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 "sharedjabund.h"
  12 #include "sharedsorabund.h"
  13 #include "sharedjclass.h"
  14 #include "sharedsorclass.h"
  15 #include "sharedjest.h"
  16 #include "sharedsorest.h"
  17 #include "sharedthetayc.h"
  18 #include "sharedthetan.h"
  19 #include "sharedmorisitahorn.h"
  20 #include "sharedbraycurtis.h"
  21
  22 //**********************************************************************************************************************
  23 vector<string> MatrixOutputCommand::getValidParameters(){
  24         try {
  25                 string Array[] =  {"label","calc","groups","outputdir","inputdir", "output"};
  26                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "MatrixOutputCommand", "getValidParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 MatrixOutputCommand::MatrixOutputCommand(){
  36         try {
  37                 //initialize outputTypes
  38                 vector<string> tempOutNames;
  39                 outputTypes["phylip"] = tempOutNames;
  40         }
  41         catch(exception& e) {
  42                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
  43                 exit(1);
  44         }
  45 }
  46 //**********************************************************************************************************************
  47 vector<string> MatrixOutputCommand::getRequiredParameters(){
  48         try {
  49                 vector<string> myArray;
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "MatrixOutputCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> MatrixOutputCommand::getRequiredFiles(){
  59         try {
  60                 string Array[] =  {"shared"};
  61                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  62                 return myArray;
  63         }
  64         catch(exception& e) {
  65                 m->errorOut(e, "MatrixOutputCommand", "getRequiredFiles");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70
  71 MatrixOutputCommand::MatrixOutputCommand(string option)  {
  72         try {
  73                 globaldata = GlobalData::getInstance();
  74                 abort = false;
  75                 allLines = 1;
  76                 labels.clear();
  77                 Groups.clear();
  78                 Estimators.clear();
  79
  80                 //allow user to run help
  81                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  82
  83                 else {
  84                         //valid paramters for this command
  85                         string Array[] =  {"label","calc","groups","outputdir","inputdir", "output"};
  86                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  87
  88                         OptionParser parser(option);
  89                         map<string,string> parameters  = parser.getParameters();
  90
  91                         ValidParameters validParameter;
  92
  93                         //check to make sure all parameters are valid for command
  94                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  95                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  96                         }
  97
  98                         //initialize outputTypes
  99                         vector<string> tempOutNames;
 100                         outputTypes["phylip"] = tempOutNames;
 101
 102                         //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 104                                 outputDir = "";
 105                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 106                         }
 107
 108                         //make sure the user has already run the read.otu command
 109                         if (globaldata->getSharedFile() == "") {
 110                                 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; }
 111                                 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; }
 112                         }
 113
 114                         //check for optional parameter and set defaults
 115                         // ...at some point should added some additional type checking...
 116                         label = validParameter.validFile(parameters, "label", false);
 117                         if (label == "not found") { label = ""; }
 118                         else {
 119                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 120                                 else { allLines = 1;  }
 121                         }
 122
 123                         output = validParameter.validFile(parameters, "output", false);         if(output == "not found"){      output = "lt"; }
 124                         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"; }
 125
 126                         //if the user has not specified any labels use the ones from read.otu
 127                         if (label == "") {
 128                                 allLines = globaldata->allLines;
 129                                 labels = globaldata->labels;
 130                         }
 131
 132                         groups = validParameter.validFile(parameters, "groups", false);
 133                         if (groups == "not found") { groups = ""; }
 134                         else {
 135                                 m->splitAtDash(groups, Groups);
 136                                 globaldata->Groups = Groups;
 137                         }
 138
 139                         calc = validParameter.validFile(parameters, "calc", false);
 140                         if (calc == "not found") { calc = "jclass-thetayc";  }
 141                         else {
 142                                  if (calc == "default")  {  calc = "jclass-thetayc";  }
 143                         }
 144                         m->splitAtDash(calc, Estimators);
 145
 146                         if (abort == false) {
 147
 148                                 validCalculator = new ValidCalculators();
 149
 150                                 int i;
 151                                 for (i=0; i<Estimators.size(); i++) {
 152                                         if (validCalculator->isValidCalculator("matrix", Estimators[i]) == true) {
 153                                                 if (Estimators[i] == "jabund") {
 154                                                         matrixCalculators.push_back(new JAbund());
 155                                                 }else if (Estimators[i] == "sorabund") {
 156                                                         matrixCalculators.push_back(new SorAbund());
 157                                                 }else if (Estimators[i] == "jclass") {
 158                                                         matrixCalculators.push_back(new Jclass());
 159                                                 }else if (Estimators[i] == "sorclass") {
