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 string RareFactCommand::getOutputFileNameTag(string type, string inputName=""){
  79         try {
  80         string outputFileName = "";
  81                 map<string, vector<string> >::iterator it;
  82
  83         //is this a type this command creates
  84         it = outputTypes.find(type);
  85         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  86         else {
  87             if (type == "rarefaction") {  outputFileName =  "rarefaction"; }
  88             else if (type == "r_chao") {  outputFileName =  "r_chao"; }
  89             else if (type == "r_ace") {  outputFileName =  "r_ace"; }
  90             else if (type == "r_jack") {  outputFileName =  "r_jack"; }
  91             else if (type == "r_shannon") {  outputFileName =  "r_shannon"; }
  92             else if (type == "r_shannoneven") {  outputFileName =  "r_shannoneven"; }
  93             else if (type == "r_smithwilson") {  outputFileName =  "r_smithwilson"; }
  94             else if (type == "r_npshannon") {  outputFileName =  "r_npshannon"; }
  95             else if (type == "r_simpson") {  outputFileName =  "r_simpson"; }
  96             else if (type == "r_simpsoneven") {  outputFileName =  "r_simpsoneven"; }
  97             else if (type == "r_invsimpson") {  outputFileName =  "r_invsimpson"; }
  98             else if (type == "r_bootstrap") {  outputFileName =  "r_bootstrap"; }
  99             else if (type == "r_coverage") {  outputFileName =  "r_coverage"; }
 100             else if (type == "r_nseqs") {  outputFileName =  "r_nseqs"; }
 101             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
 102         }
 103         return outputFileName;
 104         }
 105         catch(exception& e) {
 106                 m->errorOut(e, "RareFactCommand", "getOutputFileNameTag");
 107                 exit(1);
 108         }
 109 }
 110
 111 //**********************************************************************************************************************
 112 RareFactCommand::RareFactCommand(){
 113         try {
 114                 abort = true; calledHelp = true;
 115                 setParameters();
 116                 vector<string> tempOutNames;
 117                 outputTypes["rarefaction"] = tempOutNames;
 118                 outputTypes["r_chao"] = tempOutNames;
 119                 outputTypes["r_ace"] = tempOutNames;
 120                 outputTypes["r_jack"] = tempOutNames;
 121                 outputTypes["r_shannon"] = tempOutNames;
 122                 outputTypes["r_shannoneven"] = tempOutNames;
 123                 outputTypes["r_heip"] = tempOutNames;
 124                 outputTypes["r_smithwilson"] = tempOutNames;
 125                 outputTypes["r_npshannon"] = tempOutNames;
 126                 outputTypes["r_simpson"] = tempOutNames;
 127                 outputTypes["r_simpsoneven"] = tempOutNames;
 128                 outputTypes["r_invsimpson"] = tempOutNames;
 129                 outputTypes["r_bootstrap"] = tempOutNames;
 130                 outputTypes["r_coverage"] = tempOutNames;
 131                 outputTypes["r_nseqs"] = tempOutNames;
 132         }
 133         catch(exception& e) {
 134                 m->errorOut(e, "RareFactCommand", "RareFactCommand");
 135                 exit(1);
 136         }
 137 }
 138 //**********************************************************************************************************************
 139 RareFactCommand::RareFactCommand(string option)  {
 140         try {
 141                 abort = false; calledHelp = false;
 142                 allLines = 1;
 143
 144                 //allow user to run help
 145                 if(option == "help") { help(); abort = true; calledHelp = true; }
 146                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 147
 148                 else {
 149                         vector<string> myArray = setParameters();
 150
 151                         OptionParser parser(option);
 152                         map<string,string> parameters = parser.getParameters();
 153                         map<string,string>::iterator it;
 154
 155                         ValidParameters validParameter;
 156
 157                         //check to make sure all parameters are valid for command
 158                         for (it = parameters.begin(); it != parameters.end(); it++) {
 159                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 160                         }
 161
 162                         //initialize outputTypes
 163                         vector<string> tempOutNames;
 164                         outputTypes["rarefaction"] = tempOutNames;
 165                         outputTypes["r_chao"] = tempOutNames;
 166                         outputTypes["r_ace"] = tempOutNames;
 167                         outputTypes["r_jack"] = tempOutNames;
 168                         outputTypes["r_shannon"] = tempOutNames;
 169                         outputTypes["r_shannoneven"] = tempOutNames;
 170                         outputTypes["r_heip"] = tempOutNames;
 171                         outputTypes["r_smithwilson"] = tempOutNames;
 172                         outputTypes["r_npshannon"] = tempOutNames;
 173                         outputTypes["r_simpson"] = tempOutNames;
 174                         outputTypes["r_simpsoneven"] = tempOutNames;
 175                         outputTypes["r_invsimpson"] = tempOutNames;
 176                         outputTypes["r_bootstrap"] = tempOutNames;
 177                         outputTypes["r_coverage"] = tempOutNames;
 178                         outputTypes["r_nseqs"] = tempOutNames;
 179
 180                         //if the user changes the input directory command factory will send this info to us in the output parameter
 181                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 182                         if (inputDir == "not found"){   inputDir = "";          }
