[1573] | 1 | // UPGMATree.java |
---|
| 2 | // |
---|
| 3 | // (c) 1999-2001 PAL Development Core Team |
---|
| 4 | // |
---|
| 5 | // This package may be distributed under the |
---|
| 6 | // terms of the Lesser GNU General Public License (LGPL) |
---|
| 7 | |
---|
| 8 | // Known bugs and limitations: |
---|
| 9 | // - computational complexity O(numSeqs^3) |
---|
| 10 | // (this could be brought down to O(numSeqs^2) |
---|
| 11 | // but this needs more clever programming ...) |
---|
| 12 | |
---|
| 13 | |
---|
| 14 | package de.ugoe.cs.autoquest.tasktrees.alignment.pal.tree; |
---|
| 15 | |
---|
| 16 | import java.util.ArrayList; |
---|
| 17 | |
---|
| 18 | import de.ugoe.cs.autoquest.tasktrees.alignment.algorithms.NumberSequence; |
---|
| 19 | import de.ugoe.cs.autoquest.tasktrees.alignment.matrix.UPGMAMatrix; |
---|
| 20 | import de.ugoe.cs.autoquest.tasktrees.alignment.pal.misc.Identifier; |
---|
| 21 | |
---|
| 22 | |
---|
| 23 | /** |
---|
| 24 | * constructs a UPGMA tree from pairwise distances |
---|
| 25 | * |
---|
| 26 | * @version $Id: UPGMATree.java,v 1.9 2001/07/13 14:39:13 korbinian Exp $ |
---|
| 27 | * |
---|
| 28 | * @author Korbinian Strimmer |
---|
| 29 | * @author Alexei Drummond |
---|
| 30 | */ |
---|
[1579] | 31 | public class UPGMAAligningTree extends SimpleTree |
---|
[1573] | 32 | { |
---|
| 33 | // |
---|
| 34 | // Public stuff |
---|
| 35 | // |
---|
| 36 | |
---|
| 37 | /** |
---|
| 38 | * constructor UPGMA tree |
---|
| 39 | * @param numberseqs |
---|
| 40 | * |
---|
| 41 | * @param m distance matrix |
---|
| 42 | */ |
---|
[1579] | 43 | public UPGMAAligningTree(ArrayList<NumberSequence> numberseqs, UPGMAMatrix m) |
---|
[1573] | 44 | { |
---|
| 45 | if (m.size() < 2) |
---|
| 46 | { |
---|
| 47 | new IllegalArgumentException("LESS THAN 2 TAXA IN DISTANCE MATRIX"); |
---|
| 48 | } |
---|
| 49 | |
---|
[1577] | 50 | this.numberseqs = numberseqs; |
---|
[1573] | 51 | init(m); |
---|
| 52 | |
---|
| 53 | while (true) |
---|
| 54 | { |
---|
| 55 | findNextPair(); |
---|
| 56 | newBranchLengths(); |
---|
| 57 | |
---|
| 58 | if (numClusters == 2) |
---|
| 59 | { |
---|
| 60 | break; |
---|
| 61 | } |
---|
| 62 | |
---|
| 63 | newCluster(); |
---|
| 64 | } |
---|
| 65 | |
---|
| 66 | finish(); |
---|
| 67 | createNodeList(); |
---|
| 68 | } |
---|
| 69 | |
---|
| 70 | |
---|
| 71 | // |
---|
| 72 | // Private stuff |
---|
| 73 | // |
---|
| 74 | private ArrayList<NumberSequence> numberseqs; |
---|
| 75 | private int numClusters; |
---|
| 76 | private int besti, abi; |
---|
| 77 | private int bestj, abj; |
---|
| 78 | private int[] alias; |
---|
| 79 | private double[][] distance; |
---|
| 80 | |
---|
| 81 | private double[] height; |
---|
| 82 | private int[] oc; |
---|
| 83 | |
---|
| 84 | private double getDist(int a, int b) |
---|
| 85 | { |
---|
| 86 | return distance[alias[a]][alias[b]]; |
---|
| 87 | } |
---|
| 88 | |
---|
| 89 | private void init(UPGMAMatrix m) |
---|
| 90 | { |
---|
| 91 | numClusters = m.size(); |
---|
| 92 | |
---|
| 93 | distance = new double[numClusters][numClusters]; |
---|
| 94 | for (int i = 0; i < numClusters; i++) |
---|
| 95 | { |
---|
| 96 | for (int j = 0; j < numClusters; j++) |
---|
| 97 | { |
---|
| 98 | distance[i][j] = m.get(i,j); |
---|
| 99 | } |
---|
| 100 | } |
---|
| 101 | |
---|
| 102 | for (int i = 0; i < numClusters; i++) |
---|
| 103 | { |
---|
| 104 | Node tmp = NodeFactory.createNode(); |
---|
| 105 | tmp.setIdentifier(new Identifier(Integer.toString(i))); |
---|
[1577] | 106 | tmp.addSequence(numberseqs.get(i)); |
---|