 160                                                         matrixCalculators.push_back(new SorClass());
 161                                                 }else if (Estimators[i] == "jest") {
 162                                                         matrixCalculators.push_back(new Jest());
 163                                                 }else if (Estimators[i] == "sorest") {
 164                                                         matrixCalculators.push_back(new SorEst());
 165                                                 }else if (Estimators[i] == "thetayc") {
 166                                                         matrixCalculators.push_back(new ThetaYC());
 167                                                 }else if (Estimators[i] == "thetan") {
 168                                                         matrixCalculators.push_back(new ThetaN());
 169                                                 }else if (Estimators[i] == "morisitahorn") {
 170                                                         matrixCalculators.push_back(new MorHorn());
 171                                                 }else if (Estimators[i] == "braycurtis") {
 172                                                         matrixCalculators.push_back(new BrayCurtis());
 173                                                 }
 174                                         }
 175                                 }
 176
 177                         }
 178                 }
 179
 180         }
 181         catch(exception& e) {
 182                 m->errorOut(e, "MatrixOutputCommand", "MatrixOutputCommand");
 183                 exit(1);
 184         }
 185 }
 186
 187 //**********************************************************************************************************************
 188
 189 void MatrixOutputCommand::help(){
 190         try {
 191                 m->mothurOut("The dist.shared command can only be executed after a successful read.otu command.\n");
 192                 m->mothurOut("The dist.shared command parameters are groups, calc, output and label.  No parameters are required.\n");
 193                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like included used.\n");
 194                 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");
 195                 m->mothurOut("The dist.shared command should be in the following format: dist.shared(groups=yourGroups, calc=yourCalcs, label=yourLabels).\n");
 196                 m->mothurOut("The output parameter allows you to specify format of your distance matrix. Options are lt, and square. The default is lt.\n");
 197                 m->mothurOut("Example dist.shared(groups=A-B-C, calc=jabund-sorabund).\n");
 198                 m->mothurOut("The default value for groups is all the groups in your groupfile.\n");
 199                 m->mothurOut("The default value for calc is jclass and thetayc.\n");
 200                 validCalculator->printCalc("matrix", cout);
 201                 m->mothurOut("The dist.shared command outputs a .dist file for each calculator you specify at each distance you choose.\n");
 202                 m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
 203         }
 204         catch(exception& e) {
 205                 m->errorOut(e, "MatrixOutputCommand", "help");
 206                 exit(1);
 207         }
 208 }
 209
 210
 211 //**********************************************************************************************************************
 212
 213 MatrixOutputCommand::~MatrixOutputCommand(){
 214         if (abort == false) {
 215                 delete input; globaldata->ginput = NULL;
 216                 delete read;
 217                 delete validCalculator;
 218         }
 219 }
 220
 221 //**********************************************************************************************************************
 222
 223 int MatrixOutputCommand::execute(){
 224         try {
 225
 226                 if (abort == true) {    return 0;       }
 227
 228                 //if the users entered no valid calculators don't execute command
 229                 if (matrixCalculators.size() == 0) { m->mothurOut("No valid calculators."); m->mothurOutEndLine();  return 0; }
 230
 231                 //you have groups
 232                 read = new ReadOTUFile(globaldata->inputFileName);
 233                 read->read(&*globaldata);
 234
 235                 input = globaldata->ginput;
 236                 lookup = input->getSharedRAbundVectors();
 237                 string lastLabel = lookup[0]->getLabel();
 238
 239                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 240                 set<string> processedLabels;
 241                 set<string> userLabels = labels;
 242
 243                 if (lookup.size() < 2) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); return 0;}
 244
 245                 numGroups = lookup.size();
 246
 247                 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;  }
 248
 249                 //as long as you are not at the end of the file or done wih the lines you want
 250                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 251
 252                         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;  }
 253
 254                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 255                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 256                                 process(lookup);
 257
 258                                 processedLabels.insert(lookup[0]->getLabel());
 259                                 userLabels.erase(lookup[0]->getLabel());
 260                         }
 261
 262                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 263                                 string saveLabel = lookup[0]->getLabel();
 264
 265                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 266                                 lookup = input->getSharedRAbundVectors(lastLabel);
 267
 268                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 269                                 process(lookup);
 270
 271                                 processedLabels.insert(lookup[0]->getLabel());
 272                                 userLabels.erase(lookup[0]->getLabel());
 273
 274                                 //restore real lastlabel to save below
 275                                 lookup[0]->setLabel(saveLabel);
 276                         }
 277
 278                         lastLabel = lookup[0]->getLabel();
 279
 280                         //get next line to process
 281                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 282                         lookup = input->getSharedRAbundVectors();
 283                 }
 284
 285                 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;  }
