rarefactcommand.cpp

   1 /*
   2  *  rarefactcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "rarefactcommand.h"
  11 #include "ace.h"
  12 #include "sobs.h"
  13 #include "nseqs.h"
  14 #include "chao1.h"
  15 #include "bootstrap.h"
  16 #include "simpson.h"
  17 #include "simpsoneven.h"
  18 #include "heip.h"
  19 #include "smithwilson.h"
  20 #include "invsimpson.h"
  21 #include "npshannon.h"
  22 #include "shannoneven.h"
  23 #include "shannon.h"
  24 #include "jackknife.h"
  25 #include "coverage.h"
  26
  27
  28 //**********************************************************************************************************************
  29 vector<string> RareFactCommand::setParameters(){
  30         try {
  31                 CommandParameter plist("list", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(plist);
  32                 CommandParameter prabund("rabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(prabund);
  33                 CommandParameter psabund("sabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(psabund);
  34                 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(pshared);
  35                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  36                 CommandParameter pfreq("freq", "Number", "", "100", "", "", "",false,false); parameters.push_back(pfreq);
  37                 CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  38                 CommandParameter pcalc("calc", "Multiple", "sobs-chao-nseqs-coverage-ace-jack-shannon-shannoneven-npshannon-heip-smithwilson-simpson-simpsoneven-invsimpson-bootstrap", "sobs", "", "", "",true,false); parameters.push_back(pcalc);
  39                 CommandParameter pabund("abund", "Number", "", "10", "", "", "",false,false); parameters.push_back(pabund);
  40                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  41                 CommandParameter pgroupmode("groupmode", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pgroupmode);
  42                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  43                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  44
  45                 vector<string> myArray;
  46                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  47                 return myArray;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "RareFactCommand", "setParameters");
  51                 exit(1);
  52         }
  53 }
  54 //**********************************************************************************************************************
  55 string RareFactCommand::getHelpString(){
  56         try {
  57                 ValidCalculators validCalculator;
  58                 string helpString = "";
  59                 helpString += "The rarefaction.single command parameters are list, sabund, rabund, shared, label, iters, freq, calc, processors and abund.  list, sabund, rabund or shared is required unless you have a valid current file. \n";
  60                 helpString += "The freq parameter is used indicate when to output your data, by default it is set to 100. But you can set it to a percentage of the number of sequence. For example freq=0.10, means 10%. \n";
  61                 helpString += "The processors parameter allows you to specify the number of processors to use. The default is 1.\n";
  62                 helpString += "The rarefaction.single command should be in the following format: \n";
  63                 helpString += "rarefaction.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n";
  64                 helpString += "Example rarefaction.single(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-rchao-race-rjack-rbootstrap-rshannon-rnpshannon-rsimpson).\n";
  65                 helpString += "The default values for iters is 1000, freq is 100, and calc is rarefaction which calculates the rarefaction curve for the observed richness.\n";
  66                 validCalculator.printCalc("rarefaction");
  67                 helpString += "If you are running rarefaction.single with a shared file and would like your results collated in one file, set groupmode=t. (Default=true).\n";
  68                 helpString += "The label parameter is used to analyze specific labels in your input.\n";
  69                 helpString += "Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n";
  70                 return helpString;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "RareFactCommand", "getHelpString");
  74                 exit(1);
  75         }
  76 }
  77
  78 //**********************************************************************************************************************
  79 RareFactCommand::RareFactCommand(){
  80         try {
  81                 abort = true; calledHelp = true;
  82                 setParameters();
  83                 vector<string> tempOutNames;
  84                 outputTypes["rarefaction"] = tempOutNames;
  85                 outputTypes["r_chao"] = tempOutNames;
  86                 outputTypes["r_ace"] = tempOutNames;
  87                 outputTypes["r_jack"] = tempOutNames;
  88                 outputTypes["r_shannon"] = tempOutNames;
  89                 outputTypes["r_shannoneven"] = tempOutNames;
  90                 outputTypes["r_heip"] = tempOutNames;
  91                 outputTypes["r_smithwilson"] = tempOutNames;
  92                 outputTypes["r_npshannon"] = tempOutNames;
  93                 outputTypes["r_simpson"] = tempOutNames;
  94                 outputTypes["r_simpsoneven"] = tempOutNames;
  95                 outputTypes["r_invsimpson"] = tempOutNames;
  96                 outputTypes["r_bootstrap"] = tempOutNames;
