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;
31 import java.util.Map.Entry;
32 import java.util.SortedSet;
34 import org.forester.evoinference.matrix.distance.BasicSymmetricalDistanceMatrix;
35 import org.forester.phylogeny.Phylogeny;
36 import org.forester.phylogeny.PhylogenyNode;
37 import org.forester.util.ForesterUtil;
39 public final class NeighborJoiningR {
41 private final static DecimalFormat DF = new DecimalFormat( "0.00000" );
42 private BasicSymmetricalDistanceMatrix _d;
43 private double[][] _d_values;
44 private final DecimalFormat _df;
45 private PhylogenyNode[] _external_nodes;
46 private int[] _mappings;
49 private final boolean _verbose;
53 private double _d_min; //TODO remove me
54 private int[] _rev_mappings;
56 private NeighborJoiningR() {
61 private NeighborJoiningR( final boolean verbose, final int maximum_fraction_digits_for_distances ) {
62 if ( ( maximum_fraction_digits_for_distances < 1 ) || ( maximum_fraction_digits_for_distances > 9 ) ) {
63 throw new IllegalArgumentException( "maximum fraction digits for distances is out of range: "
64 + maximum_fraction_digits_for_distances );
67 _df = new DecimalFormat();
68 _df.setMaximumFractionDigits( maximum_fraction_digits_for_distances );
69 _df.setRoundingMode( RoundingMode.HALF_UP );
72 public final Phylogeny execute( final BasicSymmetricalDistanceMatrix distance ) {
74 final Phylogeny phylogeny = new Phylogeny();
76 System.out.println( "N=" + _n );
78 // Calculates the minimal distance.
79 // If more than one minimal distances, always the first found is used
80 final double m = updateM();
81 final int otu1 = _min_i;
82 final int otu2 = _min_j;
83 System.out.println( _min_i + " " + _min_j + " => " + DF.format( m ) + " (" + DF.format( _d_min ) + ")" );
84 // It is a condition that otu1 < otu2.
85 //System.out.println( "mapped 1 " + _mappings[ otu1 ] );
86 System.out.println( "mapped otu 2 " + _mappings[ otu2 ] );
87 final PhylogenyNode node = new PhylogenyNode();
88 //final double d = getDvalueUnmapped( otu1, _mappings[ otu2 ] );
89 final double d = _d_values[ otu1 ][ _mappings[ otu2 ] ];
90 final double d1 = ( d / 2 ) + ( ( _r[ _rev_mappings[ otu1 ] ] - _r[ otu2 ] ) / ( 2 * ( _n - 2 ) ) );
91 final double d2 = d - d1;
93 _external_nodes[ otu1 ].setDistanceToParent( d1 );
94 getExternalPhylogenyNode( otu2 ).setDistanceToParent( d2 );
97 // yes, yes, slow but only grows with n (and not n^2 or worse)...
98 _external_nodes[ otu1 ].setDistanceToParent( Double.parseDouble( _df.format( d1 ) ) );
99 getExternalPhylogenyNode( otu2 ).setDistanceToParent( Double.parseDouble( _df.format( d2 ) ) );
101 node.addAsChild( _external_nodes[ otu1 ] );
102 node.addAsChild( getExternalPhylogenyNode( otu2 ) );
104 printProgress( otu1, otu2, node );
106 System.out.println( "otu1=" + otu1 );
107 System.out.println( "otu2=" + otu2 );
108 calculateDistancesFromNewNode( otu1, otu2, d );
109 // _external_nodes[ _mappings[ otu1 ] ] = node;
110 _external_nodes[ otu1 ] = node;
111 updateMappings( otu2 );
113 System.out.println( "" );
114 System.out.println( "----------------------------------------------------------------------------------" );
115 System.out.println( "" );
117 final double d = getDvalue( 0, 1 ) / 2;
119 getExternalPhylogenyNode( 0 ).setDistanceToParent( d );
120 getExternalPhylogenyNode( 1 ).setDistanceToParent( d );
123 final double dd = Double.parseDouble( _df.format( d ) );
124 getExternalPhylogenyNode( 0 ).setDistanceToParent( dd );
125 getExternalPhylogenyNode( 1 ).setDistanceToParent( dd );
127 final PhylogenyNode root = new PhylogenyNode();
128 root.addAsChild( getExternalPhylogenyNode( 0 ) );
129 root.addAsChild( getExternalPhylogenyNode( 1 ) );
131 printProgress( 0, 1, root );
133 phylogeny.setRoot( root );
