matrixoutputcommand.cpp

   1 /*
   2  *  matrixoutputcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/20/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "matrixoutputcommand.h"
  11 #include "sharedsobscollectsummary.h"
  12 #include "sharedchao1.h"
  13 #include "sharedace.h"
  14 #include "sharednseqs.h"
  15 #include "sharedjabund.h"
  16 #include "sharedsorabund.h"
  17 #include "sharedjclass.h"
  18 #include "sharedsorclass.h"
  19 #include "sharedjest.h"
  20 #include "sharedsorest.h"
  21 #include "sharedthetayc.h"
  22 #include "sharedthetan.h"
  23 #include "sharedkstest.h"
  24 #include "whittaker.h"
  25 #include "sharedochiai.h"
  26 #include "sharedanderbergs.h"
  27 #include "sharedkulczynski.h"
  28 #include "sharedkulczynskicody.h"
  29 #include "sharedlennon.h"
  30 #include "sharedmorisitahorn.h"
  31 #include "sharedbraycurtis.h"
  32 #include "sharedjackknife.h"
  33 #include "whittaker.h"
  34 #include "odum.h"
  35 #include "canberra.h"
  36 #include "structeuclidean.h"
  37 #include "structchord.h"
  38 #include "hellinger.h"
  39 #include "manhattan.h"
  40 #include "structpearson.h"
  41 #include "soergel.h"
  42 #include "spearman.h"
  43 #include "structkulczynski.h"
  44 #include "structchi2.h"
  45 #include "speciesprofile.h"
  46 #include "hamming.h"
  47 #include "gower.h"
  48 #include "memchi2.h"
  49 #include "memchord.h"
  50 #include "memeuclidean.h"
  51 #include "mempearson.h"
  52 //**********************************************************************************************************************
  53 vector<string> MatrixOutputCommand::getValidParameters(){
  54         try {
  55                 string Array[] =  {"label","calc","groups","outputdir","inputdir", "output"};
  56                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "MatrixOutputCommand", "getValidParameters");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 MatrixOutputCommand::MatrixOutputCommand(){
  66         try {
  67                 abort = true;
  68                 //initialize outputTypes
  69                 vector<string> tempOutNames;
  70                 outputTypes["phylip"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 vector<string> MatrixOutputCommand::getRequiredParameters(){
  79         try {
  80                 vector<string> myArray;
  81                 return myArray;
  82         }
  83         catch(exception& e) {
  84                 m->errorOut(e, "MatrixOutputCommand", "getRequiredParameters");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 vector<string> MatrixOutputCommand::getRequiredFiles(){
  90         try {
  91                 string Array[] =  {"shared"};
  92                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  93                 return myArray;
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "MatrixOutputCommand", "getRequiredFiles");
  97                 exit(1);
  98         }
  99 }
 100 //**********************************************************************************************************************
 101
 102 MatrixOutputCommand::MatrixOutputCommand(string option)  {
 103         try {
 104                 globaldata = GlobalData::getInstance();
 105                 abort = false;
 106                 allLines = 1;
 107                 labels.clear();
 108                 Groups.clear();
 109                 Estimators.clear();
 110
 111                 //allow user to run help
 112                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
 113
 114                 else {
 115                         //valid paramters for this command
 116                         string Array[] =  {"label","calc","groups","outputdir","inputdir", "output"};
 117                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
 118
 119                         OptionParser parser(option);
 120                         map<string,string> parameters  = parser.getParameters();
 121
 122                         ValidParameters validParameter;
 123
 124                         //check to make sure all parameters are valid for command
 125                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 126                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 127                         }
 128
 129                         //initialize outputTypes
 130                         vector<string> tempOutNames;
 131                         outputTypes["phylip"] = tempOutNames;
 132
 133                         //if the user changes the output directory command factory will send this info to us in the output parameter
 134                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 135                                 outputDir = "";
 136                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 137                         }
 138
 139                         //make sure the user has already run the read.otu command
 140                         if (globaldata->getSharedFile() == "") {
 141                                 if (globaldata->getListFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the dist.shared command."); m->mothurOutEndLine(); abort = true; }
 142                                 else if (globaldata->getGroupFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the dist.shared command."); m->mothurOutEndLine(); abort = true; }
 143                         }
 144
 145                         //check for optional parameter and set defaults
 146                         // ...at some point should added some additional type checking...
