clusterclassic.cpp

   1 /*
   2  *  clusterclassic.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/29/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "clusterclassic.h"
  11 #include "progress.hpp"
  12
  13 /***********************************************************************/
  14 ClusterClassic::ClusterClassic(float c, string f, bool s) : method(f), smallDist(1e6), nseqs(0), sim(s) {
  15         try {
  16                 mapWanted = false;  //set to true by mgcluster to speed up overlap merge
  17
  18                 //save so you can modify as it changes in average neighbor
  19                 cutoff = c;
  20                 aboveCutoff = cutoff + 10000.0;
  21                 m = MothurOut::getInstance();
  22         }
  23         catch(exception& e) {
  24                 m->errorOut(e, "ClusterClassic", "ClusterClassic");
  25                 exit(1);
  26         }
  27 }
  28 /***********************************************************************/
  29 int ClusterClassic::readPhylipFile(string filename, NameAssignment* nameMap) {
  30         try {
  31                 double distance;
  32                 int square;
  33                 string name;
  34                 vector<string> matrixNames;
  35
  36                 ifstream fileHandle;
  37                 m->openInputFile(filename, fileHandle);
  38
  39                 fileHandle >> nseqs >> name;
  40
  41                 matrixNames.push_back(name);
  42
  43                 if(nameMap == NULL){
  44                                 list = new ListVector(nseqs);
  45                                 list->set(0, name);
  46                 }
  47                 else{
  48                                 list = new ListVector(nameMap->getListVector());
  49                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  50                 }
  51
  52                 //initialize distance matrix to cutoff
  53                 dMatrix.resize(nseqs);
  54                 //colDist temp(0, 0, aboveCutoff);
  55                 //rowSmallDists.resize(nseqs, temp);
  56                 for (int i = 1; i < nseqs; i++) {
  57                         dMatrix[i].resize(i, aboveCutoff);
  58                 }
  59
  60
  61                 char d;
  62                 while((d=fileHandle.get()) != EOF){
  63
  64                                 if(isalnum(d)){
  65                                                 square = 1;
  66                                                 fileHandle.putback(d);
  67                                                 for(int i=0;i<nseqs;i++){
  68                                                                 fileHandle >> distance;
  69                                                 }
  70                                                 break;
  71                                 }
  72                                 if(d == '\n'){
  73                                                 square = 0;
  74                                                 break;
  75                                 }
  76                 }
  77
  78                 Progress* reading;
  79
  80                 if(square == 0){
  81
  82                                 reading = new Progress("Reading matrix:     ", nseqs * (nseqs - 1) / 2);
  83
  84                                 int        index = 0;
  85
  86                                 for(int i=1;i<nseqs;i++){
  87                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
  88
  89                                                 fileHandle >> name;
  90                                                 matrixNames.push_back(name);
  91
  92
  93                                                 //there's A LOT of repeated code throughout this method...
  94                                                 if(nameMap == NULL){
  95                                                                 list->set(i, name);
  96
  97                                                                 for(int j=0;j<i;j++){
  98
  99                                                                                 if (m->control_pressed) { delete reading; fileHandle.close(); return 0;  }
 100
 101                                                                                 fileHandle >> distance;
 102
 103                                                                                 if (distance == -1) { distance = 1000000; }
 104                                                                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 105
 106                                                                                 //if(distance < cutoff){
 107                                                                                         dMatrix[i][j] = distance;
 108                                                                                         if (distance < smallDist) { smallDist = distance; }
 109                                                                                         //if (rowSmallDists[i].dist > distance) {  rowSmallDists[i].dist = distance; rowSmallDists[i].col = j; rowSmallDists[i].row = i; }
 110                                                                                         //if (rowSmallDists[j].dist > distance) {  rowSmallDists[j].dist = distance; rowSmallDists[j].col = i; rowSmallDists[j].row = j; }
 111                                                                                 //}
 112                                                                                 index++;
 113                                                                                 reading->update(index);
 114                                                                 }
 115
 116                                                 }
 117                                                 else{
 118                                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 119
