collectcommand.cpp

   1 /*
   2  *  collectcommand.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 "collectcommand.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 "invsimpson.h"
  19 #include "npshannon.h"
  20 #include "shannon.h"
  21 #include "smithwilson.h"
  22 #include "heip.h"
  23 #include "shannoneven.h"
  24 #include "jackknife.h"
  25 #include "geom.h"
  26 #include "qstat.h"
  27 #include "logsd.h"
  28 #include "bergerparker.h"
  29 #include "bstick.h"
  30 #include "goodscoverage.h"
  31 #include "efron.h"
  32 #include "boneh.h"
  33 #include "solow.h"
  34 #include "shen.h"
  35 #include "coverage.h"
  36
  37
  38 //**********************************************************************************************************************
  39 vector<string> CollectCommand::setParameters(){
  40         try {
  41                 CommandParameter plist("list", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(plist);
  42                 CommandParameter prabund("rabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(prabund);
  43                 CommandParameter psabund("sabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(psabund);
  44                 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(pshared);
  45                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  46                 CommandParameter pfreq("freq", "Number", "", "100", "", "", "",false,false); parameters.push_back(pfreq);
  47                 CommandParameter pcalc("calc", "Multiple", "sobs-chao-nseqs-coverage-ace-jack-shannon-shannoneven-npshannon-heip-smithwilson-simpson-simpsoneven-invsimpson-bootstrap-geometric-qstat-logseries-bergerparker-bstick-goodscoverage-efron-boneh-solow-shen", "sobs-chao-ace-jack-shannon-npshannon-simpson", "", "", "",true,false); parameters.push_back(pcalc);
  48                 CommandParameter pabund("abund", "Number", "", "10", "", "", "",false,false); parameters.push_back(pabund);
  49                 CommandParameter psize("size", "Number", "", "0", "", "", "",false,false); parameters.push_back(psize);
  50                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  51                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  52
  53                 vector<string> myArray;
  54                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "CollectCommand", "setParameters");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 string CollectCommand::getHelpString(){
  64         try {
  65                 string helpString = "";
  66                 ValidCalculators validCalculator;
  67                 helpString += "The collect.single command parameters are list, sabund, rabund, shared, label, freq, calc and abund.  list, sabund, rabund or shared is required unless you have a valid current file. \n";
  68                 helpString += "The collect.single command should be in the following format: \n";
  69                 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";
  70                 helpString += "collect.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n";
  71                 helpString += "Example collect(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-chao-ace-jack).\n";
  72                 helpString += "The default values for freq is 100, and calc are sobs-chao-ace-jack-shannon-npshannon-simpson.\n";
  73                 helpString += validCalculator.printCalc("single");
  74                 helpString += "The label parameter is used to analyze specific labels in your input.\n";
  75                 helpString += "Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n";
  76                 return helpString;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "CollectCommand", "getHelpString");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84 string CollectCommand::getOutputFileNameTag(string type, string inputName=""){
  85         try {
  86         string outputFileName = "";
  87                 map<string, vector<string> >::iterator it;
  88
  89         //is this a type this command creates
  90         it = outputTypes.find(type);
  91         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  92         else {
  93             if (type == "sobs")             {  outputFileName =  "sobs";            }
  94             else if (type == "chao")        {  outputFileName =  "chao";            }
  95             else if (type == "nseqs")       {  outputFileName =  "nseqs";           }
  96             else if (type == "coverage")    {  outputFileName =  "coverage";        }
  97             else if (type == "ace")         {  outputFileName =  "ace";             }
  98             else if (type == "jack")        {  outputFileName =  "jack";            }
  99             else if (type == "shannon")     {  outputFileName =  "shannon";         }
 100             else if (type == "shannoneven") {  outputFileName =  "shannoneven";     }
 101             else if (type == "npshannon")   {  outputFileName =  "npshannon";       }
 102             else if (type == "heip")        {  outputFileName =  "heip";            }
 103             else if (type == "smithwilson") {  outputFileName =  "smithwilson";     }
 104             else if (type == "simpson")     {  outputFileName =  "simpson";         }
 105             else if (type == "simpsoneven") {  outputFileName =  "simpsoneven";     }
 106             else if (type == "invsimpson")  {  outputFileName =  "invsimpson";      }
 107             else if (type == "bootstrap")   {  outputFileName =  "bootstrap";       }
 108             else if (type == "geometric")   {  outputFileName =  "geometric";       }
 109             else if (type == "qstat")       {  outputFileName =  "qstat";           }
 110             else if (type == "logseries")   {  outputFileName =  "logseries";       }
 111             else if (type == "bergerparker") {  outputFileName =  "bergerparker";   }
 112             else if (type == "bstick")      {  outputFileName =  "bstick";          }
 113             else if (type == "goodscoverage") {  outputFileName =  "goodscoverage"; }
 114             else if (type == "efron")       {  outputFileName =  "efron";           }
 115             else if (type == "boneh")       {  outputFileName =  "boneh";           }
