// FORESTER -- software libraries and applications
// for evolutionary biology research and applications.
//
-// Copyright (C) 2008-2009 Christian M. Zmasek
-// Copyright (C) 2008-2009 Burnham Institute for Medical Research
+// Copyright (C) 2014 Christian M. Zmasek
// All rights reserved
//
// This library is free software; you can redistribute it and/or
public final class NeighborJoining {
+ private final static DecimalFormat DF = new DecimalFormat( "0.00000" );
private BasicSymmetricalDistanceMatrix _d;
private double[][] _d_values;
- private double[][] _m_values;
- private double[] _r;
- private int _n;
+ private final DecimalFormat _df;
private PhylogenyNode[] _external_nodes;
private int[] _mappings;
+ private int _n;
+ private double[] _r;
private final boolean _verbose;
- private final DecimalFormat _df;
+ private int _min_i;
+ private int _min_j;
private NeighborJoining() {
_verbose = false;
_df.setRoundingMode( RoundingMode.HALF_UP );
}
- private final void printM() {
- System.out.println( "M:" );
- for( final double[] _m_value : _m_values ) {
- for( int j = 0; j < _m_values.length; j++ ) {
- System.out.print( _m_value[ j ] );
- System.out.print( " " );
- }
- System.out.println();
- }
- System.out.println();
- }
-
- private final void printD() {
- System.out.println( "D:" );
- for( final double[] _d_value : _d_values ) {
- for( int j = 0; j < _d_values.length; j++ ) {
- System.out.print( _d_value[ j ] );
- System.out.print( " " );
- }
- System.out.println();
- }
- System.out.println();
- }
-
- private final void calculateDistancesFromNewNode( final int otu1, final int otu2, final double d ) {
- // final int otu1_m = _mappings[ otu1 ];
- // final int otu2_m = _mappings[ otu2 ];
- // int i_m;
- for( int i = 0; i < _n; ++i ) {
- if ( ( i == otu1 ) || ( i == otu2 ) ) {
- continue;
- }
- if ( otu1 < i ) {
- _d_values[ _mappings[ otu1 ] ][ _mappings[ i ] ] = ( getValueFromD( otu1, i ) + getValueFromD( i, otu2 ) - d ) / 2;
- }
- else {
- _d_values[ _mappings[ i ] ][ _mappings[ otu1 ] ] = ( getValueFromD( otu1, i ) + getValueFromD( i, otu2 ) - d ) / 2;
- }
- //i_m = _mappings[ i ];
- //_d_values[ otu1_m ][ i_m ] = ( ( _d_values[ otu1_m ][ i_m ] + _d_values[ i_m ][ otu2_m ] ) - 2 ) / 2;
- }
- }
-
- private final void calculateNetDivergences() {
- double d;
- // int i_m;
- for( int i = 0; i < _n; ++i ) {
- d = 0;
- //i_m = _mappings[ i ];
- for( int n = 0; n < _n; ++n ) {
- //d += _d_values[ i_m ][ _mappings[ n ] ];
- if ( i != n ) {
- d += getValueFromD( i, n );
- }
- }
- _r[ i ] = d;
- }
- }
-
public final Phylogeny execute( final BasicSymmetricalDistanceMatrix distance ) {
reset( distance );
final Phylogeny phylogeny = new Phylogeny();
while ( _n > 2 ) {
- updateM();
// Calculates the minimal distance.
// If more than one minimal distances, always the first found is used
- // could randomize this, so that any would be returned in a randomized fashion...
- double minimum = _m_values[ 0 ][ 1 ];
- int otu1 = 0;
- int otu2 = 1;
- for( int j = 1; j < _n; ++j ) {
- for( int i = 0; i < j; ++i ) {
- if ( _m_values[ i ][ j ] < minimum ) {
- minimum = _m_values[ i ][ j ];
- otu1 = i;
- otu2 = j;
- }
- }
- }
+ updateM();
+ final int otu1 = _min_i;
+ final int otu2 = _min_j;
+ //System.out.println( _min_i + " " + _min_j );
// It is a condition that otu1 < otu2.
