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                         m->mothurConvert(temp, freq);
 257
 258                         temp = validParameter.validFile(parameters, "abund", false);                    if (temp == "not found") { temp = "10"; }
 259                         m->mothurConvert(temp, abund);
 260
 261                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 262                         m->mothurConvert(temp, nIters);
 263
 264                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 265                         m->setProcessors(temp);
 266                         m->mothurConvert(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         map<string, set<int> > labelToEnds;
 286                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 287                 else {  inputFileNames = parseSharedFile(sharedfile, labelToEnds);  format = "rabund"; }
 288                 for (map<string, set<int> >::iterator it = labelToEnds.begin(); it != labelToEnds.end(); it++) {
 289             cout << it->first << endl;
 290             for (set<int>::iterator its = (it->second).begin(); its != (it->second).end(); its++) {
 291                 cout << (*its) << endl;
 292             }
 293         }
 294                 if (m->control_pressed) { return 0; }
 295
 296                 map<int, string> file2Group; //index in outputNames[i] -> group
 297                 for (int p = 0; p < inputFileNames.size(); p++) {
 298
 299                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 300
 301                         if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        }  m->clearGroups();  return 0; }
 302
 303                         if (inputFileNames.size() > 1) {
 304                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 305                         }
 306                         int i;
 307                         ValidCalculators validCalculator;
 308
 309
 310                         for (i=0; i<Estimators.size(); i++) {
 311                                 if (validCalculator.isValidCalculator("rarefaction", Estimators[i]) == true) {
 312                                         if (Estimators[i] == "sobs") {
 313                                                 rDisplays.push_back(new RareDisplay(new Sobs(), new ThreeColumnFile(fileNameRoot+"rarefaction")));
 314                                                 outputNames.push_back(fileNameRoot+"rarefaction"); outputTypes["rarefaction"].push_back(fileNameRoot+"rarefaction");
 315                                         }else if (Estimators[i] == "chao") {
 316                                                 rDisplays.push_back(new RareDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"r_chao")));
 317                                                 outputNames.push_back(fileNameRoot+"r_chao"); outputTypes["r_chao"].push_back(fileNameRoot+"r_chao");
 318                                         }else if (Estimators[i] == "ace") {
 319                                                 if(abund < 5)
 320                                                         abund = 10;
 321                                                 rDisplays.push_back(new RareDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"r_ace")));
 322                                                 outputNames.push_back(fileNameRoot+"r_ace"); outputTypes["r_ace"].push_back(fileNameRoot+"r_ace");
 323                                         }else if (Estimators[i] == "jack") {
 324                                                 rDisplays.push_back(new RareDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"r_jack")));
 325                                                 outputNames.push_back(fileNameRoot+"r_jack"); outputTypes["r_jack"].push_back(fileNameRoot+"r_jack");
 326                                         }else if (Estimators[i] == "shannon") {
 327                                                 rDisplays.push_back(new RareDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"r_shannon")));
 328                                                 outputNames.push_back(fileNameRoot+"r_shannon"); outputTypes["r_shannon"].push_back(fileNameRoot+"r_shannon");
 329                                         }else if (Estimators[i] == "shannoneven") {
 330                                                 rDisplays.push_back(new RareDisplay(new ShannonEven(), new ThreeColumnFile(fileNameRoot+"r_shannoneven")));
 331                                                 outputNames.push_back(fileNameRoot+"r_shannoneven"); outputTypes["r_shannoneven"].push_back(fileNameRoot+"r_shannoneven");
 332                                         }else if (Estimators[i] == "heip") {
 333                                                 rDisplays.push_back(new RareDisplay(new Heip(), new ThreeColumnFile(fileNameRoot+"r_heip")));
 334                                                 outputNames.push_back(fileNameRoot+"r_heip"); outputTypes["r_heip"].push_back(fileNameRoot+"r_heip");
 335                                         }else if (Estimators[i] == "smithwilson") {
 336                                                 rDisplays.push_back(new RareDisplay(new SmithWilson(), new ThreeColumnFile(fileNameRoot+"r_smithwilson")));
 337                                                 outputNames.push_back(fileNameRoot+"r_smithwilson"); outputTypes["r_smithwilson"].push_back(fileNameRoot+"r_smithwilson");
 338                                         }else if (Estimators[i] == "npshannon") {
 339                                                 rDisplays.push_back(new RareDisplay(new NPShannon(), new ThreeColumnFile(fileNameRoot+"r_npshannon")));
 340                                                 outputNames.push_back(fileNameRoot+"r_npshannon"); outputTypes["r_npshannon"].push_back(fileNameRoot+"r_npshannon");
 341                                         }else if (Estimators[i] == "simpson") {
 342                                                 rDisplays.push_back(new RareDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"r_simpson")));
 343                                                 outputNames.push_back(fileNameRoot+"r_simpson"); outputTypes["r_simpson"].push_back(fileNameRoot+"r_simpson");
 344                                         }else if (Estimators[i] == "simpsoneven") {
 345                                                 rDisplays.push_back(new RareDisplay(new SimpsonEven(), new ThreeColumnFile(fileNameRoot+"r_simpsoneven")));
 346                                                 outputNames.push_back(fileNameRoot+"r_simpsoneven"); outputTypes["r_simpsoneven"].push_back(fileNameRoot+"r_simpsoneven");
 347                                         }else if (Estimators[i] == "invsimpson") {
 348                                                 rDisplays.push_back(new RareDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+"r_invsimpson")));
