2 // FORESTER -- software libraries and applications
3 // for evolutionary biology research and applications.
5 // Copyright (C) 2008-2009 Christian M. Zmasek
6 // Copyright (C) 2008-2009 Burnham Institute for Medical Research
9 // This library is free software; you can redistribute it and/or
10 // modify it under the terms of the GNU Lesser General Public
11 // License as published by the Free Software Foundation; either
12 // version 2.1 of the License, or (at your option) any later version.
14 // This library is distributed in the hope that it will be useful,
15 // but WITHOUT ANY WARRANTY; without even the implied warranty of
16 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 // Lesser General Public License for more details.
19 // You should have received a copy of the GNU Lesser General Public
20 // License along with this library; if not, write to the Free Software
21 // Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA
23 // Contact: phylosoft @ gmail . com
24 // WWW: https://sites.google.com/site/cmzmasek/home/software/forester
26 package org.forester.evoinference.distance;
28 import java.math.RoundingMode;
29 import java.text.DecimalFormat;
30 import java.util.ArrayList;
31 import java.util.List;
33 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
34 import org.forester.phylogeny.Phylogeny;
35 import org.forester.phylogeny.PhylogenyNode;
36 import org.forester.util.ForesterUtil;
38 public final class NeighborJoining {
40 private BasicSymmetricalDistanceMatrix _d;
41 private double[][] _d_values;
42 private final DecimalFormat _df;
43 private PhylogenyNode[] _external_nodes;
44 private double[][] _m_values;
45 private int[] _mappings;
48 private final boolean _verbose;
50 private NeighborJoining() {
55 private NeighborJoining( final boolean verbose, final int maximum_fraction_digits_for_distances ) {
56 if ( ( maximum_fraction_digits_for_distances < 1 ) || ( maximum_fraction_digits_for_distances > 9 ) ) {
57 throw new IllegalArgumentException( "maximum fraction digits for distances is out of range: "
58 + maximum_fraction_digits_for_distances );
61 _df = new DecimalFormat();
62 _df.setMaximumFractionDigits( maximum_fraction_digits_for_distances );
63 _df.setRoundingMode( RoundingMode.HALF_UP );
66 public final Phylogeny execute( final BasicSymmetricalDistanceMatrix distance ) {
68 final Phylogeny phylogeny = new Phylogeny();
71 // Calculates the minimal distance.
72 // If more than one minimal distances, always the first found is used
73 // could randomize this, so that any would be returned in a randomized fashion...
74 double minimum = _m_values[ 0 ][ 1 ];
77 for( int j = 1; j < _n; ++j ) {
78 for( int i = 0; i < j; ++i ) {
79 if ( _m_values[ i ][ j ] < minimum ) {
80 minimum = _m_values[ i ][ j ];
86 // It is a condition that otu1 < otu2.
87 final PhylogenyNode node = new PhylogenyNode();
88 final double d = _d_values[ _mappings[ otu1 ] ][ _mappings[ otu2 ] ];
89 final double d1 = ( d / 2 ) + ( ( _r[ otu1 ] - _r[ otu2 ] ) / ( 2 * ( _n - 2 ) ) );
90 final double d2 = d - d1;
92 getExternalPhylogenyNode( otu1 ).setDistanceToParent( d1 );
93 getExternalPhylogenyNode( otu2 ).setDistanceToParent( d2 );
96 // yes, yes, slow but only grows with n (and not n^2 or worse)...