134 phylogeny.setRooted( false );
138 public final List<Phylogeny> execute( final List<BasicSymmetricalDistanceMatrix> distances_list ) {
139 final List<Phylogeny> pl = new ArrayList<Phylogeny>();
140 for( final BasicSymmetricalDistanceMatrix distances : distances_list ) {
141 pl.add( execute( distances ) );
146 private final void calculateDistancesFromNewNode( final int otu1, final int otu2, final double d ) {
147 System.out.print( "new D values: " );
148 for( int j = 0; j < _n; ++j ) {
152 if ( otu1 < _mappings[ j ] ) {
153 updateDvalue( otu1, otu2, j, d );
155 // else if ( otu1 > _mappings[ j ] ) {
156 // updateDvalue( otu2, otu1, j, d );
159 System.out.println();
162 private final void updateDvalue( final int otu1, final int otu2, final int j, final double d ) {
163 final double new_d = ( getDvalueUnmapped( otu1, _mappings[ j ] ) + getDvalue( j, otu2 ) - d ) / 2;
164 System.out.print( DF.format( new_d ) + " " );
165 // System.out.println( "going to remove: " + getDvalueUnmapped( otu1, _mappings[ j ] ) + ", " + otu1 + ", "
166 // + _mappings[ j ] );
167 if ( otu1 < _mappings[ j ] ) {
168 _s.removePairing( getDvalueUnmapped( otu1, _mappings[ j ] ), otu1, _mappings[ j ] );
171 _s.removePairing( getDvalueUnmapped( otu1, _mappings[ j ] ), _mappings[ j ], otu1 );
173 // System.out.println( "going to remove: " + getDvalue( j, otu2 ) + ", " +_mappings[ otu2 ] + ", "
174 // + _mappings[ j ] );
175 if ( _mappings[ otu2 ] < _mappings[ j ] ) {
176 _s.removePairing( getDvalue( j, otu2 ), _mappings[ otu2 ], _mappings[ j ] );
179 _s.removePairing( getDvalue( j, otu2 ), _mappings[ j ], _mappings[ otu2 ] );
181 _s.addPairing( new_d, otu1, _mappings[ j ] );
182 setDvalueU( otu1, j, new_d );
185 private void setDvalueU( final int i, final int j, final double d ) {
186 if ( i < _mappings[ j ] ) {
187 _d_values[ i ][ _mappings[ j ] ] = d;
189 _d_values[ _mappings[ j ] ][ i ] = d;
192 private double getDvalue( final int i, final int j ) {
194 return _d_values[ _mappings[ i ] ][ _mappings[ j ] ];
196 return _d_values[ _mappings[ j ] ][ _mappings[ i ] ];
199 private double getDvalueUnmapped( final int i, final int j ) {
201 return _d_values[ i ][ j ];
203 return _d_values[ j ][ i ];
206 private final void calculateNetDivergences() {
207 for( int i = 0; i < _n; ++i ) {
208 _r[ i ] = calculateNetDivergence( i );
212 private double calculateNetDivergence( final int i ) {
214 for( int n = 0; n < _n; ++n ) {
216 d += getDvalue( n, i );
222 private final PhylogenyNode getExternalPhylogenyNode( final int i ) {
223 return _external_nodes[ _mappings[ i ] ];
226 private final void initExternalNodes() {
227 _external_nodes = new PhylogenyNode[ _n ];
229 for( int i = 0; i < _n; ++i ) {
230 _external_nodes[ i ] = new PhylogenyNode();
231 id = _d.getIdentifier( i );
233 _external_nodes[ i ].setName( id );
236 _external_nodes[ i ].setName( Integer.toString( i ) );
239 _rev_mappings[ i ] = i;
243 private final void printProgress( final int otu1, final int otu2, final PhylogenyNode node ) {
244 System.out.println( "Node " + printProgressNodeToString( _external_nodes[ otu1 ] ) + " joins "
245 + ( printProgressNodeToString( getExternalPhylogenyNode( otu2 ) ) ) + " [resulting in node "
246 + ( printProgressNodeToString( node ) ) + "]" );
249 private final String printProgressNodeToString( final PhylogenyNode n ) {
250 if ( n.isExternal() ) {
251 if ( ForesterUtil.isEmpty( n.getName() ) ) {
252 return Long.toString( n.getId() );
258 + ( ForesterUtil.isEmpty( n.getChildNode1().getName() ) ? n.getChildNode1().getId() : n.getChildNode1()
261 + ( ForesterUtil.isEmpty( n.getChildNode2().getName() ) ? n.getChildNode2().getId() : n.getChildNode2()
265 // only the values in the lower triangle are used.
