nmdscommand.cpp

   1 /*
   2  *  nmdscommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 1/11/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "nmdscommand.h"
  11 #include "readphylipvector.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> NMDSCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"phylip","axes","mindim","maxdim","iters","maxiters","epsilon","outputdir","inputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "NMDSCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 NMDSCommand::NMDSCommand(){
  27         try {
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["nmds"] = tempOutNames;
  32                 outputTypes["stress"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "NMDSCommand", "NMDSCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> NMDSCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"phylip"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "NMDSCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> NMDSCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "NMDSCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63
  64 NMDSCommand::NMDSCommand(string option)  {
  65         try {
  66                 abort = false;
  67
  68                 //allow user to run help
  69                 if(option == "help") { help(); abort = true; }
  70
  71                 else {
  72                         //valid paramters for this command
  73                         string Array[] =  {"phylip","axes","mindim","maxdim","iters","maxiters","epsilon","outputdir", "inputdir"};
  74                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  75
  76                         OptionParser parser(option);
  77                         map<string, string> parameters = parser. getParameters();
  78
  79                         ValidParameters validParameter;
  80                         map<string, string>::iterator it;
  81
  82                         //check to make sure all parameters are valid for command
  83                         for (it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86                         //if the user changes the input directory command factory will send this info to us in the output parameter
  87                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  88                         if (inputDir == "not found"){   inputDir = "";          }
  89                         else {
  90                                 string path;
  91                                 it = parameters.find("phylip");
  92                                 //user has given a template file
  93                                 if(it != parameters.end()){
  94                                         path = m->hasPath(it->second);
  95                                         //if the user has not given a path then, add inputdir. else leave path alone.
  96                                         if (path == "") {       parameters["phylip"] = inputDir + it->second;           }
  97                                 }
  98
  99                                 it = parameters.find("axes");
 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["axes"] = inputDir + it->second;             }
 105                                 }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["nmds"] = tempOutNames;
 111                         outputTypes["stress"] = tempOutNames;
 112
 113                         //required parameters
 114                         phylipfile = validParameter.validFile(parameters, "phylip", true);
 115                         if (phylipfile == "not open") { phylipfile = ""; abort = true; }
 116                         else if (phylipfile == "not found") { phylipfile = ""; m->mothurOut("You must provide a distance file before running the nmds command."); m->mothurOutEndLine(); abort = true; }
 117
 118                         axesfile = validParameter.validFile(parameters, "axes", true);
 119                         if (axesfile == "not open") { axesfile = ""; abort = true; }
 120                         else if (axesfile == "not found") { axesfile = "";  }
 121
 122                         //if the user changes the output directory command factory will send this info to us in the output parameter
 123                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 124                                 outputDir = "";
 125                                 outputDir += m->hasPath(phylipfile); //if user entered a file with a path then preserve it
 126                         }
 127
 128                         string temp = validParameter.validFile(parameters, "mindim", false);    if (temp == "not found") {      temp = "1";     }
 129                         convert(temp, mindim);
 130
 131                         temp = validParameter.validFile(parameters, "maxiters", false); if (temp == "not found") {      temp = "500";   }
 132                         convert(temp, maxIters);
 133
 134                         temp = validParameter.validFile(parameters, "iters", false);    if (temp == "not found") {      temp = "10";    }
 135                         convert(temp, iters);
 136
 137                         temp = validParameter.validFile(parameters, "maxdim", false);   if (temp == "not found") {      temp = "2";     }
 138                         convert(temp, maxdim);
 139
 140                         temp = validParameter.validFile(parameters, "epsilon", false);  if (temp == "not found") {      temp = "0.000000000001";        }