 349                                                 outputNames.push_back(fileNameRoot+"r_invsimpson"); outputTypes["r_invsimpson"].push_back(fileNameRoot+"r_invsimpson");
 350                                         }else if (Estimators[i] == "bootstrap") {
 351                                                 rDisplays.push_back(new RareDisplay(new Bootstrap(), new ThreeColumnFile(fileNameRoot+"r_bootstrap")));
 352                                                 outputNames.push_back(fileNameRoot+"r_bootstrap"); outputTypes["r_bootstrap"].push_back(fileNameRoot+"r_bootstrap");
 353                                         }else if (Estimators[i] == "coverage") {
 354                                                 rDisplays.push_back(new RareDisplay(new Coverage(), new ThreeColumnFile(fileNameRoot+"r_coverage")));
 355                                                 outputNames.push_back(fileNameRoot+"r_coverage"); outputTypes["r_coverage"].push_back(fileNameRoot+"r_coverage");
 356                                         }else if (Estimators[i] == "nseqs") {
 357                                                 rDisplays.push_back(new RareDisplay(new NSeqs(), new ThreeColumnFile(fileNameRoot+"r_nseqs")));
 358                                                 outputNames.push_back(fileNameRoot+"r_nseqs"); outputTypes["r_nseqs"].push_back(fileNameRoot+"r_nseqs");
 359                                         }
 360                     if (inputFileNames.size() > 1) { file2Group[outputNames.size()-1] = groups[p]; }
 361                                 }
 362                         }
 363
 364
 365                         //if the users entered no valid calculators don't execute command
 366                         if (rDisplays.size() == 0) { for(int i=0;i<rDisplays.size();i++){       delete rDisplays[i];    }  return 0; }
 367
 368                         input = new InputData(inputFileNames[p], format);
 369                         order = input->getOrderVector();
 370                         string lastLabel = order->getLabel();
 371
 372                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 373                         set<string> processedLabels;
 374                         set<string> userLabels = labels;
 375
 376                         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; }
 377
 378                         //as long as you are not at the end of the file or done wih the lines you want
 379                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 380
 381                                 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; }
 382
 383
 384                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 385
 386                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 387                     map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 388                     set<int> ends;
 389                     if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 390                                         rCurve = new Rarefact(order, rDisplays, processors, ends);
 391                                         rCurve->getCurve(freq, nIters);
 392                                         delete rCurve;
 393
 394                                         processedLabels.insert(order->getLabel());
 395                                         userLabels.erase(order->getLabel());
 396                                 }
 397
 398                                 if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 399                                         string saveLabel = order->getLabel();
 400
 401                                         delete order;
 402                                         order = (input->getOrderVector(lastLabel));
 403
 404                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 405                                         map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 406                     set<int> ends;
 407                     if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 408                                         rCurve = new Rarefact(order, rDisplays, processors, ends);
 409
 410                                         rCurve->getCurve(freq, nIters);
 411                                         delete rCurve;
 412
 413                                         processedLabels.insert(order->getLabel());
 414                                         userLabels.erase(order->getLabel());
 415
 416                                         //restore real lastlabel to save below
 417                                         order->setLabel(saveLabel);
 418                                 }
 419
 420                                 lastLabel = order->getLabel();
 421
 422                                 delete order;
 423                                 order = (input->getOrderVector());
 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                         //output error messages about any remaining user labels
 429                         set<string>::iterator it;
 430                         bool needToRun = false;
 431                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 432                                 m->mothurOut("Your file does not include the label " + *it);
 433                                 if (processedLabels.count(lastLabel) != 1) {
 434                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 435                                         needToRun = true;
 436                                 }else {
 437                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 438                                 }
 439                         }
 440
 441                         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; }
 442
 443                         //run last label if you need to
 444                         if (needToRun == true)  {
 445                                 if (order != NULL) {    delete order;   }
 446                                 order = (input->getOrderVector(lastLabel));
 447
 448                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 449                                 map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 450                 set<int> ends;
 451                 if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 452                 rCurve = new Rarefact(order, rDisplays, processors, ends);
 453
 454                                 rCurve->getCurve(freq, nIters);
 455                                 delete rCurve;
 456
 457                                 delete order;
 458                         }
 459
 460
 461                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 462                         rDisplays.clear();
 463                         delete input;
 464                 }
 465
 466
 467                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 468
 469                 //create summary file containing all the groups data for each label - this function just combines the info from the files already created.