 147                         label = validParameter.validFile(parameters, "label", false);
 148                         if (label == "not found") { label = ""; }
 149                         else {
 150                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 151                                 else { allLines = 1;  }
 152                         }
 153
 154                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
 155                         if ((output != "lt") && (output != "square")) { m->mothurOut(output + " is not a valid output form. Options are lt and square. I will use lt."); m->mothurOutEndLine(); output = "lt"; }
 156
 157                         //if the user has not specified any labels use the ones from read.otu
 158                         if (label == "") {
 159                                 allLines = globaldata->allLines;
 160                                 labels = globaldata->labels;
 161                         }
 162
 163                         groups = validParameter.validFile(parameters, "groups", false);
 164                         if (groups == "not found") { groups = ""; }
 165                         else {
 166                                 m->splitAtDash(groups, Groups);
 167                                 globaldata->Groups = Groups;
 168                         }
 169
 170                         calc = validParameter.validFile(parameters, "calc", false);
 171                         if (calc == "not found") { calc = "jclass-thetayc";  }
 172                         else {
 173                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 174                         }
 175                         m->splitAtDash(calc, Estimators);
 176
 177                         if (abort == false) {
 178
 179                                 validCalculator = new ValidCalculators();
 180
 181                                 int i;
 182                                 for (i=0; i<Estimators.size(); i++) {
 183                                         if (validCalculator->isValidCalculator("matrix", Estimators[i]) == true) {
 184                                                 if (Estimators[i] == "sharedsobs") {
 185                                                         matrixCalculators.push_back(new SharedSobsCS());
 186                                                 }else if (Estimators[i] == "sharedchao") {
 187                                                         matrixCalculators.push_back(new SharedChao1());
 188                                                 }else if (Estimators[i] == "sharedace") {
 189                                                         matrixCalculators.push_back(new SharedAce());
 190                                                 }else if (Estimators[i] == "jabund") {
 191                                                         matrixCalculators.push_back(new JAbund());
 192                                                 }else if (Estimators[i] == "sorabund") {
 193                                                         matrixCalculators.push_back(new SorAbund());
 194                                                 }else if (Estimators[i] == "jclass") {
 195                                                         matrixCalculators.push_back(new Jclass());
 196                                                 }else if (Estimators[i] == "sorclass") {
 197                                                         matrixCalculators.push_back(new SorClass());
 198                                                 }else if (Estimators[i] == "jest") {
 199                                                         matrixCalculators.push_back(new Jest());
 200                                                 }else if (Estimators[i] == "sorest") {
 201                                                         matrixCalculators.push_back(new SorEst());
 202                                                 }else if (Estimators[i] == "thetayc") {
 203                                                         matrixCalculators.push_back(new ThetaYC());
 204                                                 }else if (Estimators[i] == "thetan") {
 205                                                         matrixCalculators.push_back(new ThetaN());
 206                                                 }else if (Estimators[i] == "kstest") {
 207                                                         matrixCalculators.push_back(new KSTest());
 208                                                 }else if (Estimators[i] == "sharednseqs") {
 209                                                         matrixCalculators.push_back(new SharedNSeqs());
 210                                                 }else if (Estimators[i] == "ochiai") {
 211                                                         matrixCalculators.push_back(new Ochiai());
 212                                                 }else if (Estimators[i] == "anderberg") {
 213                                                         matrixCalculators.push_back(new Anderberg());
 214                                                 }else if (Estimators[i] == "kulczynski") {
 215                                                         matrixCalculators.push_back(new Kulczynski());
 216                                                 }else if (Estimators[i] == "kulczynskicody") {
 217                                                         matrixCalculators.push_back(new KulczynskiCody());
 218                                                 }else if (Estimators[i] == "lennon") {
 219                                                         matrixCalculators.push_back(new Lennon());
 220                                                 }else if (Estimators[i] == "morisitahorn") {
 221                                                         matrixCalculators.push_back(new MorHorn());