 141                         convert(temp, epsilon);
 142
 143                         if (mindim < 1) { m->mothurOut("mindim must be at least 1."); m->mothurOutEndLine(); abort = true; }
 144                         if (maxdim < mindim) { m->mothurOut("maxdim must be greater than mindim."); m->mothurOutEndLine(); abort = true; }
 145                 }
 146
 147         }
 148         catch(exception& e) {
 149                 m->errorOut(e, "NMDSCommand", "NMDSCommand");
 150                 exit(1);
 151         }
 152 }
 153 //**********************************************************************************************************************
 154 void NMDSCommand::help(){
 155         try {
 156                 m->mothurOut("The nmds command is modelled after the nmds code written in R by Sarah Goslee, using Non-metric multidimensional scaling function using the majorization algorithm from Borg & Groenen 1997, Modern Multidimensional Scaling."); m->mothurOutEndLine();
 157                 m->mothurOut("The nmds command parameters are phylip, axes, mindim, maxdim, maxiters, iters and epsilon."); m->mothurOutEndLine();
 158                 m->mothurOut("The phylip parameter allows you to enter your distance file."); m->mothurOutEndLine();
 159                 m->mothurOut("The axes parameter allows you to enter a file containing a starting configuration."); m->mothurOutEndLine();
 160                 m->mothurOut("The maxdim parameter allows you to select how maximum dimensions to use. Default=2"); m->mothurOutEndLine();
 161                 m->mothurOut("The mindim parameter allows you to select how minimum dimensions to use. Default=1"); m->mothurOutEndLine();
 162                 m->mothurOut("The maxiters parameter allows you to select the maximum number of iters to try with each random configuration. Default=500"); m->mothurOutEndLine();
 163                 m->mothurOut("The iters parameter allows you to select the number of random configuration to try. Default=10"); m->mothurOutEndLine();
 164                 m->mothurOut("The epsilon parameter allows you to select set an acceptable stopping point. Default=1e-12."); m->mothurOutEndLine();
 165                 m->mothurOut("Example nmds(phylip=yourDistanceFile).\n");
 166                 m->mothurOut("Note: No spaces between parameter labels (i.e. phylip), '=' and parameters (i.e.yourDistanceFile).\n\n");
 167         }
 168         catch(exception& e) {
 169                 m->errorOut(e, "NMDSCommand", "help");
 170                 exit(1);
 171         }
 172 }
 173 //**********************************************************************************************************************
 174 NMDSCommand::~NMDSCommand(){}
 175 //**********************************************************************************************************************
 176 int NMDSCommand::execute(){
 177         try {
 178
 179                 if (abort == true) { return 0; }
 180
 181                 cout.setf(ios::fixed, ios::floatfield);
 182                 cout.setf(ios::showpoint);
 183
 184                 vector<string> names;
 185                 vector< vector< double> > matrix;
 186
 187                 //read in phylip file
 188                 ReadPhylipVector readFile(phylipfile);
 189                 names = readFile.read(matrix);
 190                 if (m->control_pressed) { return 0; }
 191
 192                 //read axes
 193                 vector< vector<double> > axes;
 194                 if (axesfile != "") {  axes = readAxes(names);          }
 195
 196                 for (int i = mindim; i <= maxdim; i++) {
 197                         m->mothurOut("Processing Dimension: " + toString(i)); m->mothurOutEndLine();
 198
 199                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "dim" + toString(i) + ".nmds";
 200                         string stressFileName = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "dim" + toString(i) + ".stress.nmds";
 201                         outputNames.push_back(outputFileName); outputTypes["nmds"].push_back(outputFileName);
 202                         outputNames.push_back(stressFileName); outputTypes["stress"].push_back(stressFileName);
 203
 204                         ofstream out, out2;
 205                         m->openOutputFile(outputFileName, out);
 206                         m->openOutputFile(stressFileName, out2);
 207
 208                         out2.setf(ios::fixed, ios::floatfield);
 209                         out2.setf(ios::showpoint);
 210                         out.setf(ios::fixed, ios::floatfield);
 211                         out.setf(ios::showpoint);
 212
 213                         out2 << "Iter\tStress\tCorr" << endl;
 214
 215                         for (int j = 0; j < iters; j++) {
 216                                 m->mothurOut(toString(j+1)); m->mothurOutEndLine();
 217
 218                                 //get configuration - either randomly generate or resize to this dimension
 219                                 vector< vector<double> > thisConfig;
 220                                 if (axesfile == "") {   thisConfig = generateStartingConfiguration(names.size(), i);            }
 221                                 else                            {       thisConfig = getConfiguration(axes, i);                                                         }