 120                                                                 for(int j=0;j<i;j++){
 121                                                                                 fileHandle >> distance;
 122
 123                                                                                 if (m->control_pressed) { delete reading; fileHandle.close(); return 0;  }
 124
 125                                                                                 if (distance == -1) { distance = 1000000; }
 126                                                                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 127
 128                                                                                 //if(distance < cutoff){
 129                                                                                         if (distance < smallDist) { smallDist = distance; }
 130
 131                                                                                         int row = nameMap->get(matrixNames[i]);
 132                                                                                         int col = nameMap->get(matrixNames[j]);
 133
 134                                                                                         if (row < col) {  dMatrix[col][row] = distance; }
 135                                                                                         else { dMatrix[row][col] = distance; }
 136
 137                                                                                         //if (rowSmallDists[row].dist > distance) {  rowSmallDists[row].dist = distance; rowSmallDists[row].col = col; rowSmallDists[row].row = row; }
 138                                                                                         //if (rowSmallDists[col].dist > distance) {  rowSmallDists[col].dist = distance; rowSmallDists[col].col = row; rowSmallDists[col].row = col; }
 139                                                                                 //}
 140                                                                                 index++;
 141                                                                                 reading->update(index);
 142                                                                 }
 143                                                 }
 144                                 }
 145                 }
 146                 else{
 147
 148                                 reading = new Progress("Reading matrix:     ", nseqs * nseqs);
 149
 150                                 int index = nseqs;
 151
 152                                 for(int i=1;i<nseqs;i++){
 153                                                 fileHandle >> name;
 154                                                 matrixNames.push_back(name);
 155
 156                                                 if(nameMap == NULL){
 157                                                                 list->set(i, name);
 158                                                                 for(int j=0;j<nseqs;j++){
 159                                                                                 fileHandle >> distance;
 160
 161                                                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 162
 163                                                                                 if (distance == -1) { distance = 1000000; }
 164                                                                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 165
 166                                                                                 if(j < i){
 167                                                                                         if (distance < smallDist) { smallDist = distance; }
 168
 169                                                                                         dMatrix[i][j] = distance;
 170                                                                                         //if (rowSmallDists[i].dist > distance) {  rowSmallDists[i].dist = distance; rowSmallDists[i].col = j; rowSmallDists[i].row = i; }
 171                                                                                         //if (rowSmallDists[j].dist > distance) {  rowSmallDists[j].dist = distance; rowSmallDists[j].col = i; rowSmallDists[j].row = j; }
 172                                                                                 }
 173                                                                                 index++;
 174                                                                                 reading->update(index);
 175                                                                 }
 176
 177                                                 }
 178                                                 else{
 179                                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 180
 181                                                                 for(int j=0;j<nseqs;j++){
 182                                                                                 fileHandle >> distance;
 183
 184                                                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 185
 186                                                                            if (distance == -1) { distance = 1000000; }
 187                                                                                 else if (sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 188
 189                                                                                 if(j < i){
 190                                                                                         if (distance < smallDist) { smallDist = distance; }
 191
 192                                                                                         int row = nameMap->get(matrixNames[i]);
