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) : method(f), smallDist(1e6), nseqs(0) {
  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                 globaldata = GlobalData::getInstance();
  23         }
  24         catch(exception& e) {
  25                 m->errorOut(e, "ClusterClassic", "ClusterClassic");
  26                 exit(1);
  27         }
  28 }
  29 /***********************************************************************/
  30 int ClusterClassic::readPhylipFile(string filename, NameAssignment* nameMap) {
  31         try {
  32                 double distance;
  33                 int square;
  34                 string name;
  35                 vector<string> matrixNames;
  36
  37                 ifstream fileHandle;
  38                 m->openInputFile(filename, fileHandle);
  39
  40                 fileHandle >> nseqs >> name;
  41
  42                 matrixNames.push_back(name);
  43
  44                 if(nameMap == NULL){
  45                                 list = new ListVector(nseqs);
  46                                 list->set(0, name);
  47                 }
  48                 else{
  49                                 list = new ListVector(nameMap->getListVector());
  50                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
  51                 }
  52
  53                 //initialize distance matrix to cutoff
  54                 dMatrix.resize(nseqs);
  55                 //colDist temp(0, 0, aboveCutoff);
  56                 //rowSmallDists.resize(nseqs, temp);
  57                 for (int i = 1; i < nseqs; i++) {
  58                         dMatrix[i].resize(i, aboveCutoff);
  59                 }
  60
  61
  62                 char d;
  63                 while((d=fileHandle.get()) != EOF){
  64
  65                                 if(isalnum(d)){
  66                                                 square = 1;
  67                                                 fileHandle.putback(d);
  68                                                 for(int i=0;i<nseqs;i++){
  69                                                                 fileHandle >> distance;
  70                                                 }
  71                                                 break;
  72                                 }
  73                                 if(d == '\n'){
  74                                                 square = 0;
  75                                                 break;
  76                                 }
  77                 }
  78
  79                 Progress* reading;
  80
  81                 if(square == 0){
  82
  83                                 reading = new Progress("Reading matrix:     ", nseqs * (nseqs - 1) / 2);
  84
  85                                 int        index = 0;
  86
  87                                 for(int i=1;i<nseqs;i++){
  88                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
  89
  90                                                 fileHandle >> name;
  91                                                 matrixNames.push_back(name);
  92
  93
  94                                                 //there's A LOT of repeated code throughout this method...
  95                                                 if(nameMap == NULL){
  96                                                                 list->set(i, name);
  97
  98                                                                 for(int j=0;j<i;j++){
  99
 100                                                                                 if (m->control_pressed) { delete reading; fileHandle.close(); return 0;  }
 101
 102                                                                                 fileHandle >> distance;
 103
 104                                                                                 if (distance == -1) { distance = 1000000; }
 105                                                                                 else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 106
 107                                                                                 //if(distance < cutoff){
 108                                                                                         dMatrix[i][j] = distance;
 109                                                                                         if (distance < smallDist) { smallDist = distance; }
 110                                                                                         //if (rowSmallDists[i].dist > distance) {  rowSmallDists[i].dist = distance; rowSmallDists[i].col = j; rowSmallDists[i].row = i; }
 111                                                                                         //if (rowSmallDists[j].dist > distance) {  rowSmallDists[j].dist = distance; rowSmallDists[j].col = i; rowSmallDists[j].row = j; }
 112                                                                                 //}
 113                                                                                 index++;
 114                                                                                 reading->update(index);
 115                                                                 }
 116
 117                                                 }
 118                                                 else{
 119                                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 120
 121                                                                 for(int j=0;j<i;j++){
 122                                                                                 fileHandle >> distance;
 123
 124                                                                                 if (m->control_pressed) { delete reading; fileHandle.close(); return 0;  }
 125