 222                                 if (m->control_pressed) { out.close(); out2.close(); for (int k = 0; k < outputNames.size(); k++) {     remove(outputNames[k].c_str()); } return 0; }
 223
 224                                 //calc nmds for this dimension
 225                                 double stress;
 226                                 vector< vector<double> > endConfig = nmdsCalc(matrix, thisConfig, stress);
 227                                 if (m->control_pressed) { out.close(); out2.close(); for (int k = 0; k < outputNames.size(); k++) {     remove(outputNames[k].c_str()); } return 0; }
 228
 229                                 //calc euclid distances for new config
 230                                 vector< vector<double> > newEuclid = linearCalc.calculateEuclidianDistance(endConfig);
 231                                 if (m->control_pressed) { out.close(); out2.close(); for (int k = 0; k < outputNames.size(); k++) {     remove(outputNames[k].c_str()); } return 0; }
 232
 233                                 //calc correlation between original distances and euclidean distances from this config
 234                                 double corr = linearCalc.calcPearson(matrix, newEuclid);
 235                                 corr *= corr;
 236                                 if (m->control_pressed) { out.close(); out2.close(); for (int k = 0; k < outputNames.size(); k++) {     remove(outputNames[k].c_str()); } return 0; }
 237
 238                                 //output results
 239                                 out << "Config" << (j+1) << '\t';
 240                                 for (int k = 0; k < i; k++) { out << "X" << (k+1) << '\t'; }
 241                                 out << endl;
 242                                 out2 << (j+1) << '\t' << stress << '\t' << corr << endl;
 243
 244                                 output(endConfig, names, out);
 245
 246                                 if (m->control_pressed) { out.close(); out2.close(); for (int k = 0; k < outputNames.size(); k++) {     remove(outputNames[k].c_str()); } return 0; }
 247
 248                         }
 249
 250                         out.close(); out2.close();
 251                 }
 252
 253                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 254
 255                 m->mothurOutEndLine();
 256                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 257                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 258                 m->mothurOutEndLine();
 259
 260                 return 0;
 261         }
 262         catch(exception& e) {
 263                 m->errorOut(e, "NMDSCommand", "execute");
 264                 exit(1);
 265         }
 266 }
 267 //**********************************************************************************************************************
 268 vector< vector<double> > NMDSCommand::nmdsCalc(vector< vector<double> >& matrix, vector< vector<double> >& config, double& stress1) {
 269         try {
 270
 271                 vector< vector<double> > newConfig = config;
 272
 273                 //calc euclid distances
 274                 vector< vector<double> > euclid = linearCalc.calculateEuclidianDistance(newConfig);
 275                 if (m->control_pressed) { return newConfig; }
 276
 277                 double stress2 = calculateStress(matrix, euclid);
 278                 stress1 = stress2 + 1.0 + epsilon;
 279
 280                 int count = 0;
 281                 while ((count < maxIters) && (abs(stress1 - stress2) > epsilon)) {
 282                         count++;
 283
 284                         stress1 = stress2;
 285
 286                         if (m->control_pressed) { return newConfig; }
 287
 288                         vector< vector<double> > b; b.resize(euclid.size());
 289                         for (int i = 0; i < b.size(); i++) { b[i].resize(euclid[i].size(), 0.0); }
 290
 291                         vector<double> columnSums; columnSums.resize(euclid.size(), 0.0);
 292                         for (int i = 0; i < euclid.size(); i++) {
 293                                 for (int j = 0; j < euclid[i].size(); j++) {
 294                                         //eliminate divide by zero error
 295                                         if (euclid[i][j] != 0) {
 296                                                 b[i][j] = matrix[i][j] / euclid[i][j];
 297                                                 columnSums[j] += b[i][j];
 298                                                 b[i][j] *= -1.0;
 299                                         }
 300                                 }
 301                         }
 302
 303                         //put in diagonal sums
 304                         for (int i = 0; i < euclid.size(); i++) {  b[i][i] = columnSums[i]; }
 305
 306                         int numInLowerTriangle = matrix.size() * (matrix.size()-1) / 2.0;
 307                         double n = (1.0 + sqrt(1.0 + 8.0 * numInLowerTriangle)) / 2.0;
 308
 309                         //matrix mult
 310                         newConfig = linearCalc.matrix_mult(newConfig, b);
 311                         for (int i = 0; i < newConfig.size(); i++) {
 312                                 for (int j = 0; j < newConfig[i].size(); j++) {
 313                                         newConfig[i][j] *= (1.0 / n);