  97                 outputTypes["r_coverage"] = tempOutNames;
  98                 outputTypes["r_nseqs"] = tempOutNames;
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "RareFactCommand", "RareFactCommand");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106 RareFactCommand::RareFactCommand(string option)  {
 107         try {
 108                 abort = false; calledHelp = false;
 109                 allLines = 1;
 110
 111                 //allow user to run help
 112                 if(option == "help") { help(); abort = true; calledHelp = true; }
 113                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 114
 115                 else {
 116                         vector<string> myArray = setParameters();
 117
 118                         OptionParser parser(option);
 119                         map<string,string> parameters = parser.getParameters();
 120                         map<string,string>::iterator it;
 121
 122                         ValidParameters validParameter;
 123
 124                         //check to make sure all parameters are valid for command
 125                         for (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["rarefaction"] = tempOutNames;
 132                         outputTypes["r_chao"] = tempOutNames;
 133                         outputTypes["r_ace"] = tempOutNames;
 134                         outputTypes["r_jack"] = tempOutNames;
 135                         outputTypes["r_shannon"] = tempOutNames;
 136                         outputTypes["r_shannoneven"] = tempOutNames;
 137                         outputTypes["r_heip"] = tempOutNames;
 138                         outputTypes["r_smithwilson"] = tempOutNames;
 139                         outputTypes["r_npshannon"] = tempOutNames;
 140                         outputTypes["r_simpson"] = tempOutNames;
 141                         outputTypes["r_simpsoneven"] = tempOutNames;
 142                         outputTypes["r_invsimpson"] = tempOutNames;
 143                         outputTypes["r_bootstrap"] = tempOutNames;
 144                         outputTypes["r_coverage"] = tempOutNames;
 145                         outputTypes["r_nseqs"] = tempOutNames;
 146
 147                         //if the user changes the input directory command factory will send this info to us in the output parameter
 148                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 149                         if (inputDir == "not found"){   inputDir = "";          }
 150                         else {
 151                                 string path;
 152                                 it = parameters.find("shared");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 158                                 }
 159
 160                                 it = parameters.find("rabund");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 166                                 }
 167
 168                                 it = parameters.find("sabund");
 169                                 //user has given a template file
 170                                 if(it != parameters.end()){
 171                                         path = m->hasPath(it->second);
 172                                         //if the user has not given a path then, add inputdir. else leave path alone.
 173                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 174                                 }
 175
 176                                 it = parameters.find("list");
 177                                 //user has given a template file
 178                                 if(it != parameters.end()){
 179                                         path = m->hasPath(it->second);
 180                                         //if the user has not given a path then, add inputdir. else leave path alone.
 181                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 182                                 }
 183                         }
 184
 185                         //check for required parameters
 186                         listfile = validParameter.validFile(parameters, "list", true);
 187                         if (listfile == "not open") { listfile = ""; abort = true; }
 188                         else if (listfile == "not found") { listfile = ""; }
 189                         else {  format = "list"; inputfile = listfile; m->setListFile(listfile); }
 190
 191                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 192                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 193                         else if (sabundfile == "not found") { sabundfile = ""; }
 194                         else {  format = "sabund"; inputfile = sabundfile; m->setSabundFile(sabundfile); }
 195
 196                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 197                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 198                         else if (rabundfile == "not found") { rabundfile = ""; }
 199                         else {  format = "rabund"; inputfile = rabundfile; m->setRabundFile(rabundfile); }
 200
 201                         sharedfile = validParameter.validFile(parameters, "shared", true);
 202                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 203                         else if (sharedfile == "not found") { sharedfile = ""; }
 204                         else {  format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
 205
 206                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 207                                 //is there are current file available for any of these?