 126                                                                                 if (distance == -1) { distance = 1000000; }
 127                                                                                 else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 128
 129                                                                                 //if(distance < cutoff){
 130                                                                                         if (distance < smallDist) { smallDist = distance; }
 131
 132                                                                                         int row = nameMap->get(matrixNames[i]);
 133                                                                                         int col = nameMap->get(matrixNames[j]);
 134
 135                                                                                         if (row < col) {  dMatrix[col][row] = distance; }
 136                                                                                         else { dMatrix[row][col] = distance; }
 137
 138                                                                                         //if (rowSmallDists[row].dist > distance) {  rowSmallDists[row].dist = distance; rowSmallDists[row].col = col; rowSmallDists[row].row = row; }
 139                                                                                         //if (rowSmallDists[col].dist > distance) {  rowSmallDists[col].dist = distance; rowSmallDists[col].col = row; rowSmallDists[col].row = col; }
 140                                                                                 //}
 141                                                                                 index++;
 142                                                                                 reading->update(index);
 143                                                                 }
 144                                                 }
 145                                 }
 146                 }
 147                 else{
 148
 149                                 reading = new Progress("Reading matrix:     ", nseqs * nseqs);
 150
 151                                 int index = nseqs;
 152
 153                                 for(int i=1;i<nseqs;i++){
 154                                                 fileHandle >> name;
 155                                                 matrixNames.push_back(name);
 156
 157                                                 if(nameMap == NULL){
 158                                                                 list->set(i, name);
 159                                                                 for(int j=0;j<nseqs;j++){
 160                                                                                 fileHandle >> distance;
 161
 162                                                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 163
 164                                                                                 if (distance == -1) { distance = 1000000; }
 165                                                                                 else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 166
 167                                                                                 if(j < i){
 168                                                                                         if (distance < smallDist) { smallDist = distance; }
 169
 170                                                                                         dMatrix[i][j] = distance;
 171                                                                                         //if (rowSmallDists[i].dist > distance) {  rowSmallDists[i].dist = distance; rowSmallDists[i].col = j; rowSmallDists[i].row = i; }
 172                                                                                         //if (rowSmallDists[j].dist > distance) {  rowSmallDists[j].dist = distance; rowSmallDists[j].col = i; rowSmallDists[j].row = j; }
 173                                                                                 }
 174                                                                                 index++;
 175                                                                                 reading->update(index);
 176                                                                 }
 177
 178                                                 }
 179                                                 else{
 180                                                                 if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 181
 182                                                                 for(int j=0;j<nseqs;j++){
 183                                                                                 fileHandle >> distance;
 184
 185                                                                                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 186
 187                                                                            if (distance == -1) { distance = 1000000; }
 188                                                                                 else if (globaldata->sim) { distance = 1.0 - distance;  }  //user has entered a sim matrix that we need to convert.
 189
 190                                                                                 if(j < i){
 191                                                                                         if (distance < smallDist) { smallDist = distance; }
 192
 193                                                                                         int row = nameMap->get(matrixNames[i]);
 194                                                                                         int col = nameMap->get(matrixNames[j]);
 195
 196                                                                                         if (row < col) {  dMatrix[col][row] = distance; }