 286
 287                 //output error messages about any remaining user labels
 288                 set<string>::iterator it;
 289                 bool needToRun = false;
 290                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 291                         m->mothurOut("Your file does not include the label " + *it);
 292                         if (processedLabels.count(lastLabel) != 1) {
 293                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 294                                 needToRun = true;
 295                         }else {
 296                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 297                         }
 298                 }
 299
 300                 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;  }
 301
 302                 //run last label if you need to
 303                 if (needToRun == true)  {
 304                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {  delete lookup[i]; }  }
 305                         lookup = input->getSharedRAbundVectors(lastLabel);
 306
 307                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 308                         process(lookup);
 309                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 310                 }
 311
 312                 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;  }
 313
 314                 //reset groups parameter
 315                 globaldata->Groups.clear();
 316
 317                 m->mothurOutEndLine();
 318                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 319                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 320                 m->mothurOutEndLine();
 321
 322
 323                 return 0;
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "MatrixOutputCommand", "execute");
 327                 exit(1);
 328         }
 329 }
 330 /***********************************************************/
 331 void MatrixOutputCommand::printSims(ostream& out) {
 332         try {
 333
 334                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 335
 336                 //output num seqs
 337                 out << simMatrix.size() << endl;
 338
 339                 if (output == "lt") {
 340                         for (int m = 0; m < simMatrix.size(); m++)      {
 341                                 out << lookup[m]->getGroup() << '\t';
 342                                 for (int n = 0; n < m; n++)     {
 343                                         out << simMatrix[m][n] << '\t';
 344                                 }
 345                                 out << endl;
 346                         }
 347                 }else{
 348                         for (int m = 0; m < simMatrix.size(); m++)      {
 349                                 out << lookup[m]->getGroup() << '\t';
 350                                 for (int n = 0; n < simMatrix[m].size(); n++)   {
 351                                         out << simMatrix[m][n] << '\t';
 352                                 }
 353                                 out << endl;
 354                         }
 355                 }
 356         }
 357         catch(exception& e) {
 358                 m->errorOut(e, "MatrixOutputCommand", "printSims");
 359                 exit(1);
 360         }
 361 }
 362 /***********************************************************/
 363 int MatrixOutputCommand::process(vector<SharedRAbundVector*> thisLookup){
 364         try {
 365
 366                                 EstOutput data;
 367                                 vector<SharedRAbundVector*> subset;
 368
 369                                 //for each calculator
 370                                 for(int i = 0 ; i < matrixCalculators.size(); i++) {
 371
 372                                         //initialize simMatrix
 373                                         simMatrix.clear();
 374                                         simMatrix.resize(numGroups);
 375                                         for (int p = 0; p < simMatrix.size(); p++)      {
 376                                                 for (int j = 0; j < simMatrix.size(); j++)      {
 377                                                         simMatrix[p].push_back(0.0);
 378                                                 }
 379                                         }
 380
 381                                         for (int k = 0; k < thisLookup.size(); k++) {
 382                                                 for (int l = k; l < thisLookup.size(); l++) {
 383                                                         if (k != l) { //we dont need to similiarity of a groups to itself
 384                                                                 //get estimated similarity between 2 groups
 385
 386                                                                 if (m->control_pressed) { return 0; }
 387
 388                                                                 subset.clear(); //clear out old pair of sharedrabunds
 389                                                                 //add new pair of sharedrabunds
 390                                                                 subset.push_back(thisLookup[k]); subset.push_back(thisLookup[l]);
 391
 392                                                                 data = matrixCalculators[i]->getValues(subset); //saves the calculator outputs
 393                                                                 //save values in similarity matrix
 394                                                                 simMatrix[k][l] = 1.0 - data[0];  //convert similiarity to distance
 395                                                                 simMatrix[l][k] = 1.0 - data[0];  //convert similiarity to distance
 396                                                         }
 397                                                 }
 398                                         }
 399
 400                                         exportFileName = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName)) + matrixCalculators[i]->getName() + "." + thisLookup[0]->getLabel() + "." + output + ".dist";
 401                                         m->openOutputFile(exportFileName, out);
 402                                         outputNames.push_back(exportFileName); outputTypes["phylip"].push_back(exportFileName);
 403
 404                                         printSims(out);
 405                                         out.close();
 406
 407                                 }
 408
 409                                 return 0;
 410
 411         }
 412         catch(exception& e) {
 413                 m->errorOut(e, "MatrixOutputCommand", "process");
 414                 exit(1);
 415         }
 416 }
 417 /***********************************************************/
 418
 419
 420