 208                                 //give priority to shared, then list, then rabund, then sabund
 209                                 //if there is a current shared file, use it
 210                                 sharedfile = m->getSharedFile();
 211                                 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 212                                 else {
 213                                         listfile = m->getListFile();
 214                                         if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 215                                         else {
 216                                                 rabundfile = m->getRabundFile();
 217                                                 if (rabundfile != "") { inputfile = rabundfile; format = "rabund"; m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 218                                                 else {
 219                                                         sabundfile = m->getSabundFile();
 220                                                         if (sabundfile != "") { inputfile = sabundfile; format = "sabund"; m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 221                                                         else {
 222                                                                 m->mothurOut("No valid current files. You must provide a list, sabund, rabund or shared file before you can use the collect.single command."); m->mothurOutEndLine();
 223                                                                 abort = true;
 224                                                         }
 225                                                 }
 226                                         }
 227                                 }
 228                         }
 229
 230                         //if the user changes the output directory command factory will send this info to us in the output parameter
 231                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputfile);              }
 232
 233                         //check for optional parameter and set defaults
 234                         // ...at some point should added some additional type checking...
 235                         label = validParameter.validFile(parameters, "label", false);
 236                         if (label == "not found") { label = ""; }
 237                         else {
 238                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 239                                 else { allLines = 1;  }
 240                         }
 241
 242                         calc = validParameter.validFile(parameters, "calc", false);
 243                         if (calc == "not found") { calc = "sobs";  }
 244                         else {
 245                                  if (calc == "default")  {  calc = "sobs";  }
 246                         }
 247                         m->splitAtDash(calc, Estimators);
 248                         if (m->inUsersGroups("citation", Estimators)) {
 249                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 250                                 //remove citation from list of calcs
 251                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 252                         }
 253
 254                         string temp;
 255                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 256                         convert(temp, freq);
 257
 258                         temp = validParameter.validFile(parameters, "abund", false);                    if (temp == "not found") { temp = "10"; }
 259                         convert(temp, abund);
 260
 261                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 262                         convert(temp, nIters);
 263
 264                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 265                         m->setProcessors(temp);
 266                         convert(temp, processors);
 267
 268                         temp = validParameter.validFile(parameters, "groupmode", false);                if (temp == "not found") { temp = "T"; }
 269                         groupMode = m->isTrue(temp);
 270                 }
 271
 272         }
 273         catch(exception& e) {
 274                 m->errorOut(e, "RareFactCommand", "RareFactCommand");
 275                 exit(1);
 276         }
 277 }
 278 //**********************************************************************************************************************
 279
 280 int RareFactCommand::execute(){
 281         try {
 282
 283                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 284
 285                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 286                 else {  inputFileNames = parseSharedFile(sharedfile);  format = "rabund"; }
 287
 288                 if (m->control_pressed) { return 0; }
 289
 290                 for (int p = 0; p < inputFileNames.size(); p++) {
 291
 292                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 293
 294                         if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        }  m->clearGroups();  return 0; }
 295
 296                         if (inputFileNames.size() > 1) {
 297                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 298                         }
 299                         int i;
 300                         ValidCalculators validCalculator;
 301
 302
 303                         for (i=0; i<Estimators.size(); i++) {
 304                                 if (validCalculator.isValidCalculator("rarefaction", Estimators[i]) == true) {
 305                                         if (Estimators[i] == "sobs") {
 306                                                 rDisplays.push_back(new RareDisplay(new Sobs(), new ThreeColumnFile(fileNameRoot+"rarefaction")));
 307                                                 outputNames.push_back(fileNameRoot+"rarefaction"); outputTypes["rarefaction"].push_back(fileNameRoot+"rarefaction");
 308                                         }else if (Estimators[i] == "chao") {
 309                                                 rDisplays.push_back(new RareDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"r_chao")));
 310                                                 outputNames.push_back(fileNameRoot+"r_chao"); outputTypes["r_chao"].push_back(fileNameRoot+"r_chao");