 222                                                 }else if (Estimators[i] == "braycurtis") {
 223                                                         matrixCalculators.push_back(new BrayCurtis());
 224                                                 }else if (Estimators[i] == "whittaker") {
 225                                                         matrixCalculators.push_back(new Whittaker());
 226                                                 }else if (Estimators[i] == "odum") {
 227                                                         matrixCalculators.push_back(new Odum());
 228                                                 }else if (Estimators[i] == "canberra") {
 229                                                         matrixCalculators.push_back(new Canberra());
 230                                                 }else if (Estimators[i] == "structeuclidean") {
 231                                                         matrixCalculators.push_back(new StructEuclidean());
 232                                                 }else if (Estimators[i] == "structchord") {
 233                                                         matrixCalculators.push_back(new StructChord());
 234                                                 }else if (Estimators[i] == "hellinger") {
 235                                                         matrixCalculators.push_back(new Hellinger());
 236                                                 }else if (Estimators[i] == "manhattan") {
 237                                                         matrixCalculators.push_back(new Manhattan());
 238                                                 }else if (Estimators[i] == "structpearson") {
 239                                                         matrixCalculators.push_back(new StructPearson());
 240                                                 }else if (Estimators[i] == "soergel") {
 241                                                         matrixCalculators.push_back(new Soergel());
 242                                                 }else if (Estimators[i] == "spearman") {
 243                                                         matrixCalculators.push_back(new Spearman());
 244                                                 }else if (Estimators[i] == "structkulczynski") {
 245                                                         matrixCalculators.push_back(new StructKulczynski());
 246                                                 }else if (Estimators[i] == "speciesprofile") {
 247                                                         matrixCalculators.push_back(new SpeciesProfile());
 248                                                 }else if (Estimators[i] == "hamming") {
 249                                                         matrixCalculators.push_back(new Hamming());
 250                                                 }else if (Estimators[i] == "structchi2") {
 251                                                         matrixCalculators.push_back(new StructChi2());
 252                                                 }else if (Estimators[i] == "gower") {
 253                                                         matrixCalculators.push_back(new Gower());
 254                                                 }else if (Estimators[i] == "memchi2") {
 255                                                         matrixCalculators.push_back(new MemChi2());
 256                                                 }else if (Estimators[i] == "memchord") {
 257                                                         matrixCalculators.push_back(new MemChord());
 258                                                 }else if (Estimators[i] == "memeuclidean") {
 259                                                         matrixCalculators.push_back(new MemEuclidean());
 260                                                 }else if (Estimators[i] == "mempearson") {
 261                                                         matrixCalculators.push_back(new MemPearson());
 262                                                 }
 263                                         }
 264                                 }
 265
 266                         }
 267                 }
 268
 269         }
 270         catch(exception& e) {
 271                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
 272                 exit(1);
 273         }
 274 }
 275
 276 //**********************************************************************************************************************
 277
 278 void MatrixOutputCommand::help(){
 279         try {
 280                 m->mothurOut("The dist.shared command can only be executed after a successful read.otu command.\n");
 281                 m->mothurOut("The dist.shared command parameters are groups, calc, output and label.  No parameters are required.\n");
 282                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n");
 283                 m->mothurOut("The group names are separated by dashes. The label parameter allows you to select what distance levels you would like distance matrices created for, and is also separated by dashes.\n");
 284                 m->mothurOut("The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n");
 285                 m->mothurOut("The output parameter allows you to specify format of your distance matrix. Options are lt, and square. The default is lt.\n");
 286                 m->mothurOut("Example dist.shared(groups=A-B-C, calc=jabund-sorabund).\n");
 287                 m->mothurOut("The default value for groups is all the groups in your groupfile.\n");
 288                 m->mothurOut("The default value for calc is jclass and thetayc.\n");
 289                 validCalculator->printCalc("matrix", cout);
 290                 m->mothurOut("The dist.shared command outputs a .dist file for each calculator you specify at each distance you choose.\n");