 193                                                                                         int col = nameMap->get(matrixNames[j]);
 194
 195                                                                                         if (row < col) {  dMatrix[col][row] = distance; }
 196                                                                                         else { dMatrix[row][col] = distance; }
 197
 198                                                                                         //if (rowSmallDists[row].dist > distance) {  rowSmallDists[row].dist = distance; rowSmallDists[row].col = col; rowSmallDists[row].row = row; }
 199                                                                                         //if (rowSmallDists[col].dist > distance) {  rowSmallDists[col].dist = distance; rowSmallDists[col].col = row; rowSmallDists[col].row = col; }
 200                                                                                 }
 201                                                                                 index++;
 202                                                                                 reading->update(index);
 203                                                                 }
 204                                                 }
 205                                 }
 206                 }
 207
 208                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 209
 210                 reading->finish();
 211                 delete reading;
 212
 213                 list->setLabel("0");
 214                 rabund = new RAbundVector(list->getRAbundVector());
 215
 216                 fileHandle.close();
 217
 218                 return 0;
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "ClusterClassic", "readPhylipFile");
 222                 exit(1);
 223         }
 224
 225 }
 226 /***********************************************************************/
 227 //sets smallCol and smallRow, returns distance
 228 double ClusterClassic::getSmallCell() {
 229         try {
 230
 231                 smallDist = aboveCutoff;
 232                 smallRow = 1;
 233                 smallCol = 0;
 234
 235                 vector<colDist> mins;
 236
 237                 for(int i=1;i<nseqs;i++){
 238                         for(int j=0;j<i;j++){
 239                                 if (dMatrix[i][j] < smallDist) {
 240                                         mins.clear();
 241                                         colDist temp(i, j, dMatrix[i][j]);
 242                                         mins.push_back(temp);
 243                                         smallDist = dMatrix[i][j];
 244                                 }else if (dMatrix[i][j] == smallDist) {
 245                                         colDist temp(i, j, dMatrix[i][j]);
 246                                         mins.push_back(temp);
 247
 248                                 }
 249                         }
 250                 }
 251
 252                 if (mins.size() > 0) {
 253                         int zrand = 0;
 254                         if (mins.size() > 1) {
 255                                 //pick random number between 0 and mins.size()
 256                                 zrand = (int)((float)(rand()) / (RAND_MAX / (mins.size()-1) + 1));
 257                         }
 258
 259                         smallRow = mins[zrand].row;
 260                         smallCol = mins[zrand].col;
 261
 262                 }
 263         //cout << smallRow << '\t' << smallCol << '\t' << smallDist << endl;
 264                 //eliminate smallCell
 265                 if (smallRow < smallCol) { dMatrix[smallCol][smallRow] = aboveCutoff; }
 266                 else { dMatrix[smallRow][smallCol] = aboveCutoff; }
 267
 268                 return smallDist;
 269
 270         }
 271         catch(exception& e) {
 272                 m->errorOut(e, "ClusterClassic", "getSmallCell");
 273                 exit(1);
 274         }
 275 }
 276 /***********************************************************************/
 277 void ClusterClassic::clusterBins(){
 278         try {
 279         //      cout << smallCol << '\t' << smallRow << '\t' << smallDist << '\t' << rabund->get(smallRow) << '\t' << rabund->get(smallCol);
 280
 281                 rabund->set(smallRow, rabund->get(smallRow)+rabund->get(smallCol));
 282                 rabund->set(smallCol, 0);
 283                 /*for (int i = smallCol+1; i < rabund->size(); i++) {
 284                         rabund->set((i-1), rabund->get(i));
 285                 }
 286                 rabund->resize((rabund->size()-1));*/
 287                 rabund->setLabel(toString(smallDist));
 288
 289         //      cout << '\t' << rabund->get(smallRow) << '\t' << rabund->get(smallCol) << endl;
 290         }
 291         catch(exception& e) {
 292                 m->errorOut(e, "ClusterClassic", "clusterBins");
 293                 exit(1);
 294         }
 295 }
 296 /***********************************************************************/
 297 void ClusterClassic::clusterNames(){
 298         try {
 299         //      cout << smallCol << '\t' << smallRow << '\t' << smallDist << '\t' << list->get(smallRow) << '\t' << list->get(smallCol);
 300                 if (mapWanted) {  updateMap();  }
 301
 302                 list->set(smallRow, list->get(smallRow)+','+list->get(smallCol));
 303                 list->set(smallCol, "");
 304                 /*for (int i = smallCol+1; i < list->size(); i++) {
 305                         list->set((i-1), list->get(i));
 306                 }
 307                 list->resize((list->size()-1));*/
 308                 list->setLabel(toString(smallDist));
 309
 310         //      cout << '\t' << list->get(smallRow) << '\t' << list->get(smallCol) << endl;
 311     }
 312         catch(exception& e) {
 313                 m->errorOut(e, "ClusterClassic", "clusterNames");