 311                                         }else if (Estimators[i] == "ace") {
 312                                                 if(abund < 5)
 313                                                         abund = 10;
 314                                                 rDisplays.push_back(new RareDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"r_ace")));
 315                                                 outputNames.push_back(fileNameRoot+"r_ace"); outputTypes["r_ace"].push_back(fileNameRoot+"r_ace");
 316                                         }else if (Estimators[i] == "jack") {
 317                                                 rDisplays.push_back(new RareDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"r_jack")));
 318                                                 outputNames.push_back(fileNameRoot+"r_jack"); outputTypes["r_jack"].push_back(fileNameRoot+"r_jack");
 319                                         }else if (Estimators[i] == "shannon") {
 320                                                 rDisplays.push_back(new RareDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"r_shannon")));
 321                                                 outputNames.push_back(fileNameRoot+"r_shannon"); outputTypes["r_shannon"].push_back(fileNameRoot+"r_shannon");
 322                                         }else if (Estimators[i] == "shannoneven") {
 323                                                 rDisplays.push_back(new RareDisplay(new ShannonEven(), new ThreeColumnFile(fileNameRoot+"r_shannoneven")));
 324                                                 outputNames.push_back(fileNameRoot+"r_shannoneven"); outputTypes["r_shannoneven"].push_back(fileNameRoot+"r_shannoneven");
 325                                         }else if (Estimators[i] == "heip") {
 326                                                 rDisplays.push_back(new RareDisplay(new Heip(), new ThreeColumnFile(fileNameRoot+"r_heip")));
 327                                                 outputNames.push_back(fileNameRoot+"r_heip"); outputTypes["r_heip"].push_back(fileNameRoot+"r_heip");
 328                                         }else if (Estimators[i] == "smithwilson") {
 329                                                 rDisplays.push_back(new RareDisplay(new SmithWilson(), new ThreeColumnFile(fileNameRoot+"r_smithwilson")));
 330                                                 outputNames.push_back(fileNameRoot+"r_smithwilson"); outputTypes["r_smithwilson"].push_back(fileNameRoot+"r_smithwilson");
 331                                         }else if (Estimators[i] == "npshannon") {
 332                                                 rDisplays.push_back(new RareDisplay(new NPShannon(), new ThreeColumnFile(fileNameRoot+"r_npshannon")));
 333                                                 outputNames.push_back(fileNameRoot+"r_npshannon"); outputTypes["r_npshannon"].push_back(fileNameRoot+"r_npshannon");
 334                                         }else if (Estimators[i] == "simpson") {
 335                                                 rDisplays.push_back(new RareDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"r_simpson")));
 336                                                 outputNames.push_back(fileNameRoot+"r_simpson"); outputTypes["r_simpson"].push_back(fileNameRoot+"r_simpson");
 337                                         }else if (Estimators[i] == "simpsoneven") {
 338                                                 rDisplays.push_back(new RareDisplay(new SimpsonEven(), new ThreeColumnFile(fileNameRoot+"r_simpsoneven")));
 339                                                 outputNames.push_back(fileNameRoot+"r_simpsoneven"); outputTypes["r_simpsoneven"].push_back(fileNameRoot+"r_simpsoneven");
 340                                         }else if (Estimators[i] == "invsimpson") {
 341                                                 rDisplays.push_back(new RareDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+"r_invsimpson")));
 342                                                 outputNames.push_back(fileNameRoot+"r_invsimpson"); outputTypes["r_invsimpson"].push_back(fileNameRoot+"r_invsimpson");
 343                                         }else if (Estimators[i] == "bootstrap") {
 344                                                 rDisplays.push_back(new RareDisplay(new Bootstrap(), new ThreeColumnFile(fileNameRoot+"r_bootstrap")));
 345                                                 outputNames.push_back(fileNameRoot+"r_bootstrap"); outputTypes["r_bootstrap"].push_back(fileNameRoot+"r_bootstrap");
 346                                         }else if (Estimators[i] == "coverage") {
 347                                                 rDisplays.push_back(new RareDisplay(new Coverage(), new ThreeColumnFile(fileNameRoot+"r_coverage")));
 348                                                 outputNames.push_back(fileNameRoot+"r_coverage"); outputTypes["r_coverage"].push_back(fileNameRoot+"r_coverage");
 349                                         }else if (Estimators[i] == "nseqs") {
 350                                                 rDisplays.push_back(new RareDisplay(new NSeqs(), new ThreeColumnFile(fileNameRoot+"r_nseqs")));
 351                                                 outputNames.push_back(fileNameRoot+"r_nseqs"); outputTypes["r_nseqs"].push_back(fileNameRoot+"r_nseqs");
 352                                         }
 353                                 }
 354                         }
 355
 356
 357                         //if the users entered no valid calculators don't execute command
 358                         if (rDisplays.size() == 0) { for(int i=0;i<rDisplays.size();i++){       delete rDisplays[i];    }  return 0; }
 359
 360                         input = new InputData(inputFileNames[p], format);
 361                         order = input->getOrderVector();
 362                         string lastLabel = order->getLabel();
 363
 364                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 365                         set<string> processedLabels;
 366                         set<string> userLabels = labels;
 367
 368                         if (m->control_pressed) { for(int i=0;i<rDisplays.size();i++){  delete rDisplays[i];    }  delete input;  delete order;  for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 369