 116             else if (type == "solow")       {  outputFileName =  "solow";           }
 117             else if (type == "shen")        {  outputFileName =  "shen";            }
 118             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
 119         }
 120         return outputFileName;
 121         }
 122         catch(exception& e) {
 123                 m->errorOut(e, "CollectCommand", "getOutputFileNameTag");
 124                 exit(1);
 125         }
 126 }
 127 //**********************************************************************************************************************
 128 CollectCommand::CollectCommand(){
 129         try {
 130                 abort = true; calledHelp = true;
 131                 setParameters();
 132                 vector<string> tempOutNames;
 133                 outputTypes["sobs"] = tempOutNames;
 134                 outputTypes["chao"] = tempOutNames;
 135                 outputTypes["nseqs"] = tempOutNames;
 136                 outputTypes["coverage"] = tempOutNames;
 137                 outputTypes["ace"] = tempOutNames;
 138                 outputTypes["jack"] = tempOutNames;
 139                 outputTypes["shannon"] = tempOutNames;
 140                 outputTypes["shannoneven"] = tempOutNames;
 141                 outputTypes["npshannon"] = tempOutNames;
 142                 outputTypes["heip"] = tempOutNames;
 143                 outputTypes["smithwilson"] = tempOutNames;
 144                 outputTypes["simpson"] = tempOutNames;
 145                 outputTypes["simpsoneven"] = tempOutNames;
 146                 outputTypes["invsimpson"] = tempOutNames;
 147                 outputTypes["bootstrap"] = tempOutNames;
 148                 outputTypes["geometric"] = tempOutNames;
 149                 outputTypes["qstat"] = tempOutNames;
 150                 outputTypes["logseries"] = tempOutNames;
 151                 outputTypes["bergerparker"] = tempOutNames;
 152                 outputTypes["bstick"] = tempOutNames;
 153                 outputTypes["goodscoverage"] = tempOutNames;
 154                 outputTypes["efron"] = tempOutNames;
 155                 outputTypes["boneh"] = tempOutNames;
 156                 outputTypes["solow"] = tempOutNames;
 157                 outputTypes["shen"] = tempOutNames;
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "CollectCommand", "CollectCommand");
 161                 exit(1);
 162         }
 163 }
 164 //**********************************************************************************************************************
 165 CollectCommand::CollectCommand(string option)  {
 166         try {
 167                 abort = false; calledHelp = false;
 168                 allLines = 1;
 169
 170                 //allow user to run help
 171                 if(option == "help") { help(); calledHelp = true; abort = true; }
 172                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 173
 174                 else {
 175                         vector<string> myArray = setParameters();
 176
 177                         OptionParser parser(option);
 178                         map<string,string> parameters = parser.getParameters();
 179                         map<string,string>::iterator it;
 180
 181                         ValidParameters validParameter;
 182
 183                         //check to make sure all parameters are valid for command
 184                         for (it = parameters.begin(); it != parameters.end(); it++) {
 185                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 186                         }
 187
 188                         //initialize outputTypes
 189                         vector<string> tempOutNames;
 190                         outputTypes["sobs"] = tempOutNames;
 191                         outputTypes["chao"] = tempOutNames;
 192                         outputTypes["nseqs"] = tempOutNames;
 193                         outputTypes["coverage"] = tempOutNames;
 194                         outputTypes["ace"] = tempOutNames;
 195                         outputTypes["jack"] = tempOutNames;
 196                         outputTypes["shannon"] = tempOutNames;
 197                         outputTypes["shannoneven"] = tempOutNames;
 198                         outputTypes["npshannon"] = tempOutNames;
 199                         outputTypes["heip"] = tempOutNames;
 200                         outputTypes["smithwilson"] = tempOutNames;
 201                         outputTypes["simpson"] = tempOutNames;
 202                         outputTypes["simpsoneven"] = tempOutNames;
 203                         outputTypes["invsimpson"] = tempOutNames;
 204                         outputTypes["bootstrap"] = tempOutNames;
 205                         outputTypes["geometric"] = tempOutNames;
 206                         outputTypes["qstat"] = tempOutNames;
 207                         outputTypes["logseries"] = tempOutNames;
 208                         outputTypes["bergerparker"] = tempOutNames;
 209                         outputTypes["bstick"] = tempOutNames;
 210                         outputTypes["goodscoverage"] = tempOutNames;
 211                         outputTypes["efron"] = tempOutNames;
 212                         outputTypes["boneh"] = tempOutNames;
 213                         outputTypes["solow"] = tempOutNames;
 214                         outputTypes["shen"] = tempOutNames;
 215
 216                         //if the user changes the input directory command factory will send this info to us in the output parameter
 217                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 218                         if (inputDir == "not found"){   inputDir = "";          }
 219                         else {
 220                                 string path;
 221                                 it = parameters.find("shared");
 222                                 //user has given a template file
 223                                 if(it != parameters.end()){
 224                                         path = m->hasPath(it->second);
 225                                         //if the user has not given a path then, add inputdir. else leave path alone.
