pcacommand.cpp

   1 /*
   2  *  pcacommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 1/7/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "pcacommand.h"
  11 #include "inputdata.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> PCACommand::setParameters(){
  15         try {
  16                 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(pshared);
  17                 CommandParameter prelabund("relabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(prelabund);
  18                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  19                 CommandParameter pmetric("metric", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pmetric);
  20                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "PCACommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string PCACommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The pca command parameters are shared, relabund, label, groups and metric.  shared or relabund is required unless you have a valid current file.";
  38                 helpString += "The label parameter is used to analyze specific labels in your input. Default is the first label in your shared or relabund file. Multiple labels may be separated by dashes.\n";
  39                 helpString += "The groups parameter allows you to specify which groups you would like analyzed. Groupnames are separated by dashes.\n";
  40                 helpString += "The metric parameter allows you to indicate if would like the pearson correlation coefficient calculated. Default=True";
  41                 helpString += "Example pca(groups=yourGroups).\n";
  42                 helpString += "Example pca(groups=A-B-C).\n";
  43                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  44                 return helpString;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "PCACommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 PCACommand::PCACommand(){
  53         try {
  54                 abort = true; calledHelp = true;
  55                 setParameters();
  56                 vector<string> tempOutNames;
  57                 outputTypes["pca"] = tempOutNames;
  58                 outputTypes["loadings"] = tempOutNames;
  59         }
  60         catch(exception& e) {
  61                 m->errorOut(e, "PCACommand", "PCACommand");
  62                 exit(1);
  63         }
  64 }
  65 //**********************************************************************************************************************
  66
  67 PCACommand::PCACommand(string option)  {
  68         try {
  69                 abort = false; calledHelp = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; calledHelp = true; }
  73                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  74
  75                 else {
  76                         vector<string> myArray = setParameters();
  77
  78                         OptionParser parser(option);
  79                         map<string, string> parameters = parser. getParameters();
  80
  81                         ValidParameters validParameter;
  82                         map<string, string>::iterator it;
  83
  84                         //check to make sure all parameters are valid for command
  85                         for (it = parameters.begin(); it != parameters.end(); it++) {
  86                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  87                         }
  88
  89                         //initialize outputTypes
  90                         vector<string> tempOutNames;
  91                         outputTypes["pca"] = tempOutNames;
  92                         outputTypes["loadings"] = tempOutNames;
  93
  94                         //if the user changes the input directory command factory will send this info to us in the output parameter
  95                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  96                         if (inputDir == "not found"){   inputDir = "";          }
  97                         else {
  98                                 string path;
  99                                 it = parameters.find("shared");
 100                                 //user has given a template file
 101                                 if(it != parameters.end()){
 102                                         path = m->hasPath(it->second);
 103                                         //if the user has not given a path then, add inputdir. else leave path alone.
 104                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 105                                 }
 106
 107                                 it = parameters.find("relabund");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["relabund"] = inputDir + it->second;         }
 113                                 }
 114                         }
 115
 116                         //check for required parameters
 117                         sharedfile = validParameter.validFile(parameters, "shared", true);
 118                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 119                         else if (sharedfile == "not found") { sharedfile = ""; }
 120                         else {  mode = "sharedfile"; inputFile = sharedfile; }
 121
 122                         relabundfile = validParameter.validFile(parameters, "relabund", true);
 123                         if (relabundfile == "not open") { relabundfile = ""; abort = true; }
 124                         else if (relabundfile == "not found") { relabundfile = ""; }
 125                         else {  mode = "relabund"; inputFile = relabundfile; }
 126
 127
 128                         if ((sharedfile == "") && (relabundfile == "")) {
 129                                 //is there are current file available for any of these?
