _d_values = _d.getValues();
219 _m_values = new double[ _n ][ _n ];
223 private final void updateM() {
224 calculateNetDivergences();
225 final int n_minus_2 = _n - 2;
226 for( int j = 1; j < _n; ++j ) {
227 final double r_j = _r[ j ];
228 final int m_j = _mappings[ j ];
229 for( int i = 0; i < j; ++i ) {
230 _m_values[ i ][ j ] = _d_values[ _mappings[ i ] ][ m_j ] - ( ( _r[ i ] + r_j ) / n_minus_2 );
235 // otu2 will, in effect, be "deleted" from the matrix.
236 private final void updateMappings( final int otu2 ) {
237 for( int i = otu2; i < ( _mappings.length - 1 ); ++i ) {
238 _mappings[ i ] = _mappings[ i + 1 ];
242 public final static NeighborJoining createInstance() {
243 return new NeighborJoining();
246 public final static NeighborJoining createInstance( final boolean verbose,
247 final int maximum_fraction_digits_for_distances ) {
248 return new NeighborJoining( verbose, maximum_fraction_digits_for_distances );