 370                         //as long as you are not at the end of the file or done wih the lines you want
 371                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 372
 373                                 if (m->control_pressed) { for(int i=0;i<rDisplays.size();i++){  delete rDisplays[i];    }  delete input;  delete order;  for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 374
 375
 376                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 377
 378                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 379                                         rCurve = new Rarefact(order, rDisplays, processors);
 380                                         rCurve->getCurve(freq, nIters);
 381                                         delete rCurve;
 382
 383                                         processedLabels.insert(order->getLabel());
 384                                         userLabels.erase(order->getLabel());
 385                                 }
 386
 387                                 if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 388                                         string saveLabel = order->getLabel();
 389
 390                                         delete order;
 391                                         order = (input->getOrderVector(lastLabel));
 392
 393                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 394                                         rCurve = new Rarefact(order, rDisplays, processors);
 395                                         rCurve->getCurve(freq, nIters);
 396                                         delete rCurve;
 397
 398                                         processedLabels.insert(order->getLabel());
 399                                         userLabels.erase(order->getLabel());
 400
 401                                         //restore real lastlabel to save below
 402                                         order->setLabel(saveLabel);
 403                                 }
 404
 405                                 lastLabel = order->getLabel();
 406
 407                                 delete order;
 408                                 order = (input->getOrderVector());
 409                         }
 410
 411                         if (m->control_pressed) { for(int i=0;i<rDisplays.size();i++){  delete rDisplays[i];    }  delete input;   for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 412
 413                         //output error messages about any remaining user labels
 414                         set<string>::iterator it;
 415                         bool needToRun = false;
 416                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 417                                 m->mothurOut("Your file does not include the label " + *it);
 418                                 if (processedLabels.count(lastLabel) != 1) {
 419                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 420                                         needToRun = true;
 421                                 }else {
 422                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 423                                 }
 424                         }
 425
 426                         if (m->control_pressed) { for(int i=0;i<rDisplays.size();i++){  delete rDisplays[i];    }  delete input;   for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 427
 428                         //run last label if you need to
 429                         if (needToRun == true)  {
 430                                 if (order != NULL) {    delete order;   }
 431                                 order = (input->getOrderVector(lastLabel));
 432
 433                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 434                                 rCurve = new Rarefact(order, rDisplays, processors);
 435                                 rCurve->getCurve(freq, nIters);
 436                                 delete rCurve;
 437
 438                                 delete order;
 439                         }
 440
 441
 442                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 443                         rDisplays.clear();
 444                         delete input;
 445                 }
 446
 447
 448                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 449
 450                 //create summary file containing all the groups data for each label - this function just combines the info from the files already created.
 451                 if ((sharedfile != "") && (groupMode)) {   outputNames = createGroupFile(outputNames);  }
 452
 453                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 454
 455                 m->mothurOutEndLine();
 456                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 457                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 458                 m->mothurOutEndLine();
 459
 460                 return 0;
 461         }
 462         catch(exception& e) {
 463                 m->errorOut(e, "RareFactCommand", "execute");
 464                 exit(1);
 465         }
 466 }
 467 //**********************************************************************************************************************
 468 vector<string> RareFactCommand::createGroupFile(vector<string>& outputNames) {
 469         try {
 470
 471                 vector<string> newFileNames;
 472
 473                 //find different types of files
 474                 map<string, vector<string> > typesFiles;
 475                 for (int i = 0; i < outputNames.size(); i++) {
 476                         string extension = m->getExtension(outputNames[i]);
 477
 478                         ifstream in;
 479                         m->openInputFile(outputNames[i], in);
 480
 481                         string labels = m->getline(in);
 482                         string newLine = labels.substr(0, labels.find_first_of('\t'));
 483
 484                         newLine += "\tGroup" + labels.substr(labels.find_first_of('\t'));
 485
 486                         typesFiles[extension].push_back(outputNames[i]);
 487
 488                         string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 489
 490                         //print headers
 491                         ofstream out;
