cluster.cpp

   1 /*
   2  *  cluster.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 8/14/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 #include "cluster.hpp"
  11 #include "rabundvector.hpp"
  12 #include "listvector.hpp"
  13
  14 /***********************************************************************/
  15
  16 Cluster::Cluster(RAbundVector* rav, ListVector* lv, SparseDistanceMatrix* dm, float c, string f, float cs) :
  17 rabund(rav), list(lv), dMatrix(dm), method(f), adjust(cs)
  18 {
  19         try {
  20
  21         mapWanted = false;  //set to true by mgcluster to speed up overlap merge
  22
  23         //save so you can modify as it changes in average neighbor
  24         cutoff = c;
  25         m = MothurOut::getInstance();
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "Cluster", "Cluster");
  29                 exit(1);
  30         }
  31 }
  32 /***********************************************************************/
  33 void Cluster::clusterBins(){
  34         try {
  35                 rabund->set(smallCol, rabund->get(smallRow)+rabund->get(smallCol));
  36                 rabund->set(smallRow, 0);
  37                 rabund->setLabel(toString(smallDist));
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "Cluster", "clusterBins");
  41                 exit(1);
  42         }
  43 }
  44 /***********************************************************************/
  45
  46 void Cluster::clusterNames(){
  47         try {
  48                 if (mapWanted) {  updateMap();  }
  49
  50                 list->set(smallCol, list->get(smallRow)+','+list->get(smallCol));
  51                 list->set(smallRow, "");
  52                 list->setLabel(toString(smallDist));
  53     }
  54         catch(exception& e) {
  55                 m->errorOut(e, "Cluster", "clusterNames");
  56                 exit(1);
  57         }
  58 }
  59 /***********************************************************************/
  60 void Cluster::update(double& cutOFF){
  61         try {
  62         smallCol = dMatrix->getSmallestCell(smallRow);
  63         nColCells = dMatrix->seqVec[smallCol].size();
  64         nRowCells = dMatrix->seqVec[smallRow].size();
  65
  66                 vector<int> foundCol(nColCells, 0);
  67         //cout << dMatrix->getNNodes() << " small cell: " << smallRow << '\t' << smallCol << endl;
  68                 int search;
  69                 bool changed;
  70
  71                 for (int i=nRowCells-1;i>=0;i--) {
  72             if (m->control_pressed) { break; }
  73
  74                         //if you are not the smallCell
  75                         if (dMatrix->seqVec[smallRow][i].index != smallCol) {
  76                 search = dMatrix->seqVec[smallRow][i].index;
  77
  78                                 bool merged = false;
  79                                 for (int j=0;j<nColCells;j++) {
  80
  81                                         if (dMatrix->seqVec[smallCol][j].index != smallRow) {  //if you are not the smallest distance
  82                                                 if (dMatrix->seqVec[smallCol][j].index == search) {
  83                                                         foundCol[j] = 1;
  84                                                         merged = true;
  85                                                         changed = updateDistance(dMatrix->seqVec[smallCol][j], dMatrix->seqVec[smallRow][i]);
  86                             dMatrix->updateCellCompliment(smallCol, j);
  87                                                         break;
  88                                                 }else if (dMatrix->seqVec[smallCol][j].index < search) { //we don't have a distance for this cell
  89                             if (adjust != -1.0) { //adjust
  90                                 merged = true;
  91                                 PDistCell value(search, adjust); //create a distance for the missing value
  92                                 int location = dMatrix->addCellSorted(smallCol, value);
  93                                 changed = updateDistance(dMatrix->seqVec[smallCol][location], dMatrix->seqVec[smallRow][i]);
  94                                 dMatrix->updateCellCompliment(smallCol, location);
  95                                 nColCells++;
  96                                 foundCol.push_back(0); //add a new found column
  97                                 //adjust value
  98                                 for (int k = foundCol.size()-1; k > location; k--) { foundCol[k] = foundCol[k-1]; }
  99                                 foundCol[location] = 1;
 100                             }
 101                             j+=nColCells;
 102                         }
 103                                         }
 104                                 }
 105                                 //if not merged it you need it for warning
 106                                 if ((!merged) && (method == "average" || method == "weighted")) {
 107                                         if (cutOFF > dMatrix->seqVec[smallRow][i].dist) {
 108                                                 cutOFF = dMatrix->seqVec[smallRow][i].dist;
 109                         //cout << "changing cutoff to " << cutOFF << endl;
 110                                         }
 111
 112                                 }
 113                                 dMatrix->rmCell(smallRow, i);
 114                         }
 115                 }
 116                 clusterBins();
 117                 clusterNames();
 118
 119                 // Special handling for singlelinkage case, not sure whether this
 120                 // could be avoided
 121                 for (int i=nColCells-1;i>=0;i--) {
 122                         if (foundCol[i] == 0) {
 123                 if (adjust != -1.0) { //adjust
 124                     PDistCell value(smallCol, adjust); //create a distance for the missing value
 125                     changed = updateDistance(dMatrix->seqVec[smallCol][i], value);
 126                     dMatrix->updateCellCompliment(smallCol, i);
 127                 }else {
 128                     if (method == "average" || method == "weighted") {
 129                         if (dMatrix->seqVec[smallCol][i].index != smallRow) { //if you are not hte smallest distance
 130                             if (cutOFF > dMatrix->seqVec[smallCol][i].dist) {
 131                                 cutOFF = dMatrix->seqVec[smallCol][i].dist;
 132                             }
 133                         }
 134                     }
 135                 }
 136                 dMatrix->rmCell(smallCol, i);
 137                         }
 138                 }
 139
 140         }
 141         catch(exception& e) {
 142                 m->errorOut(e, "Cluster", "update");
 143                 exit(1);
 144         }
 145 }
 146 /***********************************************************************/
 147 void Cluster::setMapWanted(bool f)  {
 148         try {
 149                 mapWanted = f;
 150
 151         //initialize map
 152                 for (int k = 0; k < list->getNumBins(); k++) {
 153
 154             string names = list->get(k);
 155
 156             //parse bin
 157             string individual = "";
 158             int binNameslength = names.size();
 159             for(int j=0;j<binNameslength;j++){
 160                 if(names[j] == ','){
 161                     seq2Bin[individual] = k;
 162                     individual = "";
 163                 }
 164                 else{  individual += names[j];  }
 165             }
 166             //get last name
 167             seq2Bin[individual] = k;
 168                 }
 169
 170         }
 171         catch(exception& e) {
 172                 m->errorOut(e, "Cluster", "setMapWanted");
 173                 exit(1);
 174         }
 175 }
 176 /***********************************************************************/
 177 void Cluster::updateMap() {
 178     try {
 179                 //update location of seqs in smallRow since they move to smallCol now
 180                 string names = list->get(smallRow);
 181
 182         string individual = "";
 183         int binNameslength = names.size();
 184         for(int j=0;j<binNameslength;j++){
 185             if(names[j] == ','){
 186                 seq2Bin[individual] = smallCol;
 187                 individual = "";
 188             }
 189             else{  individual += names[j];  }
 190         }
 191         //get last name
 192         seq2Bin[individual] = smallCol;
 193
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "Cluster", "updateMap");
 197                 exit(1);
 198         }
 199 }
 200 /***********************************************************************/
 201
 202
 203