 183                         else {
 184                                 string path;
 185                                 it = parameters.find("shared");
 186                                 //user has given a template file
 187                                 if(it != parameters.end()){
 188                                         path = m->hasPath(it->second);
 189                                         //if the user has not given a path then, add inputdir. else leave path alone.
 190                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 191                                 }
 192
 193                                 it = parameters.find("rabund");
 194                                 //user has given a template file
 195                                 if(it != parameters.end()){
 196                                         path = m->hasPath(it->second);
 197                                         //if the user has not given a path then, add inputdir. else leave path alone.
 198                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 199                                 }
 200
 201                                 it = parameters.find("sabund");
 202                                 //user has given a template file
 203                                 if(it != parameters.end()){
 204                                         path = m->hasPath(it->second);
 205                                         //if the user has not given a path then, add inputdir. else leave path alone.
 206                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 207                                 }
 208
 209                                 it = parameters.find("list");
 210                                 //user has given a template file
 211                                 if(it != parameters.end()){
 212                                         path = m->hasPath(it->second);
 213                                         //if the user has not given a path then, add inputdir. else leave path alone.
 214                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 215                                 }
 216                         }
 217
 218                         //check for required parameters
 219                         listfile = validParameter.validFile(parameters, "list", true);
 220                         if (listfile == "not open") { listfile = ""; abort = true; }
 221                         else if (listfile == "not found") { listfile = ""; }
 222                         else {  format = "list"; inputfile = listfile; m->setListFile(listfile); }
 223
 224                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 225                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 226                         else if (sabundfile == "not found") { sabundfile = ""; }
 227                         else {  format = "sabund"; inputfile = sabundfile; m->setSabundFile(sabundfile); }
 228
 229                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 230                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 231                         else if (rabundfile == "not found") { rabundfile = ""; }
 232                         else {  format = "rabund"; inputfile = rabundfile; m->setRabundFile(rabundfile); }
 233
 234                         sharedfile = validParameter.validFile(parameters, "shared", true);
 235                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 236                         else if (sharedfile == "not found") { sharedfile = ""; }
 237                         else {  format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
 238
 239                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 240                                 //is there are current file available for any of these?
 241                                 //give priority to shared, then list, then rabund, then sabund
 242                                 //if there is a current shared file, use it
 243                                 sharedfile = m->getSharedFile();
 244                                 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 245                                 else {
 246                                         listfile = m->getListFile();
 247                                         if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 248                                         else {
 249                                                 rabundfile = m->getRabundFile();
 250                                                 if (rabundfile != "") { inputfile = rabundfile; format = "rabund"; m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 251                                                 else {
 252                                                         sabundfile = m->getSabundFile();
 253                                                         if (sabundfile != "") { inputfile = sabundfile; format = "sabund"; m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 254                                                         else {
 255                                                                 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();
 256                                                                 abort = true;
 257                                                         }
 258                                                 }
 259                                         }
 260                                 }
 261                         }
 262
 263                         //if the user changes the output directory command factory will send this info to us in the output parameter
 264                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputfile);              }
 265
 266                         //check for optional parameter and set defaults
 267                         // ...at some point should added some additional type checking...