 314                                 }
 315                         }
 316
 317                         euclid = linearCalc.calculateEuclidianDistance(newConfig);
 318
 319                         stress2 = calculateStress(matrix, euclid);
 320                 }
 321
 322                 return newConfig;
 323         }
 324         catch(exception& e) {
 325                 m->errorOut(e, "NMDSCommand", "generateStartingConfiguration");
 326                 exit(1);
 327         }
 328 }
 329
 330 //**********************************************************************************************************************
 331 //generate random config
 332 vector< vector<double> > NMDSCommand::generateStartingConfiguration(int numNames, int dimension) {
 333         try {
 334                 vector< vector<double> > axes;  axes.resize(dimension);
 335                 for (int i = 0; i < axes.size(); i++) {  axes[i].resize(numNames); }
 336
 337                 //generate random number between -1 and 1, precision 6
 338                 for (int i = 0; i < axes.size(); i++) {
 339                         for (int j = 0; j < axes[i].size(); j++) {
 340
 341                                 if (m->control_pressed) { return axes; }
 342
 343                                 //generate random int between 0 and 99999
 344                                 int myrand = (int)((float)(rand()) / ((RAND_MAX / 99998) + 1));
 345
 346                                 //generate random sign
 347                                 int mysign = (int)((float)(rand()) / ((RAND_MAX / 99998) + 1));
 348
 349                                 //if mysign is even then sign = positive, else sign = negative
 350                                 if ((mysign % 2) == 0) { mysign = 1.0; }
 351                                 else { mysign = -1.0; }
 352
 353                                 axes[i][j] = mysign * myrand / (float) 100000;
 354                         }
 355                 }
 356
 357                 return axes;
 358         }
 359         catch(exception& e) {
 360                 m->errorOut(e, "NMDSCommand", "generateStartingConfiguration");
 361                 exit(1);
 362         }
 363 }
 364 //**********************************************************************************************************************
 365 //normalize configuration
 366 int NMDSCommand::normalizeConfiguration(vector< vector<double> >& axes, int numNames, int dimension) {
 367         try {
 368                 vector<double> averageAxes; averageAxes.resize(dimension, 0.0);
 369
 370                 //find average
 371                 for (int i = 0; i < axes.size(); i++) {
 372                         for (int j = 0; j < axes[i].size(); j++) {      averageAxes[i] += axes[i][j];   }
 373
 374                         averageAxes[i] /= (float) numNames;
 375                 }
 376
 377                 //normalize axes
 378                 double sumDenom = 0.0;
 379                 for (int i = 0; i < axes.size(); i++) {
 380                         for (int j = 0; j < axes[i].size(); j++) {
 381                                 sumDenom += ((axes[i][j] - averageAxes[i]) * (axes[i][j] - averageAxes[i]));
 382                         }
 383                 }
 384
 385                 double denom = sqrt((sumDenom / (float) (axes.size() * numNames)));
 386
 387                 for (int i = 0; i < axes.size(); i++) {
 388                         for (int j = 0; j < axes[i].size(); j++) {
 389                                 axes[i][j] = (axes[i][j] - averageAxes[i]) / denom;
 390                         }
 391                 }
 392
 393                 return 0;
 394         }
 395         catch(exception& e) {
 396                 m->errorOut(e, "NMDSCommand", "normalizeConfiguration");
 397                 exit(1);
 398         }
 399 }
 400 //**********************************************************************************************************************
 401 //get configuration
 402 vector< vector<double> > NMDSCommand::getConfiguration(vector< vector<double> >& axes, int dimension) {
 403         try {
 404                 vector< vector<double> > newAxes; newAxes.resize(dimension);
 405
 406                 for (int i = 0; i < dimension; i++) {
 407                         newAxes[i] = axes[i];
 408                 }
 409
 410                 return newAxes;
 411         }
 412         catch(exception& e) {
 413                 m->errorOut(e, "NMDSCommand", "getConfiguration");
 414                 exit(1);
 415         }
 416 }
 417 //**********************************************************************************************************************
 418 //find raw stress, and normalize using
 419 double NMDSCommand::calculateStress(vector< vector<double> >& matrix, vector< vector<double> >& config) {
 420         try {
 421                 double normStress = 0.0;
 422                 double denom = 0.0;
 423                 double rawStress = 0.0;
 424
 425                 //find raw stress
 426                 for (int i = 0; i < matrix.size(); i++) {
 427                         for (int j = 0; j < matrix[i].size(); j++) {
 428                                 if (m->control_pressed) { return normStress; }
 429
 430                                 rawStress += ((matrix[i][j] - config[i][j]) * (matrix[i][j] - config[i][j]));
 431                                 denom += (config[i][j] * config[i][j]);