 197                                                                                         else { dMatrix[row][col] = distance; }
 198
 199                                                                                         //if (rowSmallDists[row].dist > distance) {  rowSmallDists[row].dist = distance; rowSmallDists[row].col = col; rowSmallDists[row].row = row; }
 200                                                                                         //if (rowSmallDists[col].dist > distance) {  rowSmallDists[col].dist = distance; rowSmallDists[col].col = row; rowSmallDists[col].row = col; }
 201                                                                                 }
 202                                                                                 index++;
 203                                                                                 reading->update(index);
 204                                                                 }
 205                                                 }
 206                                 }
 207                 }
 208
 209                 if (m->control_pressed) {  fileHandle.close();  delete reading; return 0; }
 210
 211                 reading->finish();
 212                 delete reading;
 213
 214                 list->setLabel("0");
 215                 rabund = new RAbundVector(list->getRAbundVector());
 216
 217                 fileHandle.close();
 218
 219                 return 0;
 220         }
 221         catch(exception& e) {
 222                 m->errorOut(e, "ClusterClassic", "readPhylipFile");
 223                 exit(1);
 224         }
 225
 226 }
 227 /***********************************************************************/
 228 //sets smallCol and smallRow, returns distance
 229 double ClusterClassic::getSmallCell() {
 230         try {
 231
 232                 smallDist = aboveCutoff;
 233                 smallRow = 1;
 234                 smallCol = 0;
 235
 236                 vector<colDist> mins;
 237
 238                 for(int i=1;i<nseqs;i++){
 239                         for(int j=0;j<i;j++){
 240                                 if (dMatrix[i][j] < smallDist) {
 241                                         mins.clear();
 242                                         colDist temp(i, j, dMatrix[i][j]);
 243                                         mins.push_back(temp);
 244                                         smallDist = dMatrix[i][j];
 245                                 }else if (dMatrix[i][j] == smallDist) {
 246                                         colDist temp(i, j, dMatrix[i][j]);
 247                                         mins.push_back(temp);
 248
 249                                 }
 250                         }
 251                 }
 252
 253                 if (mins.size() > 0) {
 254                         int zrand = 0;
 255                         if (mins.size() > 1) {
 256                                 //pick random number between 0 and mins.size()
 257                                 zrand = (int)((float)(rand()) / (RAND_MAX / (mins.size()-1) + 1));
 258                         }
 259
 260                         smallRow = mins[zrand].row;
 261                         smallCol = mins[zrand].col;
 262
 263                 }
 264         //cout << smallRow << '\t' << smallCol << '\t' << smallDist << endl;
 265                 //eliminate smallCell
 266                 if (smallRow < smallCol) { dMatrix[smallCol][smallRow] = aboveCutoff; }
 267                 else { dMatrix[smallRow][smallCol] = aboveCutoff; }
 268
 269                 return smallDist;
 270
 271         }
 272         catch(exception& e) {
 273                 m->errorOut(e, "ClusterClassic", "getSmallCell");
 274                 exit(1);
 275         }
 276 }
 277 /***********************************************************************/
 278 void ClusterClassic::clusterBins(){
 279         try {
 280         //      cout << smallCol << '\t' << smallRow << '\t' << smallDist << '\t' << rabund->get(smallRow) << '\t' << rabund->get(smallCol);
 281
 282                 rabund->set(smallRow, rabund->get(smallRow)+rabund->get(smallCol));
 283                 rabund->set(smallCol, 0);
 284                 /*for (int i = smallCol+1; i < rabund->size(); i++) {
 285                         rabund->set((i-1), rabund->get(i));
 286                 }
 287                 rabund->resize((rabund->size()-1));*/
 288                 rabund->setLabel(toString(smallDist));
 289
 290         //      cout << '\t' << rabund->get(smallRow) << '\t' << rabund->get(smallCol) << endl;
 291         }
 292         catch(exception& e) {
 293                 m->errorOut(e, "ClusterClassic", "clusterBins");
 294                 exit(1);
 295         }
 296 }
 297 /***********************************************************************/
 298 void ClusterClassic::clusterNames(){
 299         try {
 300         //      cout << smallCol << '\t' << smallRow << '\t' << smallDist << '\t' << list->get(smallRow) << '\t' << list->get(smallCol);
 301                 if (mapWanted) {  updateMap();  }
 302
 303                 list->set(smallRow, list->get(smallRow)+','+list->get(smallCol));
 304                 list->set(smallCol, "");
 305                 /*for (int i = smallCol+1; i < list->size(); i++) {
 306                         list->set((i-1), list->get(i));
 307                 }
 308                 list->resize((list->size()-1));*/
 309                 list->setLabel(toString(smallDist));
 310
 311         //      cout << '\t' << list->get(smallRow) << '\t' << list->get(smallCol) << endl;
 312     }
 313         catch(exception& e) {
 314                 m->errorOut(e, "ClusterClassic", "clusterNames");
 315                 exit(1);
 316         }
 317 }