 268                         label = validParameter.validFile(parameters, "label", false);
 269                         if (label == "not found") { label = ""; }
 270                         else {
 271                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 272                                 else { allLines = 1;  }
 273                         }
 274
 275                         calc = validParameter.validFile(parameters, "calc", false);
 276                         if (calc == "not found") { calc = "sobs";  }
 277                         else {
 278                                  if (calc == "default")  {  calc = "sobs";  }
 279                         }
 280                         m->splitAtDash(calc, Estimators);
 281                         if (m->inUsersGroups("citation", Estimators)) {
 282                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 283                                 //remove citation from list of calcs
 284                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 285                         }
 286
 287                         string temp;
 288                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 289                         m->mothurConvert(temp, freq);
 290
 291                         temp = validParameter.validFile(parameters, "abund", false);                    if (temp == "not found") { temp = "10"; }
 292                         m->mothurConvert(temp, abund);
 293
 294                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 295                         m->mothurConvert(temp, nIters);
 296
 297                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 298                         m->setProcessors(temp);
 299                         m->mothurConvert(temp, processors);
 300
 301                         temp = validParameter.validFile(parameters, "groupmode", false);                if (temp == "not found") { temp = "T"; }
 302                         groupMode = m->isTrue(temp);
 303                 }
 304
 305         }
 306         catch(exception& e) {
 307                 m->errorOut(e, "RareFactCommand", "RareFactCommand");
 308                 exit(1);
 309         }
 310 }
 311 //**********************************************************************************************************************
 312
 313 int RareFactCommand::execute(){
 314         try {
 315
 316                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 317
 318         map<string, set<int> > labelToEnds;
 319                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 320                 else {  inputFileNames = parseSharedFile(sharedfile, labelToEnds);  format = "rabund"; }
 321
 322         if (m->control_pressed) { return 0; }
 323
 324                 map<int, string> file2Group; //index in outputNames[i] -> group
 325                 for (int p = 0; p < inputFileNames.size(); p++) {
 326
 327                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 328
 329                         if (m->control_pressed) {  outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);        }  m->clearGroups();  return 0; }
 330
 331                         if (inputFileNames.size() > 1) {
 332                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 333                         }
 334                         int i;
 335                         ValidCalculators validCalculator;
 336
 337
 338                         for (i=0; i<Estimators.size(); i++) {
 339                                 if (validCalculator.isValidCalculator("rarefaction", Estimators[i]) == true) {
 340                                         if (Estimators[i] == "sobs") {
 341                                                 rDisplays.push_back(new RareDisplay(new Sobs(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("rarefaction"))));
 342                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("rarefaction")); outputTypes["rarefaction"].push_back(fileNameRoot+getOutputFileNameTag("rarefaction"));
 343                                         }else if (Estimators[i] == "chao") {
 344                                                 rDisplays.push_back(new RareDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_chao"))));
 345                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_chao")); outputTypes["r_chao"].push_back(fileNameRoot+getOutputFileNameTag("r_chao"));
 346                                         }else if (Estimators[i] == "ace") {
 347                                                 if(abund < 5)
 348                                                         abund = 10;
 349                                                 rDisplays.push_back(new RareDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_ace"))));
 350                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_ace")); outputTypes["r_ace"].push_back(fileNameRoot+getOutputFileNameTag("r_ace"));
 351                                         }else if (Estimators[i] == "jack") {
 352                                                 rDisplays.push_back(new RareDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_jack"))));
 353                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_jack")); outputTypes["r_jack"].push_back(fileNameRoot+getOutputFileNameTag("r_jack"));
 354                                         }else if (Estimators[i] == "shannon") {
 355                                                 rDisplays.push_back(new RareDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_shannon"))));
 356                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_shannon")); outputTypes["r_shannon"].push_back(fileNameRoot+getOutputFileNameTag("r_shannon"));
 357                                         }else if (Estimators[i] == "shannoneven") {
 358                                                 rDisplays.push_back(new RareDisplay(new ShannonEven(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_shannoneven"))));
 359                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_shannoneven")); outputTypes["r_shannoneven"].push_back(fileNameRoot+getOutputFileNameTag("r_shannoneven"));