 291                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 292         }
 293         catch(exception& e) {
 294                 m->errorOut(e, "MatrixOutputCommand", "help");
 295                 exit(1);
 296         }
 297 }
 298
 299
 300 //**********************************************************************************************************************
 301
 302 MatrixOutputCommand::~MatrixOutputCommand(){
 303         if (abort == false) {
 304                 delete input; globaldata->ginput = NULL;
 305                 delete read;
 306                 delete validCalculator;
 307         }
 308 }
 309
 310 //**********************************************************************************************************************
 311
 312 int MatrixOutputCommand::execute(){
 313         try {
 314
 315                 if (abort == true) {    return 0;       }
 316
 317                 //if the users entered no valid calculators don't execute command
 318                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
 319
 320                 //you have groups
 321                 read = new ReadOTUFile(globaldata->inputFileName);
 322                 read->read(&*globaldata);
 323
 324                 input = globaldata->ginput;
 325                 lookup = input->getSharedRAbundVectors();
 326                 string lastLabel = lookup[0]->getLabel();
 327
 328                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 329                 set<string> processedLabels;
 330                 set<string> userLabels = labels;
 331
 332                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0;}
 333
 334                 numGroups = lookup.size();
 335
 336                 if (m->control_pressed) { delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } globaldata->Groups.clear(); return 0;  }
 337
 338                 //as long as you are not at the end of the file or done wih the lines you want
 339                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 340
 341                         if (m->control_pressed) { outputTypes.clear(); delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } globaldata->Groups.clear(); return 0;  }
 342
 343                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 344                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 345                                 process(lookup);
 346
 347                                 processedLabels.insert(lookup[0]->getLabel());
 348                                 userLabels.erase(lookup[0]->getLabel());
 349                         }
 350
 351                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 352                                 string saveLabel = lookup[0]->getLabel();
 353
 354                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 355                                 lookup = input->getSharedRAbundVectors(lastLabel);
 356
 357                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 358                                 process(lookup);
 359
 360                                 processedLabels.insert(lookup[0]->getLabel());
 361                                 userLabels.erase(lookup[0]->getLabel());
 362
 363                                 //restore real lastlabel to save below
 364                                 lookup[0]->setLabel(saveLabel);
 365                         }
 366
 367                         lastLabel = lookup[0]->getLabel();
 368
 369                         //get next line to process
 370                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 371                         lookup = input->getSharedRAbundVectors();
 372                 }
 373
 374                 if (m->control_pressed) { outputTypes.clear(); delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < outputNames.size(); i++) {     remove(outputNames[i].c_str()); } globaldata->Groups.clear(); return 0;  }
 375
 376                 //output error messages about any remaining user labels
 377                 set<string>::iterator it;
 378                 bool needToRun = false;
 379                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 380                         m->mothurOut("Your file does not include the label " + *it);
 381                         if (processedLabels.count(lastLabel) != 1) {
 382                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 383                                 needToRun = true;
 384                         }else {
 385                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 386                         }
 387                 }
 388
 389                 if (m->control_pressed) { outputTypes.clear(); delete read; delete input; globaldata->ginput = NULL;  for (int i = 0; i < outputNames.size(); i++) {    remove(outputNames[i].c_str()); } globaldata->Groups.clear(); return 0;  }
 390
 391                 //run last label if you need to
 392                 if (needToRun == true)  {
 393                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  }
 394                         lookup = input->getSharedRAbundVectors(lastLabel);
 395
 396                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 397                         process(lookup);
 398                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 399                 }
 400
 401                 if (m->control_pressed) { outputTypes.clear();  delete read; delete input; globaldata->ginput = NULL;  for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str()); } globaldata->Groups.clear(); return 0;  }
 402
 403                 //reset groups parameter
 404                 globaldata->Groups.clear();
 405
 406                 m->mothurOutEndLine();
