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