 470                 if ((sharedfile != "") && (groupMode)) {   outputNames = createGroupFile(outputNames, file2Group);  }
 471
 472                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 473
 474                 m->mothurOutEndLine();
 475                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 476                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 477                 m->mothurOutEndLine();
 478
 479                 return 0;
 480         }
 481         catch(exception& e) {
 482                 m->errorOut(e, "RareFactCommand", "execute");
 483                 exit(1);
 484         }
 485 }
 486 //**********************************************************************************************************************
 487 vector<string> RareFactCommand::createGroupFile(vector<string>& outputNames, map<int, string> file2Group) {
 488         try {
 489
 490                 vector<string> newFileNames;
 491
 492                 //find different types of files
 493                 map<string, map<string, string> > typesFiles;
 494         map<string, vector< vector<string> > > fileLabels; //combofile name to labels. each label is a vector because it may be unique lci hci.
 495         vector<string> groupNames;
 496                 for (int i = 0; i < outputNames.size(); i++) {
 497
 498                         string extension = m->getExtension(outputNames[i]);
 499             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 500                         m->mothurRemove(combineFileName); //remove old file
 501
 502                         ifstream in;
 503                         m->openInputFile(outputNames[i], in);
 504
 505                         string labels = m->getline(in);
 506
 507                         istringstream iss (labels,istringstream::in);
 508             string newLabel = ""; vector<string> theseLabels;
 509             while(!iss.eof()) {  iss >> newLabel; m->gobble(iss); theseLabels.push_back(newLabel); }
 510             vector< vector<string> > allLabels;
 511             vector<string> thisSet; thisSet.push_back(theseLabels[0]); allLabels.push_back(thisSet); thisSet.clear(); //makes "numSampled" its own grouping
 512             for (int j = 1; j < theseLabels.size()-1; j++) {
 513                 if (theseLabels[j+1] == "lci") {
 514                     thisSet.push_back(theseLabels[j]);
 515                     thisSet.push_back(theseLabels[j+1]);
 516                     thisSet.push_back(theseLabels[j+2]);
 517                     j++; j++;
 518                 }else{ //no lci or hci for this calc.
 519                     thisSet.push_back(theseLabels[j]);
 520                 }
 521                 allLabels.push_back(thisSet);
 522                 thisSet.clear();
 523             }
 524             fileLabels[combineFileName] = allLabels;
 525
 526             map<string, map<string, string> >::iterator itfind = typesFiles.find(extension);
 527             if (itfind != typesFiles.end()) {
 528                 (itfind->second)[outputNames[i]] = file2Group[i];
 529             }else {
 530                 map<string, string> temp;
 531                 temp[outputNames[i]] = file2Group[i];
 532                 typesFiles[extension] = temp;
 533             }
 534             if (!(m->inUsersGroups(file2Group[i], groupNames))) {  groupNames.push_back(file2Group[i]); }
 535                 }
 536
 537                 //for each type create a combo file
 538
 539                 for (map<string, map<string, string> >::iterator it = typesFiles.begin(); it != typesFiles.end(); it++) {
 540
 541                         ofstream out;
 542                         string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + it->first;
 543                         m->openOutputFileAppend(combineFileName, out);
 544                         newFileNames.push_back(combineFileName);
 545                         map<string, string> thisTypesFiles = it->second; //it->second maps filename to group
 546             set<int> numSampledSet;
 547
 548                         //open each type summary file
 549                         map<string, map<int, vector< vector<string> > > > files; //maps file name to lines in file
 550                         int maxLines = 0;
 551                         for (map<string, string>::iterator itFileNameGroup = thisTypesFiles.begin(); itFileNameGroup != thisTypesFiles.end(); itFileNameGroup++) {
 552
 553                 string thisfilename = itFileNameGroup->first;
 554                 string group = itFileNameGroup->second;
 555
 556                                 ifstream temp;
 557                                 m->openInputFile(thisfilename, temp);
 558
 559                                 //read through first line - labels
 560                                 m->getline(temp);       m->gobble(temp);
 561
 562                                 map<int, vector< vector<string> > > thisFilesLines;
 563                                 while (!temp.eof()){
 564                     int numSampled = 0;
 565                     temp >> numSampled; m->gobble(temp);
 566
 567                     vector< vector<string> > theseReads;
 568                     vector<string> thisSet; thisSet.push_back(toString(numSampled)); theseReads.push_back(thisSet); thisSet.clear();
 569                     for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //output thing like 0.03-A lci-A hci-A
 570                         vector<string> reads;
 571                         string next = "";
 572                         for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) { //output modified labels
 573                             temp >> next; m->gobble(temp);
 574                             reads.push_back(next);