[1573] | 107 | getRoot().addChild(tmp); |
---|
| 108 | } |
---|
| 109 | |
---|
| 110 | alias = new int[numClusters]; |
---|
| 111 | for (int i = 0; i < numClusters; i++) |
---|
| 112 | { |
---|
| 113 | alias[i] = i; |
---|
| 114 | } |
---|
| 115 | |
---|
| 116 | height = new double[numClusters]; |
---|
| 117 | oc = new int[numClusters]; |
---|
| 118 | for (int i = 0; i < numClusters; i++) |
---|
| 119 | { |
---|
| 120 | height[i] = 0.0; |
---|
| 121 | oc[i] = 1; |
---|
| 122 | } |
---|
| 123 | } |
---|
| 124 | |
---|
| 125 | private void finish() |
---|
| 126 | { |
---|
| 127 | distance = null; |
---|
| 128 | } |
---|
| 129 | |
---|
| 130 | private void findNextPair() |
---|
| 131 | { |
---|
| 132 | besti = 0; |
---|
| 133 | bestj = 1; |
---|
| 134 | double dmin = getDist(0, 1); |
---|
| 135 | for (int i = 0; i < numClusters-1; i++) |
---|
| 136 | { |
---|
| 137 | for (int j = i+1; j < numClusters; j++) |
---|
| 138 | { |
---|
| 139 | if (getDist(i, j) < dmin) |
---|
| 140 | { |
---|
| 141 | dmin = getDist(i, j); |
---|
| 142 | besti = i; |
---|
| 143 | bestj = j; |
---|
| 144 | } |
---|
| 145 | } |
---|
| 146 | } |
---|
| 147 | abi = alias[besti]; |
---|
| 148 | abj = alias[bestj]; |
---|
[1577] | 149 | //System.out.println("Found best pair: " + abi + "/" +abj + " - "+ besti+ "/"+bestj +" with distance " + dmin); |
---|
[1573] | 150 | |
---|
| 151 | } |
---|
| 152 | |
---|
| 153 | private void newBranchLengths() |
---|
| 154 | { |
---|
| 155 | double dij = getDist(besti, bestj); |
---|
| 156 | |
---|
| 157 | getRoot().getChild(besti).setBranchLength(dij/2.0-height[abi]); |
---|
| 158 | getRoot().getChild(bestj).setBranchLength(dij/2.0-height[abj]); |
---|
| 159 | } |
---|
| 160 | |
---|
| 161 | private void newCluster() |
---|
| 162 | { |
---|
| 163 | // Update distances |
---|
| 164 | for (int k = 0; k < numClusters; k++) |
---|
| 165 | { |
---|
| 166 | if (k != besti && k != bestj) |
---|
| 167 | { |
---|
| 168 | int ak = alias[k]; |
---|
| 169 | double updated = updatedDistance(besti,bestj,k); |
---|
| 170 | distance[ak][abi] = distance[abi][ak] = updated; |
---|
| 171 | } |
---|
| 172 | } |
---|
| 173 | distance[abi][abi] = 0.0; |
---|
| 174 | |
---|
| 175 | // Update UPGMA variables |
---|
| 176 | height[abi] = getDist(besti, bestj)/2.0; |
---|
| 177 | oc[abi] += oc[abj]; |
---|
| 178 | |
---|
| 179 | // Index besti now represent the new cluster |
---|
| 180 | getRoot().joinChildren(besti, bestj); |
---|
| 181 | |
---|
| 182 | // Update alias |
---|
| 183 | for (int i = bestj; i < numClusters-1; i++) |
---|
| 184 | { |
---|
| 185 | alias[i] = alias[i+1]; |
---|
| 186 | } |
---|
| 187 | |
---|
| 188 | numClusters--; |
---|
| 189 | } |
---|
| 190 | |
---|
| 191 | |
---|
| 192 | /** |
---|
| 193 | * compute updated distance between the new cluster (i,j) |
---|
| 194 | * to any other cluster k |
---|
| 195 | */ |
---|
| 196 | private double updatedDistance(int i, int j, int k) |
---|
| 197 | { |
---|
| 198 | int ai = alias[i]; |
---|
| 199 | int aj = alias[j]; |
---|
| 200 | |
---|
| 201 | double ocsum = (double) (oc[ai]+oc[aj]); |
---|
| 202 | double idist = getDist(k,i); |
---|
| 203 | double jdist = getDist(k,j); |
---|
| 204 | //TODO: Dirty hack to deal with infinity, insert proper solution here |
---|
| 205 | if(Double.isInfinite(idist)) { |
---|
| 206 | idist = 100; |
---|
| 207 | } |
---|
| 208 | if(Double.isInfinite(jdist)) { |
---|
| 209 | jdist = 100; |
---|
| 210 | } |
---|
| 211 | |
---|
| 212 | return (oc[ai]/ocsum)*idist+ |
---|
| 213 | (oc[aj]/ocsum)*jdist; |
---|
| 214 | } |
---|
| 215 | } |
---|