 432                         }
 433                 }
 434
 435                 //normalize stress
 436                 if ((rawStress != 0.0) && (denom != 0.0)) {
 437                         normStress = sqrt((rawStress / denom));
 438                 }
 439
 440                 return normStress;
 441         }
 442         catch(exception& e) {
 443                 m->errorOut(e, "NMDSCommand", "calculateStress");
 444                 exit(1);
 445         }
 446 }
 447
 448 //**********************************************************************************************************************
 449 int NMDSCommand::output(vector< vector<double> >& config, vector<string>& names, ofstream& out) {
 450         try {
 451
 452                 for (int i = 0; i < names.size(); i++) {
 453
 454                         out << names[i] << '\t';
 455
 456                         for (int j = 0; j < config.size(); j++) {
 457                                 out << config[j][i] << '\t';
 458                         }
 459
 460                         out << endl;
 461                 }
 462
 463                 out << endl << endl;
 464
 465                 return 0;
 466         }
 467         catch(exception& e) {
 468                 m->errorOut(e, "NMDSCommand", "output");
 469                 exit(1);
 470         }
 471 }
 472 /*****************************************************************/
 473 vector< vector<double> > NMDSCommand::readAxes(vector<string> names){
 474         try {
 475                 ifstream in;
 476                 m->openInputFile(axesfile, in);
 477
 478                 string headerLine = m->getline(in); m->gobble(in);
 479
 480                 //count the number of axis you are reading
 481                 bool done = false;
 482                 int count = 0;
 483                 while (!done) {
 484                         int pos = headerLine.find("axis");
 485                         if (pos != string::npos) {
 486                                 count++;
 487                                 headerLine = headerLine.substr(pos+4);
 488                         }else { done = true; }
 489                 }
 490
 491                 if (maxdim > count) {
 492                         m->mothurOut("You requested maxdim = " + toString(maxdim) + ", but your file only includes " + toString(count) + ". Using " + toString(count) + "."); m->mothurOutEndLine();
 493                         maxdim = count;
 494                         if (maxdim < mindim) { m->mothurOut("Also adjusting mindim to " + toString(maxdim-1) + "."); m->mothurOutEndLine(); }
 495                 }
 496
 497                 vector< vector<double> > axes;  axes.resize(maxdim);
 498                 for (int i = 0; i < axes.size(); i++) { axes[i].resize(names.size(), 0.0); }
 499
 500                 map <string, vector<double> > orderedAxes;
 501                 map     <string, vector<double> >::iterator it;
 502
 503                 while (!in.eof()) {
 504
 505                         if (m->control_pressed) { in.close(); return axes; }
 506
 507                         string group = "";
 508                         in >> group; m->gobble(in);
 509
 510                         bool ignore = false;
 511                         if (!m->inUsersGroups(group, names)) { ignore = true; m->mothurOut(group + " is in your axes file and not in your distance file, ignoring."); m->mothurOutEndLine(); }
 512
 513                         vector<double> thisGroupsAxes;
 514                         for (int i = 0; i < count; i++) {
 515                                 float temp = 0.0;
 516                                 in >> temp;
 517
 518                                 //only save the axis we want
 519                                 if (i < maxdim) {  thisGroupsAxes.push_back(temp); }
 520                         }
 521
 522                         if (!ignore) {  orderedAxes[group] = thisGroupsAxes; }
 523
 524                         m->gobble(in);
 525                 }
 526                 in.close();
 527
 528                 //sanity check
 529                 if (names.size() != orderedAxes.size()) { m->mothurOut("[ERROR]: your axes file does not match your distance file, aborting."); m->mothurOutEndLine(); m->control_pressed = true; return axes; }
 530
 531                 //put axes info in same order as distance file, just in case
 532                 for (int i = 0; i < names.size(); i++) {
 533                         it = orderedAxes.find(names[i]);
 534
 535                         if (it != orderedAxes.end()) {
 536                                 vector<double> thisGroupsAxes = it->second;
 537
 538                                 for (int j = 0; j < thisGroupsAxes.size(); j++) {
 539                                         axes[j][i] = thisGroupsAxes[j];
 540                                 }
 541
 542                         }else { m->mothurOut("[ERROR]: your axes file does not match your distance file, aborting."); m->mothurOutEndLine(); m->control_pressed = true; return axes; }
 543                 }
 544
 545                 return axes;
 546         }
 547         catch(exception& e) {
 548                 m->errorOut(e, "NMDSCommand", "readAxes");
 549                 exit(1);
 550         }
 551 }
 552 /**********************************************************************************************************************/
 553
 554
 555