 226                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 227                                 }
 228
 229                                 it = parameters.find("rabund");
 230                                 //user has given a template file
 231                                 if(it != parameters.end()){
 232                                         path = m->hasPath(it->second);
 233                                         //if the user has not given a path then, add inputdir. else leave path alone.
 234                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 235                                 }
 236
 237                                 it = parameters.find("sabund");
 238                                 //user has given a template file
 239                                 if(it != parameters.end()){
 240                                         path = m->hasPath(it->second);
 241                                         //if the user has not given a path then, add inputdir. else leave path alone.
 242                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 243                                 }
 244
 245                                 it = parameters.find("list");
 246                                 //user has given a template file
 247                                 if(it != parameters.end()){
 248                                         path = m->hasPath(it->second);
 249                                         //if the user has not given a path then, add inputdir. else leave path alone.
 250                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 251                                 }
 252                         }
 253
 254                         //check for required parameters
 255                         listfile = validParameter.validFile(parameters, "list", true);
 256                         if (listfile == "not open") { listfile = ""; abort = true; }
 257                         else if (listfile == "not found") { listfile = ""; }
 258                         else {  format = "list"; inputfile = listfile; m->setListFile(listfile); }
 259
 260                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 261                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 262                         else if (sabundfile == "not found") { sabundfile = ""; }
 263                         else {  format = "sabund"; inputfile = sabundfile; m->setSabundFile(sabundfile); }
 264
 265                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 266                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 267                         else if (rabundfile == "not found") { rabundfile = ""; }
 268                         else {  format = "rabund"; inputfile = rabundfile; m->setRabundFile(rabundfile); }
 269
 270                         sharedfile = validParameter.validFile(parameters, "shared", true);
 271                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 272                         else if (sharedfile == "not found") { sharedfile = ""; }
 273                         else {  format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
 274
 275
 276                         //if the user changes the output directory command factory will send this info to us in the output parameter
 277                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 278
 279                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 280                                 //is there are current file available for any of these?
 281                                 //give priority to shared, then list, then rabund, then sabund
 282                                 //if there is a current shared file, use it
 283                                 sharedfile = m->getSharedFile();
 284                                 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 285                                 else {
 286                                         listfile = m->getListFile();
 287                                         if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 288                                         else {
 289                                                 rabundfile = m->getRabundFile();
 290                                                 if (rabundfile != "") { inputfile = rabundfile; format = "rabund"; m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 291                                                 else {
 292                                                         sabundfile = m->getSabundFile();
 293                                                         if (sabundfile != "") { inputfile = sabundfile; format = "sabund"; m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 294                                                         else {
 295                                                                 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();
 296                                                                 abort = true;
 297                                                         }
 298                                                 }
 299                                         }
 300                                 }
 301                         }
 302
 303                         //check for optional parameter and set defaults
 304                         // ...at some point should added some additional type checking...