 407                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 408                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 409                 m->mothurOutEndLine();
 410
 411
 412                 return 0;
 413         }
 414         catch(exception& e) {
 415                 m->errorOut(e, "MatrixOutputCommand", "execute");
 416                 exit(1);
 417         }
 418 }
 419 /***********************************************************/
 420 void MatrixOutputCommand::printSims(ostream& out) {
 421         try {
 422
 423                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 424
 425                 //output num seqs
 426                 out << simMatrix.size() << endl;
 427
 428                 if (output == "lt") {
 429                         for (int m = 0; m < simMatrix.size(); m++)      {
 430                                 out << lookup[m]->getGroup() << '\t';
 431                                 for (int n = 0; n < m; n++)     {
 432                                         out << simMatrix[m][n] << '\t';
 433                                 }
 434                                 out << endl;
 435                         }
 436                 }else{
 437                         for (int m = 0; m < simMatrix.size(); m++)      {
 438                                 out << lookup[m]->getGroup() << '\t';
 439                                 for (int n = 0; n < simMatrix[m].size(); n++)   {
 440                                         out << simMatrix[m][n] << '\t';
 441                                 }
 442                                 out << endl;
 443                         }
 444                 }
 445         }
 446         catch(exception& e) {
 447                 m->errorOut(e, "MatrixOutputCommand", "printSims");
 448                 exit(1);
 449         }
 450 }
 451 /***********************************************************/
 452 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
 453         try {
 454
 455                                 EstOutput data;
 456                                 vector<SharedRAbundVector*> subset;
 457
 458                                 //for each calculator
 459                                 for(int i = 0 ; i < matrixCalculators.size(); i++) {
 460
 461                                         //initialize simMatrix
 462                                         simMatrix.clear();
 463                                         simMatrix.resize(numGroups);
 464                                         for (int p = 0; p < simMatrix.size(); p++)      {
 465                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 466                                                         simMatrix[p].push_back(0.0);
 467                                                 }
 468                                         }
 469
 470                                         for (int k = 0; k < thisLookup.size(); k++) {
 471                                                 for (int l = k; l < thisLookup.size(); l++) {
 472                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 473                                                                 //get estimated similarity between 2 groups
 474
 475                                                                 if (m->control_pressed) { return 0; }
 476
 477                                                                 subset.clear(); //clear out old pair of sharedrabunds
 478                                                                 //add new pair of sharedrabunds
 479                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 480
 481                                                                 //if this calc needs all groups to calculate the pair load all groups
 482                                                                 if (matrixCalculators[i]->getNeedsAll()) {
 483                                                                         //load subset with rest of lookup for those calcs that need everyone to calc for a pair
 484                                                                         for (int w = 0; w < thisLookup.size(); w++) {
 485                                                                                 if ((w != k) && (w != l)) { subset.push_back(thisLookup[w]); }
 486                                                                         }
 487                                                                 }
 488
 489                                                                 data = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
 490                                                                 //save values in similarity matrix
 491                                                                 simMatrix[k][l] = 1.0 - data[0];  //convert similiarity to distance
 492                                                                 simMatrix[l][k] = 1.0 - data[0];  //convert similiarity to distance
 493                                                         }
 494                                                 }
 495                                         }
 496
 497                                         exportFileName = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".dist";
 498                                         m->openOutputFile(exportFileName, out);
 499                                         outputNames.push_back(exportFileName); outputTypes["phylip"].push_back(exportFileName);
 500
 501                                         printSims(out);
 502                                         out.close();
 503
 504                                 }
 505
 506                                 return 0;
 507
 508         }
 509         catch(exception& e) {
 510                 m->errorOut(e, "MatrixOutputCommand", "process");
 511                 exit(1);
 512         }
 513 }
 514 /***********************************************************/
 515
 516
 517