 575                         }
 576                         theseReads.push_back(reads);
 577                     }
 578                     thisFilesLines[numSampled] = theseReads;
 579                     m->gobble(temp);
 580
 581                     numSampledSet.insert(numSampled);
 582                                 }
 583
 584                                 files[group] = thisFilesLines;
 585
 586                                 //save longest file for below
 587                                 if (maxLines < thisFilesLines.size()) { maxLines = thisFilesLines.size(); }
 588
 589                                 temp.close();
 590                                 m->mothurRemove(thisfilename);
 591                         }
 592
 593             //output new labels line
 594             out << fileLabels[combineFileName][0][0] << '\t';
 595             for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //output thing like 0.03-A lci-A hci-A
 596                 for (int n = 0; n < groupNames.size(); n++) { // for each group
 597                     for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) { //output modified labels
 598                         out << fileLabels[combineFileName][k][l] << '-' << groupNames[n] << '\t';
 599                     }
 600                 }
 601             }
 602                         out << endl;
 603
 604                         //for each label
 605                         for (set<int>::iterator itNumSampled = numSampledSet.begin(); itNumSampled != numSampledSet.end(); itNumSampled++) {
 606
 607                 out << (*itNumSampled) << '\t';
 608
 609                 if (m->control_pressed) { break; }
 610
 611                 for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //each chunk
 612                                     //grab data for each group
 613                     for (map<string, map<int, vector< vector<string> > > >::iterator itFileNameGroup = files.begin(); itFileNameGroup != files.end(); itFileNameGroup++) {
 614
 615                         string group = itFileNameGroup->first;
 616
 617                         map<int, vector< vector<string> > >::iterator itLine = files[group].find(*itNumSampled);
 618                         if (itLine != files[group].end()) {
 619                             for (int l = 0; l < (itLine->second)[k].size(); l++) {
 620                                 out << (itLine->second)[k][l] << '\t';
 621
 622                             }
 623                         }else {
 624                             for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) {
 625                                 out << "NA" << '\t';
 626                             }
 627                         }
 628                     }
 629                 }
 630                 out << endl;
 631                         }
 632                         out.close();
 633                 }
 634
 635                 //return combine file name
 636                 return newFileNames;
 637
 638         }
 639         catch(exception& e) {
 640                 m->errorOut(e, "RareFactCommand", "createGroupFile");
 641                 exit(1);
 642         }
 643 }
 644 //**********************************************************************************************************************
 645 vector<string> RareFactCommand::parseSharedFile(string filename, map<string, set<int> >& label2Ends) {
 646         try {
 647                 vector<string> filenames;
 648
 649                 map<string, ofstream*> filehandles;
 650                 map<string, ofstream*>::iterator it3;
 651
 652                 input = new InputData(filename, "sharedfile");
 653                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 654
 655                 string sharedFileRoot = m->getRootName(filename);
 656
 657                 //clears file before we start to write to it below
 658                 for (int i=0; i<lookup.size(); i++) {
 659                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 660                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 661                 }
 662
 663                 ofstream* temp;
 664                 for (int i=0; i<lookup.size(); i++) {
 665                         temp = new ofstream;
 666                         filehandles[lookup[i]->getGroup()] = temp;
 667                         groups.push_back(lookup[i]->getGroup());
 668                 }
 669
 670                 while(lookup[0] != NULL) {
 671
 672                         for (int i = 0; i < lookup.size(); i++) {
 673                                 RAbundVector rav = lookup[i]->getRAbundVector();
 674                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 675                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 676                                 (*(filehandles[lookup[i]->getGroup()])).close();
 677                 label2Ends[lookup[i]->getLabel()].insert(rav.getNumSeqs());
 678                         }
 679
 680                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 681                         lookup = input->getSharedRAbundVectors();
 682                 }
 683
 684                 //free memory
 685                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 686                         delete it3->second;
 687                 }
 688
 689                 delete input;
 690                 m->clearGroups();
 691
 692                 return filenames;
 693         }
 694         catch(exception& e) {
 695                 m->errorOut(e, "RareFactCommand", "parseSharedFile");
 696                 exit(1);
 697         }
 698 }
 699 //**********************************************************************************************************************
 700
 701
 702