 305                         label = validParameter.validFile(parameters, "label", false);
 306                         if (label == "not found") { label = ""; }
 307                         else {
 308                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 309                                 else { allLines = 1;  }
 310                         }
 311
 312                         //NOTE: if you add new calc options, don't forget to add them to the parameter initialize in setParameters or the gui won't be able to use them
 313                         calc = validParameter.validFile(parameters, "calc", false);
 314                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 315                         else {
 316                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 317                         }
 318                         m->splitAtDash(calc, Estimators);
 319                         if (m->inUsersGroups("citation", Estimators)) {
 320                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 321                                 //remove citation from list of calcs
 322                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 323                         }
 324
 325                         string temp;
 326                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 327                         m->mothurConvert(temp, freq);
 328
 329                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
 330                         m->mothurConvert(temp, abund);
 331
 332                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 333                         m->mothurConvert(temp, size);
 334                 }
 335
 336         }
 337         catch(exception& e) {
 338                 m->errorOut(e, "CollectCommand", "CollectCommand");
 339                 exit(1);
 340         }
 341 }
 342 //**********************************************************************************************************************
 343
 344 int CollectCommand::execute(){
 345         try {
 346
 347                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 348
 349                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 350                 else {  inputFileNames = parseSharedFile(sharedfile);  format = "rabund"; }
 351
 352                 for (int p = 0; p < inputFileNames.size(); p++) {
 353
 354                         if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        }  m->clearGroups();  return 0; }
 355
 356                         if (outputDir == "") { outputDir += m->hasPath(inputFileNames[p]); }
 357                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 358                         //globaldata->inputFileName = inputFileNames[p];
 359
 360                         if (inputFileNames.size() > 1) {
 361                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 362                         }
 363
 364                         ValidCalculators validCalculator;
 365
 366                         for (int i=0; i<Estimators.size(); i++) {
 367                                 if (validCalculator.isValidCalculator("single", Estimators[i]) == true) {
 368                                         if (Estimators[i] == "sobs") {
 369                                                 cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("sobs"))));
 370                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("sobs")); outputTypes["sobs"].push_back(fileNameRoot+getOutputFileNameTag("sobs"));
 371                                         }else if (Estimators[i] == "chao") {
 372                                                 cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("chao"))));
 373                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("chao")); outputTypes["chao"].push_back(fileNameRoot+getOutputFileNameTag("chao"));
 374                                         }else if (Estimators[i] == "nseqs") {
 375                                                 cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("nseqs"))));
 376                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("nseqs")); outputTypes["nseqs"].push_back(fileNameRoot+getOutputFileNameTag("nseqs"));
 377                                         }else if (Estimators[i] == "coverage") {
 378                                                 cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("coverage"))));
 379                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("coverage")); outputTypes["coverage"].push_back(fileNameRoot+getOutputFileNameTag("coverage"));
 380                                         }else if (Estimators[i] == "ace") {
 381                                                 cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("ace"))));
 382                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("ace")); outputTypes["ace"].push_back(fileNameRoot+getOutputFileNameTag("ace"));
 383                                         }else if (Estimators[i] == "jack") {
 384                                                 cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("jack"))));
 385                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("jack")); outputTypes["jack"].push_back(fileNameRoot+getOutputFileNameTag("jack"));
 386                                         }else if (Estimators[i] == "shannon") {
 387                                                 cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("shannon"))));
 388                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("shannon")); outputTypes["shannon"].push_back(fileNameRoot+getOutputFileNameTag("shannon"));
 389                                         }else if (Estimators[i] == "shannoneven") {
 390                                                 cDisplays.push_back(new CollectDisplay(new ShannonEven(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("shannoneven"))));
 391                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("shannoneven")); outputTypes["shannoneven"].push_back(fileNameRoot+getOutputFileNameTag("shannoneven"));
 392                                         }else if (Estimators[i] == "npshannon") {
 393                                                 cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("npshannon"))));
 394                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("npshannon")); outputTypes["npshannon"].push_back(fileNameRoot+getOutputFileNameTag("npshannon"));
 395                                         }else if (Estimators[i] == "heip") {
 396                                                 cDisplays.push_back(new CollectDisplay(new Heip(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("heip"))));
 397                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("heip")); outputTypes["heip"].push_back(fileNameRoot+getOutputFileNameTag("heip"));
 398                                         }else if (Estimators[i] == "smithwilson") {
 399                                                 cDisplays.push_back(new CollectDisplay(new SmithWilson(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("smithwilson"))));
 400                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("smithwilson")); outputTypes["smithwilson"].push_back(fileNameRoot+getOutputFileNameTag("smithwilson"));