 360                                         }else if (Estimators[i] == "heip") {
 361                                                 rDisplays.push_back(new RareDisplay(new Heip(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_heip"))));
 362                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_heip")); outputTypes["r_heip"].push_back(fileNameRoot+getOutputFileNameTag("r_heip"));
 363                                         }else if (Estimators[i] == "smithwilson") {
 364                                                 rDisplays.push_back(new RareDisplay(new SmithWilson(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_smithwilson"))));
 365                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_smithwilson")); outputTypes["r_smithwilson"].push_back(fileNameRoot+getOutputFileNameTag("r_smithwilson"));
 366                                         }else if (Estimators[i] == "npshannon") {
 367                                                 rDisplays.push_back(new RareDisplay(new NPShannon(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_npshannon"))));
 368                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_npshannon")); outputTypes["r_npshannon"].push_back(fileNameRoot+getOutputFileNameTag("r_npshannon"));
 369                                         }else if (Estimators[i] == "simpson") {
 370                                                 rDisplays.push_back(new RareDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_simpson"))));
 371                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_simpson")); outputTypes["r_simpson"].push_back(fileNameRoot+getOutputFileNameTag("r_simpson"));
 372                                         }else if (Estimators[i] == "simpsoneven") {
 373                                                 rDisplays.push_back(new RareDisplay(new SimpsonEven(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_simpsoneven"))));
 374                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_simpsoneven")); outputTypes["r_simpsoneven"].push_back(fileNameRoot+getOutputFileNameTag("r_simpsoneven"));
 375                                         }else if (Estimators[i] == "invsimpson") {
 376                                                 rDisplays.push_back(new RareDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_invsimpson"))));
 377                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_invsimpson")); outputTypes["r_invsimpson"].push_back(fileNameRoot+getOutputFileNameTag("r_invsimpson"));
 378                                         }else if (Estimators[i] == "bootstrap") {
 379                                                 rDisplays.push_back(new RareDisplay(new Bootstrap(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_bootstrap"))));
 380                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_bootstrap")); outputTypes["r_bootstrap"].push_back(fileNameRoot+getOutputFileNameTag("r_bootstrap"));
 381                                         }else if (Estimators[i] == "coverage") {
 382                                                 rDisplays.push_back(new RareDisplay(new Coverage(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_coverage"))));
 383                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_coverage")); outputTypes["r_coverage"].push_back(fileNameRoot+getOutputFileNameTag("r_coverage"));
 384                                         }else if (Estimators[i] == "nseqs") {
 385                                                 rDisplays.push_back(new RareDisplay(new NSeqs(), new ThreeColumnFile(fileNameRoot+getOutputFileNameTag("r_nseqs"))));
 386                                                 outputNames.push_back(fileNameRoot+getOutputFileNameTag("r_nseqs")); outputTypes["r_nseqs"].push_back(fileNameRoot+getOutputFileNameTag("r_nseqs"));
 387                                         }
 388                     if (inputFileNames.size() > 1) { file2Group[outputNames.size()-1] = groups[p]; }
 389                                 }
 390                         }
 391
 392
 393                         //if the users entered no valid calculators don't execute command
 394                         if (rDisplays.size() == 0) { for(int i=0;i<rDisplays.size();i++){       delete rDisplays[i];    }  return 0; }
 395
 396                         input = new InputData(inputFileNames[p], format);
 397                         order = input->getOrderVector();
 398                         string lastLabel = order->getLabel();
 399
 400                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 401                         set<string> processedLabels;
 402                         set<string> userLabels = labels;
 403
 404                         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; }
 405
 406                         //as long as you are not at the end of the file or done wih the lines you want
 407                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 408
 409                                 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; }
 410
 411
 412                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 413
 414                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 415                     map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 416                     set<int> ends;
 417                     if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 418                                         rCurve = new Rarefact(order, rDisplays, processors, ends);
 419                                         rCurve->getCurve(freq, nIters);
 420                                         delete rCurve;
 421
 422                                         processedLabels.insert(order->getLabel());
 423                                         userLabels.erase(order->getLabel());
 424                                 }
 425
 426                                 if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 427                                         string saveLabel = order->getLabel();