97 getExternalPhylogenyNode( otu1 ).setDistanceToParent( Double.parseDouble( _df.format( d1 ) ) );
98 getExternalPhylogenyNode( otu2 ).setDistanceToParent( Double.parseDouble( _df.format( d2 ) ) );
100 node.addAsChild( getExternalPhylogenyNode( otu1 ) );
101 node.addAsChild( getExternalPhylogenyNode( otu2 ) );
103 printProgress( otu1, otu2 );
105 calculateDistancesFromNewNode( otu1, otu2, d );
106 _external_nodes[ _mappings[ otu1 ] ] = node;
107 updateMappings( otu2 );
110 final double d = _d_values[ _mappings[ 0 ] ][ _mappings[ 1 ] ] / 2;
112 getExternalPhylogenyNode( 0 ).setDistanceToParent( d );
113 getExternalPhylogenyNode( 1 ).setDistanceToParent( d );
116 final double dd = Double.parseDouble( _df.format( d ) );
117 getExternalPhylogenyNode( 0 ).setDistanceToParent( dd );
118 getExternalPhylogenyNode( 1 ).setDistanceToParent( dd );
120 final PhylogenyNode root = new PhylogenyNode();
121 root.addAsChild( getExternalPhylogenyNode( 0 ) );
122 root.addAsChild( getExternalPhylogenyNode( 1 ) );
124 printProgress( 0, 1 );
126 phylogeny.setRoot( root );
127 phylogeny.setRooted( false );
131 public final List<Phylogeny> execute( final List<BasicSymmetricalDistanceMatrix> distances_list ) {
132 final List<Phylogeny> pl = new ArrayList<Phylogeny>();
133 for( final BasicSymmetricalDistanceMatrix distances : distances_list ) {
134 pl.add( execute( distances ) );
139 private final void calculateDistancesFromNewNode( final int otu1, final int otu2, final double d ) {
140 final int m_otu1 = _mappings[ otu1 ];
141 final int m_otu2 = _mappings[ otu2 ];
142 for( int i = 0; i < _n; ++i ) {
143 if ( ( i == otu1 ) || ( i == otu2 ) ) {
146 final int m_i = _mappings[ i ];
149 _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
152 _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
157 _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
160 _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
166 private final void calculateNetDivergences() {
168 for( int i = 0; i < _n; ++i ) {
170 final int m_i = _mappings[ i ];
171 for( int n = 0; n < _n; ++n ) {
174 d += _d_values[ _mappings[ n ] ][ m_i ];
177 d += _d_values[ m_i ][ _mappings[ n ] ];
185 private final PhylogenyNode getExternalPhylogenyNode( final int i ) {
186 return _external_nodes[ _mappings[ i ] ];
189 private final void initExternalNodes() {
190 _external_nodes = new PhylogenyNode[ _n ];
192 for( int i = 0; i < _n; ++i ) {
193 _external_nodes[ i ] = new PhylogenyNode();
194 id = _d.getIdentifier( i );
196 _external_nodes[ i ].setName( id );
199 _external_nodes[ i ].setName( Integer.toString( i ) );
205 private final void printD() {
206 System.out.println( "D:" );
207 for( final double[] _d_value : _d_values ) {
208 for( int j = 0; j < _d_values.length; j++ ) {
209 System.out.print( _d_value[ j ] );
210 System.out.print( " " );
212 System.out.println();
214 System.out.println();
217 private final void printM() {
218 System.out.println( "M:" );
219 for( final double[] _m_value : _m_values ) {
220 for( int j = 0; j < _m_values.length; j++ ) {
221 System.out.print( _m_value[ j ] );
222 System.out.print( " " );
224 System.out.println();
226 System.out.println();
229 private final void printProgress( final int otu1, final int otu2 ) {
230 final PhylogenyNode n1 = getExternalPhylogenyNode( otu1 );
231 final PhylogenyNode n2 = getExternalPhylogenyNode( otu2 );
232 System.out.println( "Node " + ( ForesterUtil.isEmpty( n1.getName() ) ? n1.getId() : n1.getName() ) + " joins "
233 + ( ForesterUtil.isEmpty( n2.getName() ) ? n2.getId() : n2.getName() ) );
236 // only the values in the lower triangle are used.
237 // !matrix values will be changed!
238 private final void reset( final BasicSymmetricalDistanceMatrix distances ) {
239 _n = distances.getSize();
241 _r = new double[ _n ];
242 _mappings = new int[ _n ];
243 _d_values = _d.getValues();
244 _m_values = new double[ _n ][ _n ];
248 private final void updateM() {
249 calculateNetDivergences();
250 final int n_minus_2 = _n - 2;
251 for( int j = 1; j < _n; ++j ) {
252 final double r_j = _r[ j ];
253 final int m_j = _mappings[ j ];
254 for( int i = 0; i < j; ++i ) {
255 _m_values[ i ][ j ] = _d_values[ _mappings[ i ] ][ m_j ] - ( ( _r[ i ] + r_j ) / n_minus_2 );
260 // otu2 will, in effect, be "deleted" from the matrix.
261 private final void updateMappings( final int otu2 ) {
262 for( int i = otu2; i < ( _mappings.length - 1 ); ++i ) {
263 _mappings[ i ] = _mappings[ i + 1 ];
267 public final static NeighborJoining createInstance() {
268 return new NeighborJoining();
271 public final static NeighborJoining createInstance( final boolean verbose,
272 final int maximum_fraction_digits_for_distances ) {
273 return new NeighborJoining( verbose, maximum_fraction_digits_for_distances );