 130                                 //give priority to shared, then list, then rabund, then sabund
 131                                 //if there is a current shared file, use it
 132                                 sharedfile = m->getSharedFile();
 133                                 if (sharedfile != "") { inputFile = sharedfile; mode = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 134                                 else {
 135                                         relabundfile = m->getRelAbundFile();
 136                                         if (relabundfile != "") { inputFile = relabundfile; mode = "relabund"; m->mothurOut("Using " + relabundfile + " as input file for the relabund parameter."); m->mothurOutEndLine(); }
 137                                         else {
 138                                                 m->mothurOut("No valid current files. You must provide a relabund or shared file."); m->mothurOutEndLine();
 139                                                 abort = true;
 140                                         }
 141                                 }
 142                         }
 143
 144                         //if the user changes the output directory command factory will send this info to us in the output parameter
 145                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 146                                 outputDir = "";
 147                                 outputDir += m->hasPath(inputFile); //if user entered a file with a path then preserve it
 148                         }
 149
 150                         string temp = validParameter.validFile(parameters, "metric", false);    if (temp == "not found"){       temp = "T";                             }
 151                         metric = m->isTrue(temp);
 152
 153                         label = validParameter.validFile(parameters, "label", false);
 154                         if (label == "not found") { label = ""; if(labels.size() == 0) {  m->mothurOut("You did not provide a label, I will use the first label in your inputfile."); m->mothurOutEndLine(); } }
 155                         else { m->splitAtDash(label, labels); }
 156
 157                         groups = validParameter.validFile(parameters, "groups", false);
 158                         if (groups == "not found") { groups = "";  }
 159                         else { m->splitAtDash(groups, Groups);  }
 160                         m->Groups = Groups;
 161
 162                 }
 163
 164         }
 165         catch(exception& e) {
 166                 m->errorOut(e, "PCACommand", "PCACommand");
 167                 exit(1);
 168         }
 169 }
 170 //**********************************************************************************************************************
 171 int PCACommand::execute(){
 172         try {
 173
 174                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 175
 176                 cout.setf(ios::fixed, ios::floatfield);
 177                 cout.setf(ios::showpoint);
 178                 cerr.setf(ios::fixed, ios::floatfield);
 179                 cerr.setf(ios::showpoint);
 180
 181                 //get first line of shared file
 182                 vector< vector<double> > matrix;
 183                 InputData* input;
 184                 if (mode == "sharedfile")                       {
 185                         input = new InputData(inputFile, "sharedfile");
 186                 }else if (mode == "relabund")   {
 187                         input = new InputData(inputFile, "relabund");
 188                 }else {  m->mothurOut("[ERROR]: filetype not recognized."); m->mothurOutEndLine();  return 0; }
 189
 190                 vector<SharedRAbundFloatVector*> lookupFloat = input->getSharedRAbundFloatVectors();
 191                 string lastLabel = lookupFloat[0]->getLabel();
 192
 193                 set<string> processedLabels;
 194                 set<string> userLabels = labels;
 195
 196                 //if the user gave no labels, then use the first one read
 197                 if (labels.size() == 0) {
 198                         label = lastLabel;
 199
 200                         process(lookupFloat);
 201                 }
 202
 203                 //as long as you are not at the end of the file or done wih the lines you want
 204                 while((lookupFloat[0] != NULL) && (userLabels.size() != 0)) {
 205
 206                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());  } delete input; for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }  lookupFloat.clear(); return 0;  }
 207
 208                         if(labels.count(lookupFloat[0]->getLabel()) == 1){
 209                                 processedLabels.insert(lookupFloat[0]->getLabel());
 210                                 userLabels.erase(lookupFloat[0]->getLabel());
 211
 212                                 process(lookupFloat);
 213                         }
 214
 215                         if ((m->anyLabelsToProcess(lookupFloat[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 216                                 string saveLabel = lookupFloat[0]->getLabel();
 217
 218                                 for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  }  lookupFloat.clear();