266 // !matrix values will be changed!
267 private final void reset( final BasicSymmetricalDistanceMatrix distances ) {
268 _n = distances.getSize();
270 _r = new double[ _n ];
271 _mappings = new int[ _n ];
272 _rev_mappings = new int[ _n ];
273 _d_values = _d.getValues();
275 _s.initialize( distances );
277 System.out.println();
279 System.out.println( "----------------------------------------------------------------------------------" );
280 System.out.println();
281 System.out.println();
284 final private void printM() {
285 for( int j = 0; j < _d_values.length; ++j ) {
286 System.out.print( _external_nodes[ j ] );
287 System.out.print( "\t\t" );
288 for( int i = 0; i < _d_values[ j ].length; ++i ) {
289 System.out.print( DF.format( _d_values[ i ][ j ] ) );
290 System.out.print( " " );
292 System.out.println();
294 for( int j = 0; j < _n; ++j ) {
295 System.out.print( getExternalPhylogenyNode( j ) );
296 System.out.print( "\t\t" );
297 for( int i = 0; i < _n; ++i ) {
298 System.out.print( DF.format( _d_values[ _mappings[ i ] ][ _mappings[ j ] ] ) );
299 System.out.print( " " );
301 System.out.print( "\t\t" );
302 for( final Entry<Integer, SortedSet<Integer>> entry : _s.getSentrySet( _mappings[ j ] ) ) {
303 System.out.print( DF.format( ( double ) entry.getKey() / S.FACTOR ) + "=" );
304 boolean first = true;
305 for( final int v : entry.getValue() ) {
307 System.out.print( "," );
310 System.out.print( v );
312 System.out.print( " " );
314 System.out.println();
318 private final double updateM() {
319 calculateNetDivergences();
320 Double min_m = Double.MAX_VALUE;
323 final int n_minus_2 = _n - 2;
325 for( int j = 1; j < _n; ++j ) {
326 final double r_j = _r[ j ];
327 final int m_j = _mappings[ j ];
328 System.out.print( "j=" + j + " mj=" + m_j + ": " );
329 for( final Entry<Integer, SortedSet<Integer>> entry : _s.getSentrySet( m_j ) ) {
330 for( final int sorted_i : entry.getValue() ) {
331 System.out.print( sorted_i + " " );
332 System.out.print( "(" + DF.format( getDvalueUnmapped( sorted_i, m_j ) ) + ") " );
333 final double m = getDvalueUnmapped( sorted_i, m_j )
334 - ( ( _r[ _rev_mappings[ sorted_i ] ] + r_j ) / n_minus_2 );
335 if ( ( m < min_m ) ) {
336 _d_min = getDvalueUnmapped( sorted_i, m_j );
343 System.out.println();
344 for( final Entry<Integer, SortedSet<Integer>> entry : _s.getSentrySet( m_j ) ) {
345 for( final int sorted_i : entry.getValue() ) {
346 System.out.print( sorted_i );
347 System.out.print( "->" );
348 System.out.print( DF.format( _r[ sorted_i ] ) );
349 System.out.print( " " );
352 System.out.println();
354 for( int i = 0; i < j; ++i ) {
355 final double m = getDvalue( i, j ) - ( ( _r[ i ] + r_j ) / n_minus_2 );
358 _d_min = getDvalue( i, j );
364 System.out.println();
368 // otu2 will, in effect, be "deleted" from the matrix.
369 private final void updateMappings( final int otu2 ) {
370 for( int i = otu2; i < ( _mappings.length - 1 ); ++i ) {
371 System.out.print( _mappings[ i ] );
372 _mappings[ i ] = _mappings[ i + 1 ];
373 System.out.println( "----->" + _mappings[ i ] );
375 for( int i = 0; i < _mappings.length; ++i ) {
376 System.out.println( i + "-->" + _mappings[ i ] );
378 for( int i = 0; i < _n; ++i ) {
379 _rev_mappings[ _mappings[ i ] ] = i;
383 public final static NeighborJoiningR createInstance() {
384 return new NeighborJoiningR();
387 public final static NeighborJoiningR createInstance( final boolean verbose,
388 final int maximum_fraction_digits_for_distances ) {
389 return new NeighborJoiningR( verbose, maximum_fraction_digits_for_distances );