 401                                         }else if (Estimators[i] == "simpson") {
 402                                                 cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("simpson"))));
 403                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("simpson")); outputTypes["simpson"].push_back(fileNameRoot+getOutputFileNameTag("simpson"));
 404                                         }else if (Estimators[i] == "simpsoneven") {
 405                                                 cDisplays.push_back(new CollectDisplay(new SimpsonEven(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("simpsoneven"))));
 406                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("simpsoneven")); outputTypes["simpsoneven"].push_back(fileNameRoot+getOutputFileNameTag("simpsoneven"));
 407                                         }else if (Estimators[i] == "invsimpson") {
 408                                                 cDisplays.push_back(new CollectDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("invsimpson"))));
 409                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("invsimpson")); outputTypes["invsimpson"].push_back(fileNameRoot+getOutputFileNameTag("invsimpson"));
 410                                         }else if (Estimators[i] == "bootstrap") {
 411                                                 cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("bootstrap"))));
 412                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("bootstrap")); outputTypes["bootstrap"].push_back(fileNameRoot+getOutputFileNameTag("bootstrap"));
 413                                         }else if (Estimators[i] == "geometric") {
 414                                                 cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("geometric"))));
 415                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("geometric")); outputTypes["geometric"].push_back(fileNameRoot+getOutputFileNameTag("geometric"));
 416                                         }else if (Estimators[i] == "qstat") {
 417                                                 cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("qstat"))));
 418                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("qstat")); outputTypes["qstat"].push_back(fileNameRoot+getOutputFileNameTag("qstat"));
 419                                         }else if (Estimators[i] == "logseries") {
 420                                                 cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("logseries"))));
 421                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("logseries")); outputTypes["logseries"].push_back(fileNameRoot+getOutputFileNameTag("logseries"));
 422                                         }else if (Estimators[i] == "bergerparker") {
 423                                                 cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("bergerparker"))));
 424                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("bergerparker")); outputTypes["bergerparker"].push_back(fileNameRoot+getOutputFileNameTag("bergerparker"));
 425                                         }else if (Estimators[i] == "bstick") {
 426                                                 cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("bstick"))));
 427                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("bstick")); outputTypes["bstick"].push_back(fileNameRoot+getOutputFileNameTag("bstick"));
 428                                         }else if (Estimators[i] == "goodscoverage") {
 429                                                 cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+getOutputFileNameTag("goodscoverage"))));
 430                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("goodscoverage")); outputTypes["goodscoverage"].push_back(fileNameRoot+getOutputFileNameTag("goodscoverage"));
 431                                         }else if (Estimators[i] == "efron") {
 432                                                 cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+getOutputFileNameTag("efron"))));
 433                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("efron")); outputTypes["efron"].push_back(fileNameRoot+getOutputFileNameTag("efron"));
 434                                         }else if (Estimators[i] == "boneh") {
 435                                                 cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+getOutputFileNameTag("boneh"))));
 436                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("boneh")); outputTypes["boneh"].push_back(fileNameRoot+getOutputFileNameTag("boneh"));
 437                                         }else if (Estimators[i] == "solow") {
 438                                                 cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+getOutputFileNameTag("solow"))));
 439                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("solow")); outputTypes["solow"].push_back(fileNameRoot+getOutputFileNameTag("solow"));
 440                                         }else if (Estimators[i] == "shen") {
 441                                                 cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+getOutputFileNameTag("shen"))));
 442                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("shen")); outputTypes["shen"].push_back(fileNameRoot+getOutputFileNameTag("shen"));
 443                                         }
 444                                 }
 445                         }
 446
 447                         //if the users entered no valid calculators don't execute command
 448                         if (cDisplays.size() == 0) { return 0; }
 449
 450                         input = new InputData(inputFileNames[p], format);
 451                         order = input->getOrderVector();
 452                         string lastLabel = order->getLabel();
 453
 454                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 455                         set<string> processedLabels;
 456                         set<string> userLabels = labels;
 457
 458                         if (m->control_pressed) {
 459                                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 460                                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } outputTypes.clear();
 461                                 delete input;
 462                                 delete order;
 463                                 m->clearGroups();
 464                                 return 0;
 465                         }
 466
 467
 468                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 469
 470                                 if (m->control_pressed) {
 471                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 472                                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } outputTypes.clear();
 473                                         delete input;
 474                                         delete order;
 475                                         m->clearGroups();
 476                                         return 0;
 477                                 }
 478
 479
 480                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 481