 219                                 lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 220
 221                                 process(lookupFloat);
 222
 223                                 processedLabels.insert(lookupFloat[0]->getLabel());
 224                                 userLabels.erase(lookupFloat[0]->getLabel());
 225
 226                                 //restore real lastlabel to save below
 227                                 lookupFloat[0]->setLabel(saveLabel);
 228                         }
 229
 230                         lastLabel = lookupFloat[0]->getLabel();
 231
 232                         //get next line to process
 233                         //prevent memory leak
 234                         for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } lookupFloat.clear();
 235                         lookupFloat = input->getSharedRAbundFloatVectors();
 236                 }
 237
 238
 239                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } delete input; for (int i = 0; i < lookupFloat.size(); i++) {  delete lookupFloat[i];  } lookupFloat.clear(); return 0;  }
 240
 241                 //output error messages about any remaining user labels
 242                 set<string>::iterator it;
 243                 bool needToRun = false;
 244                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 245                         m->mothurOut("Your file does not include the label " + *it);
 246                         if (processedLabels.count(lastLabel) != 1) {
 247                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 248                                 needToRun = true;
 249                         }else {
 250                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 251                         }
 252                 }
 253
 254                 //run last label if you need to
 255                 if (needToRun == true)  {
 256                         for (int i = 0; i < lookupFloat.size(); i++) {  if (lookupFloat[i] != NULL) {   delete lookupFloat[i];  } }  lookupFloat.clear();
 257                         lookupFloat = input->getSharedRAbundFloatVectors(lastLabel);
 258
 259                         process(lookupFloat);
 260
 261                         for (int i = 0; i < lookupFloat.size(); i++) {  if (lookupFloat[i] != NULL) {   delete lookupFloat[i];  } } lookupFloat.clear();
 262                 }
 263
 264                 for (int i = 0; i < lookupFloat.size(); i++) {  if (lookupFloat[i] != NULL) {   delete lookupFloat[i];  } } lookupFloat.clear();
 265                 delete input;
 266
 267                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 268
 269                 m->mothurOutEndLine();
 270                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 271                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 272                 m->mothurOutEndLine();
 273
 274                 return 0;
 275         }
 276         catch(exception& e) {
 277                 m->errorOut(e, "PCACommand", "execute");
 278                 exit(1);
 279         }
 280 }
 281
 282 /**********************************************************************************************************************
 283 vector< vector<double> > PCACommand::createMatrix(vector<SharedRAbundFloatVector*> lookupFloat){
 284         try {
 285                 vector< vector<double> > matrix; matrix.resize(lookupFloat.size());
 286
 287                 //fill matrix with shared files relative abundances
 288                 for (int i = 0; i < lookupFloat.size(); i++) {
 289                         for (int j = 0; j < lookupFloat[i]->getNumBins(); j++) {
 290                                 matrix[i].push_back(lookupFloat[i]->getAbundance(j));
 291                         }
 292                 }
 293
 294                 vector< vector<double> > transposeMatrix; transposeMatrix.resize(matrix[0].size());
 295                 for (int i = 0; i < transposeMatrix.size(); i++) {
 296                         for (int j = 0; j < matrix.size(); j++) {
 297                                 transposeMatrix[i].push_back(matrix[j][i]);
 298                         }
 299                 }
 300
 301                 matrix = linearCalc.matrix_mult(matrix, transposeMatrix);
 302
 303                 return matrix;
 304         }
 305         catch(exception& e) {
 306                 m->errorOut(e, "PCACommand", "createMatrix");
 307                 exit(1);
 308         }
 309 }*/
 310 //**********************************************************************************************************************
 311
 312 int PCACommand::process(vector<SharedRAbundFloatVector*>& lookupFloat){
 313         try {
 314                 m->mothurOut("\nProcessing " + lookupFloat[0]->getLabel()); m->mothurOutEndLine();
 315
 316                 int numOTUs = lookupFloat[0]->getNumBins();
 317                 int numSamples = lookupFloat.size();
 318
 319                 vector< vector<double> > matrix(numSamples);
 320                 vector<double> colMeans(numOTUs);
 321
 322
 323                 //fill matrix with shared relative abundances, re-center
 324                 for (int i = 0; i < lookupFloat.size(); i++) {
 325                         matrix[i].resize(numOTUs, 0);
 326
 327                         for (int j = 0; j < numOTUs; j++) {