 318 /***********************************************************************/
 319 void ClusterClassic::update(double& cutOFF){
 320         try {
 321 //print();
 322                 getSmallCell();
 323
 324                 int r, c;
 325                 r = smallRow; c = smallCol;
 326
 327                 for(int i=0;i<nseqs;i++){
 328                         if(i != r && i != c){
 329                                 double distRow, distCol, newDist;
 330                                 if (i > r) { distRow = dMatrix[i][r]; }
 331                                 else { distRow =  dMatrix[r][i]; }
 332
 333                                 if (i > c) { distCol = dMatrix[i][c]; dMatrix[i][c] = aboveCutoff; } //like removeCell
 334                                 else { distCol =  dMatrix[c][i]; dMatrix[c][i] = aboveCutoff; }
 335
 336                                 if(method == "furthest"){
 337                                         newDist = max(distRow, distCol);
 338                                 }
 339                                 else if (method == "average"){
 340                                         int rowBin = rabund->get(r);
 341                                         int colBin = rabund->get(c);
 342                                         newDist = (colBin * distCol + rowBin * distRow) / (rowBin + colBin);
 343                                 }
 344                                 else if (method == "weighted"){
 345                                         newDist = (distCol + distRow) / 2.0;
 346                                 }
 347                                 else if (method == "nearest"){
 348                                         newDist = min(distRow, distCol);
 349                                 }
 350                                 //cout << "newDist = " << newDist << endl;
 351                                 if (i > r) { dMatrix[i][r] = newDist; }
 352                                 else { dMatrix[r][i] = newDist; }
 353
 354                         }
 355                 }
 356
 357                 clusterBins();
 358                 clusterNames();
 359
 360                 //resize each row
 361                 /*for(int i=0;i<nseqs;i++){
 362                         for(int j=c+1;j<dMatrix[i].size();j++){
 363                                 dMatrix[i][j-1]=dMatrix[i][j];
 364                         }
 365                 }
 366
 367                 //resize each col
 368                 for(int i=c+1;i<nseqs;i++){
 369                         for(int j=0;j < dMatrix[i-1].size();j++){
 370                                 dMatrix[i-1][j]=dMatrix[i][j];
 371                         }
 372                 }
 373
 374                 nseqs--;
 375                 dMatrix.pop_back();*/
 376
 377         }
 378         catch(exception& e) {
 379                 m->errorOut(e, "ClusterClassic", "update");
 380                 exit(1);
 381         }
 382 }
 383 /***********************************************************************/
 384 void ClusterClassic::setMapWanted(bool f)  {
 385         try {
 386                 mapWanted = f;
 387
 388                 //initialize map
 389                 for (int i = 0; i < list->getNumBins(); i++) {
 390
 391                         //parse bin
 392                         string names = list->get(i);
 393                         while (names.find_first_of(',') != -1) {
 394                                 //get name from bin
 395                                 string name = names.substr(0,names.find_first_of(','));
 396                                 //save name and bin number
 397                                 seq2Bin[name] = i;
 398                                 names = names.substr(names.find_first_of(',')+1, names.length());
 399                         }
 400
 401                         //get last name
 402                         seq2Bin[names] = i;
 403                 }
 404
 405         }
 406         catch(exception& e) {
 407                 m->errorOut(e, "ClusterClassic", "setMapWanted");
 408                 exit(1);
 409         }
 410 }
 411 /***********************************************************************/
 412 void ClusterClassic::updateMap() {
 413 try {
 414                 //update location of seqs in smallRow since they move to smallCol now
 415                 string names = list->get(smallRow);
 416                 while (names.find_first_of(',') != -1) {
 417                         //get name from bin
 418                         string name = names.substr(0,names.find_first_of(','));
 419                         //save name and bin number
 420                         seq2Bin[name] = smallCol;
 421                         names = names.substr(names.find_first_of(',')+1, names.length());
 422                 }
 423
 424                 //get last name
 425                 seq2Bin[names] = smallCol;
 426
 427         }
 428         catch(exception& e) {
 429                 m->errorOut(e, "ClusterClassic", "updateMap");
 430                 exit(1);
 431         }
 432 }
 433 /***********************************************************************/
 434 void ClusterClassic::print() {
 435 try {
 436                 //update location of seqs in smallRow since they move to smallCol now
 437                 for (int i = 0; i < dMatrix.size(); i++) {
 438                         cout << "row = " << i << '\t';
 439                         for (int j = 0; j < dMatrix[i].size(); j++) {
 440                                 cout << dMatrix[i][j] << '\t';
 441                         }
 442                         cout << endl;
 443                 }
 444         }
 445         catch(exception& e) {
 446                 m->errorOut(e, "ClusterClassic", "updateMap");
 447                 exit(1);
 448         }
 449 }
 450 /***********************************************************************/
 451