 314                 exit(1);
 315         }
 316 }
 317 /***********************************************************************/
 318 void ClusterClassic::update(double& cutOFF){
 319         try {
 320 //print();
 321                 getSmallCell();
 322
 323                 int r, c;
 324                 r = smallRow; c = smallCol;
 325
 326                 for(int i=0;i<nseqs;i++){
 327                         if(i != r && i != c){
 328                                 double distRow, distCol, newDist;
 329                                 if (i > r) { distRow = dMatrix[i][r]; }
 330                                 else { distRow =  dMatrix[r][i]; }
 331
 332                                 if (i > c) { distCol = dMatrix[i][c]; dMatrix[i][c] = aboveCutoff; } //like removeCell
 333                                 else { distCol =  dMatrix[c][i]; dMatrix[c][i] = aboveCutoff; }
 334
 335                                 if(method == "furthest"){
 336                                         newDist = max(distRow, distCol);
 337                                 }
 338                                 else if (method == "average"){
 339                                         int rowBin = rabund->get(r);
 340                                         int colBin = rabund->get(c);
 341                                         newDist = (colBin * distCol + rowBin * distRow) / (rowBin + colBin);
 342                                 }
 343                                 else if (method == "weighted"){
 344                                         newDist = (distCol + distRow) / 2.0;
 345                                 }
 346                                 else if (method == "nearest"){
 347                                         newDist = min(distRow, distCol);
 348                                 }
 349                                 //cout << "newDist = " << newDist << endl;
 350                                 if (i > r) { dMatrix[i][r] = newDist; }
 351                                 else { dMatrix[r][i] = newDist; }
 352
 353                         }
 354                 }
 355
 356                 clusterBins();
 357                 clusterNames();
 358
 359                 //resize each row
 360                 /*for(int i=0;i<nseqs;i++){
 361                         for(int j=c+1;j<dMatrix[i].size();j++){
 362                                 dMatrix[i][j-1]=dMatrix[i][j];
 363                         }
 364                 }
 365
 366                 //resize each col
 367                 for(int i=c+1;i<nseqs;i++){
 368                         for(int j=0;j < dMatrix[i-1].size();j++){
 369                                 dMatrix[i-1][j]=dMatrix[i][j];
 370                         }
 371                 }
 372
 373                 nseqs--;
 374                 dMatrix.pop_back();*/
 375
 376         }
 377         catch(exception& e) {
 378                 m->errorOut(e, "ClusterClassic", "update");
 379                 exit(1);
 380         }
 381 }
 382 /***********************************************************************/
 383 void ClusterClassic::setMapWanted(bool f)  {
 384         try {
 385                 mapWanted = f;
 386
 387                 //initialize map
 388                 for (int i = 0; i < list->getNumBins(); i++) {
 389
 390                         //parse bin
 391                         string names = list->get(i);
 392                         while (names.find_first_of(',') != -1) {
 393                                 //get name from bin
 394                                 string name = names.substr(0,names.find_first_of(','));
 395                                 //save name and bin number
 396                                 seq2Bin[name] = i;
 397                                 names = names.substr(names.find_first_of(',')+1, names.length());
 398                         }
 399
 400                         //get last name
 401                         seq2Bin[names] = i;
 402                 }
 403
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "ClusterClassic", "setMapWanted");
 407                 exit(1);
 408         }
 409 }
 410 /***********************************************************************/
 411 void ClusterClassic::updateMap() {
 412 try {
 413                 //update location of seqs in smallRow since they move to smallCol now
 414                 string names = list->get(smallRow);
 415                 while (names.find_first_of(',') != -1) {
 416                         //get name from bin
 417                         string name = names.substr(0,names.find_first_of(','));
 418                         //save name and bin number
 419                         seq2Bin[name] = smallCol;
 420                         names = names.substr(names.find_first_of(',')+1, names.length());
 421                 }
 422
 423                 //get last name
 424                 seq2Bin[names] = smallCol;
 425
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "ClusterClassic", "updateMap");
 429                 exit(1);
 430         }
 431 }
 432 /***********************************************************************/
 433 void ClusterClassic::print() {
 434 try {
 435                 //update location of seqs in smallRow since they move to smallCol now
 436                 for (int i = 0; i < dMatrix.size(); i++) {
 437                         m->mothurOut("row = " + toString(i) + "\t");
 438                         for (int j = 0; j < dMatrix[i].size(); j++) {
 439                                 m->mothurOut(toString(dMatrix[i][j]) + "\t");
 440                         }
 441                         m->mothurOutEndLine();
 442                 }
 443         }
 444         catch(exception& e) {
 445                 m->errorOut(e, "ClusterClassic", "updateMap");
 446                 exit(1);
 447         }
 448 }
 449 /***********************************************************************/
 450