final PhylogenyNode node = new PhylogenyNode();
- final double d = getValueFromD( otu1, otu2 );
+ final double d = _d_values[ _mappings[ otu1 ] ][ _mappings[ otu2 ] ];
final double d1 = ( d / 2 ) + ( ( _r[ otu1 ] - _r[ otu2 ] ) / ( 2 * ( _n - 2 ) ) );
final double d2 = d - d1;
if ( _df == null ) {
updateMappings( otu2 );
--_n;
}
- final double d = getValueFromD( 0, 1 ) / 2;
+ final double d = _d_values[ _mappings[ 0 ] ][ _mappings[ 1 ] ] / 2;
if ( _df == null ) {
getExternalPhylogenyNode( 0 ).setDistanceToParent( d );
getExternalPhylogenyNode( 1 ).setDistanceToParent( d );
return pl;
}
- private final PhylogenyNode getExternalPhylogenyNode( final int i ) {
- return _external_nodes[ _mappings[ i ] ];
+ private final void calculateDistancesFromNewNode( final int otu1, final int otu2, final double d ) {
+ final int m_otu1 = _mappings[ otu1 ];
+ final int m_otu2 = _mappings[ otu2 ];
+ for( int i = 0; i < _n; ++i ) {
+ if ( ( i == otu1 ) || ( i == otu2 ) ) {
+ continue;
+ }
+ final int m_i = _mappings[ i ];
+ if ( otu1 < i ) {
+ if ( otu2 > i ) {
+ _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
+ //System.out.print( DF.format( _d_values[ m_otu1 ][ m_i ] ) );
+ }
+ else {
+ _d_values[ m_otu1 ][ m_i ] = ( _d_values[ m_otu1 ][ m_i ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
+ //System.out.print( DF.format( _d_values[ m_otu1 ][ m_i ] ) );
+ }
+ }
+ else {
+ if ( otu2 > i ) {
+ _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_i ][ m_otu2 ] - d ) / 2;
+ //System.out.print( DF.format( _d_values[ m_i ][ m_otu1 ] ) );
+ }
+ else {
+ _d_values[ m_i ][ m_otu1 ] = ( _d_values[ m_i ][ m_otu1 ] + _d_values[ m_otu2 ][ m_i ] - d ) / 2;
+ // System.out.print( DF.format( _d_values[ m_otu1 ][ m_i ] ) );
+ }
+ }
+ //System.out.print( " " );
+ }
}
- private final double getValueFromD( final int otu1, final int otu2 ) {
- if ( otu1 > otu2 ) {
- return _d_values[ _mappings[ otu2 ] ][ _mappings[ otu1 ] ];
+ private final void calculateNetDivergences() {
+ double d;
+ for( int i = 0; i < _n; ++i ) {
+ d = 0;
+ final int m_i = _mappings[ i ];
+ for( int n = 0; n < _n; ++n ) {
+ if ( i != n ) {
+ if ( i > n ) {
+ d += _d_values[ _mappings[ n ] ][ m_i ];
+ }
+ else {
+ d += _d_values[ m_i ][ _mappings[ n ] ];
+ }
+ }
+ }
+ _r[ i ] = d;
}
- return _d_values[ _mappings[ otu1 ] ][ _mappings[ otu2 ] ];
+ }
+
+ private final PhylogenyNode getExternalPhylogenyNode( final int i ) {
+ return _external_nodes[ _mappings[ i ] ];
}
private final void initExternalNodes() {
_external_nodes[ i ].setName( id );
}
else {
- _external_nodes[ i ].setName( "" + i );
+ _external_nodes[ i ].setName( Integer.toString( i ) );
}
_mappings[ i ] = i;
}
}
private final void printProgress( final int otu1, final int otu2 ) {
- final PhylogenyNode n1 = getExternalPhylogenyNode( otu1 );
- final PhylogenyNode n2 = getExternalPhylogenyNode( otu2 );
- System.out.println( "Node " + ( ForesterUtil.isEmpty( n1.getName() ) ? n1.getId() : n1.getName() ) + " joins "
- + ( ForesterUtil.isEmpty( n2.getName() ) ? n2.getId() : n2.getName() ) );
+ System.out.println( "Node " + printProgressNodeToString( getExternalPhylogenyNode( otu1 ) ) + " joins "
+ + ( printProgressNodeToString( getExternalPhylogenyNode( otu2 ) ) ) );
+ }
+
+ private final String printProgressNodeToString( final PhylogenyNode n ) {
+ if ( n.isExternal() ) {
+ if ( ForesterUtil.isEmpty( n.getName() ) ) {
+ return Long.toString( n.getId() );
+ }
+ return n.getName();
+ }
+ return n.getId()
+ + " ("
+ + ( ForesterUtil.isEmpty( n.getChildNode1().getName() ) ? n.getChildNode1().getId() : n.getChildNode1()
+ .getName() )
+ + "+"
+ + ( ForesterUtil.isEmpty( n.getChildNode2().getName() ) ? n.getChildNode2().getId() : n.getChildNode2()
+ .getName() ) + ")";
}
// only the values in the lower triangle are used.
_r = new double[ _n ];
_mappings = new int[ _n ];
_d_values = _d.getValues();
- _m_values = new double[ _n ][ _n ];
initExternalNodes();
}
private final void updateM() {
calculateNetDivergences();
- final double r_j;
- final int j_m;
- final int _n_2 = _n - 2;
+ Double min = Double.MAX_VALUE;
+ _min_i = -1;
+ _min_j = -1;
+ final int n_minus_2 = _n - 2;
for( int j = 1; j < _n; ++j ) {
- //r_j = _r[ j ];
- // j_m = _mappings[ j ];
+ final double r_j = _r[ j ];
+ final int m_j = _mappings[ j ];
for( int i = 0; i < j; ++i ) {
- _m_values[ i ][ j ] = getValueFromD( i, j ) - ( ( _r[ i ] + _r[ j ] ) / ( _n - 2 ) );
- //_m_values[ i ][ j ] = _d_values[ _mappings[ i ] ][ j_m ] - ( ( _r[ i ] + r_j ) / ( _n_2 ) );
+ final double m = _d_values[ _mappings[ i ] ][ m_j ] - ( ( _r[ i ] + r_j ) / n_minus_2 );
+ if ( m < min ) {
+ min = m;
+ _min_i = i;
+ _min_j = j;
+ }
}
}
+ // for( int j = 1; j < _n; ++j ) {
+ // final double r_j = _r[ j ];
+ // final int m_j = _mappings[ j ];
+ // for( int i = 0; i < j; ++i ) {
+ // System.out.print( i );
+ // System.out.print( "->" );
+ // System.out.print( DF.format( _r[ i ] ) );
+ // System.out.print( " " );
+ // }
+ // System.out.println();
+ // }
}
// otu2 will, in effect, be "deleted" from the matrix.
git://source.jalview.org / jalview.git / blobdiff