 328                                 matrix[i][j] = lookupFloat[i]->getAbundance(j);
 329                                 colMeans[j] += matrix[i][j];
 330                         }
 331                 }
 332
 333
 334                 for(int j=0;j<numOTUs;j++){
 335                         colMeans[j] = colMeans[j] / (double)numSamples;
 336                 }
 337
 338                 vector<vector<double> > centered = matrix;
 339                 for(int i=0;i<numSamples;i++){
 340                         for(int j=0;j<numOTUs;j++){
 341                                 centered[i][j] = centered[i][j] - colMeans[j];
 342                         }
 343                 }
 344
 345
 346                 vector< vector<double> > transpose(numOTUs);
 347                 for (int i = 0; i < numOTUs; i++) {
 348                         transpose[i].resize(numSamples, 0);
 349
 350                         for (int j = 0; j < numSamples; j++) {
 351                                 transpose[i][j] = centered[j][i];
 352                         }
 353                 }
 354
 355                 vector<vector<double> > crossProduct = linearCalc.matrix_mult(transpose, centered);
 356
 357                 vector<double> d;
 358                 vector<double> e;
 359
 360                 linearCalc.tred2(crossProduct, d, e);           if (m->control_pressed) { return 0; }
 361                 linearCalc.qtli(d, e, crossProduct);            if (m->control_pressed) { return 0; }
 362
 363                 vector<vector<double> > X = linearCalc.matrix_mult(centered, crossProduct);
 364
 365                 if (m->control_pressed) { return 0; }
 366
 367                 string fbase = outputDir + m->getRootName(m->getSimpleName(inputFile));
 368                 string outputFileName = fbase + lookupFloat[0]->getLabel();
 369                 output(outputFileName, m->Groups, X, d);
 370
 371                 if (metric) {
 372
 373                         vector<vector<double> > observedEuclideanDistance = linearCalc.getObservedEuclideanDistance(centered);
 374
 375                         for (int i = 1; i < 4; i++) {
 376
 377                                 vector< vector<double> > PCAEuclidDists = linearCalc.calculateEuclidianDistance(X, i); //G is the pca file
 378
 379                                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 380
 381                                 double corr = linearCalc.calcPearson(PCAEuclidDists, observedEuclideanDistance);
 382
 383                                 m->mothurOut("Rsq " + toString(i) + " axis: " + toString(corr * corr)); m->mothurOutEndLine();
 384
 385                                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 386                         }
 387                 }
 388
 389                 return 0;
 390         }
 391         catch(exception& e) {
 392                 m->errorOut(e, "PCACommand", "process");
 393                 exit(1);
 394         }
 395 }
 396 /*********************************************************************************************************************************/
 397
 398 void PCACommand::output(string fnameRoot, vector<string> name_list, vector<vector<double> >& G, vector<double> d) {
 399         try {
 400
 401                 int numEigenValues = d.size();
 402                 double dsum = 0.0000;
 403                 for(int i=0;i<numEigenValues;i++){
 404                         dsum += d[i];
 405                 }
 406
 407                 ofstream pcaData((fnameRoot+".pca.axes").c_str(), ios::trunc);
 408                 pcaData.setf(ios::fixed, ios::floatfield);
 409                 pcaData.setf(ios::showpoint);
 410                 outputNames.push_back(fnameRoot+".pca.axes");
 411                 outputTypes["pca"].push_back(fnameRoot+".pca.axes");
 412
 413                 ofstream pcaLoadings((fnameRoot+".pca.loadings").c_str(), ios::trunc);
 414                 pcaLoadings.setf(ios::fixed, ios::floatfield);
 415                 pcaLoadings.setf(ios::showpoint);
 416                 outputNames.push_back(fnameRoot+".pca.loadings");
 417                 outputTypes["loadings"].push_back(fnameRoot+".pca.loadings");
 418
 419                 pcaLoadings << "axis\tloading\n";
 420                 for(int i=0;i<numEigenValues;i++){
 421                         pcaLoadings << i+1 << '\t' << d[i] * 100.0 / dsum << endl;
 422                 }
 423
 424                 pcaData << "group";
 425                 for(int i=0;i<numEigenValues;i++){
 426                         pcaData << '\t' << "axis" << i+1;
 427                 }
 428                 pcaData << endl;
 429
 430                 for(int i=0;i<numEigenValues;i++){
 431                         pcaData << name_list[i] << '\t';
 432                         for(int j=0;j<numEigenValues;j++){
 433                                 pcaData << G[i][j] << '\t';
 434                         }
 435                         pcaData << endl;
 436                 }
 437         }
 438         catch(exception& e) {
 439                 m->errorOut(e, "PCACommand", "output");
 440                 exit(1);
 441         }
 442 }
 443 /*********************************************************************************************************************************/
 444
 445