 482                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 483                                         cCurve = new Collect(order, cDisplays);
 484                                         cCurve->getCurve(freq);
 485                                         delete cCurve;
 486
 487                                         processedLabels.insert(order->getLabel());
 488                                         userLabels.erase(order->getLabel());
 489
 490
 491                                 }
 492                                 //you have a label the user want that is smaller than this label and the last label has not already been processed
 493                                 if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 494                                         string saveLabel = order->getLabel();
 495
 496                                         delete order;
 497                                         order = (input->getOrderVector(lastLabel));
 498
 499                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 500                                         cCurve = new Collect(order, cDisplays);
 501                                         cCurve->getCurve(freq);
 502                                         delete cCurve;
 503
 504
 505                                         processedLabels.insert(order->getLabel());
 506                                         userLabels.erase(order->getLabel());
 507
 508                                         //restore real lastlabel to save below
 509                                         order->setLabel(saveLabel);
 510                                 }
 511
 512                                 lastLabel = order->getLabel();
 513
 514                                 delete order;
 515                                 order = (input->getOrderVector());
 516                         }
 517
 518
 519                         if (m->control_pressed) {
 520                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 521                                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } outputTypes.clear();
 522                                         delete input;
 523                                         m->clearGroups();
 524                                         return 0;
 525                         }
 526
 527                         //output error messages about any remaining user labels
 528                         set<string>::iterator it;
 529                         bool needToRun = false;
 530                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 531                                 m->mothurOut("Your file does not include the label " + *it);
 532                                 if (processedLabels.count(lastLabel) != 1) {
 533                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 534                                         needToRun = true;
 535                                 }else {
 536                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 537                                 }
 538                         }
 539
 540                         //run last label if you need to
 541                         if (needToRun == true)  {
 542                                 if (order != NULL) {    delete order;   }
 543                                 order = (input->getOrderVector(lastLabel));
 544
 545                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 546
 547                                 cCurve = new Collect(order, cDisplays);
 548                                 cCurve->getCurve(freq);
 549                                 delete cCurve;
 550
 551                                 if (m->control_pressed) {
 552                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 553                                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        } outputTypes.clear();
 554                                         delete input;
 555                                         delete order;
 556                                         m->clearGroups();
 557                                         return 0;
 558                                 }
 559                                 delete order;
 560                         }
 561
 562                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 563                         cDisplays.clear();
 564                         delete input;
 565                 }
 566
 567                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 568
 569                 m->mothurOutEndLine();
 570                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 571                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 572                 m->mothurOutEndLine();
 573
 574
 575                 return 0;
 576         }
 577         catch(exception& e) {
 578                 m->errorOut(e, "CollectCommand", "execute");
 579                 exit(1);
 580         }
 581 }
 582
 583 //**********************************************************************************************************************
 584 vector<string> CollectCommand::parseSharedFile(string filename) {
 585         try {
 586                 vector<string> filenames;
 587
 588                 map<string, ofstream*> filehandles;
 589                 map<string, ofstream*>::iterator it3;
 590
 591                 input = new InputData(filename, "sharedfile");
 592                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 593
 594                 string sharedFileRoot = m->getRootName(filename);
 595
 596                 //clears file before we start to write to it below
 597                 for (int i=0; i<lookup.size(); i++) {
 598                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 599                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 600                 }
 601
 602                 ofstream* temp;
 603                 for (int i=0; i<lookup.size(); i++) {
 604                         temp = new ofstream;
 605                         filehandles[lookup[i]->getGroup()] = temp;
 606                         groups.push_back(lookup[i]->getGroup());
 607                 }
 608
 609                 while(lookup[0] != NULL) {
 610
 611                         for (int i = 0; i < lookup.size(); i++) {
 612                                 RAbundVector rav = lookup[i]->getRAbundVector();
 613                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 614                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 615                                 (*(filehandles[lookup[i]->getGroup()])).close();
 616                         }
 617
 618                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 619                         lookup = input->getSharedRAbundVectors();
 620                 }
 621
 622                 //free memory
 623                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 624                         delete it3->second;
 625                 }
 626
 627                 delete input;
 628
 629                 return filenames;
 630         }
 631         catch(exception& e) {
 632                 m->errorOut(e, "CollectCommand", "parseSharedFile");
 633                 exit(1);
 634         }
 635 }
 636 //**********************************************************************************************************************
 637