 492                         m->openOutputFile(combineFileName, out);
 493                         out << newLine << endl;
 494                         out.close();
 495
 496                 }
 497
 498                 //for each type create a combo file
 499                 for (map<string, vector<string> >::iterator it = typesFiles.begin(); it != typesFiles.end(); it++) {
 500
 501                         ofstream out;
 502                         string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + it->first;
 503                         m->openOutputFileAppend(combineFileName, out);
 504                         newFileNames.push_back(combineFileName);
 505
 506                         vector<string> thisTypesFiles = it->second;
 507
 508                         //open each type summary file
 509                         map<string, vector<string> > files; //maps file name to lines in file
 510                         int maxLines = 0;
 511                         for (int i=0; i<thisTypesFiles.size(); i++) {
 512
 513                                 ifstream temp;
 514                                 m->openInputFile(thisTypesFiles[i], temp);
 515
 516                                 //read through first line - labels
 517                                 m->getline(temp);       m->gobble(temp);
 518
 519                                 vector<string> thisFilesLines;
 520                                 string group = m->getRootName(thisTypesFiles[i]);
 521                                 group = group.substr(0, group.length()-1);
 522                                 group = group.substr(group.find_last_of('.')+1);
 523
 524                                 thisFilesLines.push_back(group);
 525                                 while (!temp.eof()){
 526
 527                                         string thisLine = m->getline(temp);
 528
 529                                         thisFilesLines.push_back(thisLine);
 530
 531                                         m->gobble(temp);
 532                                 }
 533
 534                                 files[thisTypesFiles[i]] = thisFilesLines;
 535
 536                                 //save longest file for below
 537                                 if (maxLines < thisFilesLines.size()) { maxLines = thisFilesLines.size(); }
 538
 539                                 temp.close();
 540                                 m->mothurRemove(thisTypesFiles[i]);
 541                         }
 542
 543
 544                         //for each label
 545                         for (int k = 1; k < maxLines; k++) {
 546
 547                                 //grab data for each group
 548                                 for (int i=0; i<thisTypesFiles.size(); i++) {
 549
 550                                         string output = "NA\t";
 551                                         if (k < files[thisTypesFiles[i]].size()) {
 552                                                 string line = files[thisTypesFiles[i]][k];
 553                                                 output = line.substr(0, line.find_first_of('\t'));
 554                                                 output += '\t' + files[thisTypesFiles[i]][0] + '\t' + line.substr(line.find_first_of('\t'));
 555                                         }else{
 556                                                 output += '\t' + files[thisTypesFiles[i]][0] + '\t';
 557                                         }
 558                                         out << output << endl;
 559                                 }
 560                         }
 561
 562                         out.close();
 563
 564                 }
 565
 566                 //return combine file name
 567                 return newFileNames;
 568
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "RareFactCommand", "createGroupFile");
 572                 exit(1);
 573         }
 574 }
 575 //**********************************************************************************************************************
 576 vector<string> RareFactCommand::parseSharedFile(string filename) {
 577         try {
 578                 vector<string> filenames;
 579
 580                 map<string, ofstream*> filehandles;
 581                 map<string, ofstream*>::iterator it3;
 582
 583                 input = new InputData(filename, "sharedfile");
 584                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 585
 586                 string sharedFileRoot = m->getRootName(filename);
 587
 588                 //clears file before we start to write to it below
 589                 for (int i=0; i<lookup.size(); i++) {
 590                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 591                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 592                 }
 593
 594                 ofstream* temp;
 595                 for (int i=0; i<lookup.size(); i++) {
 596                         temp = new ofstream;
 597                         filehandles[lookup[i]->getGroup()] = temp;
 598                         groups.push_back(lookup[i]->getGroup());
 599                 }
 600
 601                 while(lookup[0] != NULL) {
 602
 603                         for (int i = 0; i < lookup.size(); i++) {
 604                                 RAbundVector rav = lookup[i]->getRAbundVector();
 605                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 606                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 607                                 (*(filehandles[lookup[i]->getGroup()])).close();
 608                         }
 609
 610                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 611                         lookup = input->getSharedRAbundVectors();
 612                 }
 613
 614                 //free memory
 615                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 616                         delete it3->second;
 617                 }
 618
 619                 delete input;
 620                 m->clearGroups();
 621
 622                 return filenames;
 623         }
 624         catch(exception& e) {
 625                 m->errorOut(e, "RareFactCommand", "parseSharedFile");
 626                 exit(1);
 627         }
 628 }
 629 //**********************************************************************************************************************
 630
 631
 632