 428
 429                                         delete order;
 430                                         order = (input->getOrderVector(lastLabel));
 431
 432                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 433                                         map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 434                     set<int> ends;
 435                     if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 436                                         rCurve = new Rarefact(order, rDisplays, processors, ends);
 437
 438                                         rCurve->getCurve(freq, nIters);
 439                                         delete rCurve;
 440
 441                                         processedLabels.insert(order->getLabel());
 442                                         userLabels.erase(order->getLabel());
 443
 444                                         //restore real lastlabel to save below
 445                                         order->setLabel(saveLabel);
 446                                 }
 447
 448                                 lastLabel = order->getLabel();
 449
 450                                 delete order;
 451                                 order = (input->getOrderVector());
 452                         }
 453
 454                         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; }
 455
 456                         //output error messages about any remaining user labels
 457                         set<string>::iterator it;
 458                         bool needToRun = false;
 459                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 460                                 m->mothurOut("Your file does not include the label " + *it);
 461                                 if (processedLabels.count(lastLabel) != 1) {
 462                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 463                                         needToRun = true;
 464                                 }else {
 465                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 466                                 }
 467                         }
 468
 469                         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; }
 470
 471                         //run last label if you need to
 472                         if (needToRun == true)  {
 473                                 if (order != NULL) {    delete order;   }
 474                                 order = (input->getOrderVector(lastLabel));
 475
 476                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 477                                 map<string, set<int> >::iterator itEndings = labelToEnds.find(order->getLabel());
 478                 set<int> ends;
 479                 if (itEndings != labelToEnds.end()) { ends = itEndings->second; }
 480                 rCurve = new Rarefact(order, rDisplays, processors, ends);
 481
 482                                 rCurve->getCurve(freq, nIters);
 483                                 delete rCurve;
 484
 485                                 delete order;
 486                         }
 487
 488
 489                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 490                         rDisplays.clear();
 491                         delete input;
 492                 }
 493
 494
 495                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 496
 497                 //create summary file containing all the groups data for each label - this function just combines the info from the files already created.
 498                 if ((sharedfile != "") && (groupMode)) {   outputNames = createGroupFile(outputNames, file2Group);  }
 499
 500                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]); } return 0; }
 501
 502                 m->mothurOutEndLine();
 503                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 504                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 505                 m->mothurOutEndLine();
 506
 507                 return 0;
 508         }
 509         catch(exception& e) {
 510                 m->errorOut(e, "RareFactCommand", "execute");
 511                 exit(1);
 512         }
 513 }
 514 //**********************************************************************************************************************
 515 vector<string> RareFactCommand::createGroupFile(vector<string>& outputNames, map<int, string> file2Group) {
 516         try {
 517
 518                 vector<string> newFileNames;
 519
 520                 //find different types of files
 521                 map<string, map<string, string> > typesFiles;
 522         map<string, vector< vector<string> > > fileLabels; //combofile name to labels. each label is a vector because it may be unique lci hci.
 523         vector<string> groupNames;
 524                 for (int i = 0; i < outputNames.size(); i++) {
 525
 526                         string extension = m->getExtension(outputNames[i]);
 527             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 528                         m->mothurRemove(combineFileName); //remove old file
 529
 530                         ifstream in;
 531                         m->openInputFile(outputNames[i], in);
 532
 533                         string labels = m->getline(in);
 534
 535                         istringstream iss (labels,istringstream::in);
 536             string newLabel = ""; vector<string> theseLabels;
 537             while(!iss.eof()) {  iss >> newLabel; m->gobble(iss); theseLabels.push_back(newLabel); }
 538             vector< vector<string> > allLabels;
 539             vector<string> thisSet; thisSet.push_back(theseLabels[0]); allLabels.push_back(thisSet); thisSet.clear(); //makes "numSampled" its own grouping
 540             for (int j = 1; j < theseLabels.size()-1; j++) {
 541                 if (theseLabels[j+1] == "lci") {
 542                     thisSet.push_back(theseLabels[j]);
 543                     thisSet.push_back(theseLabels[j+1]);
 544                     thisSet.push_back(theseLabels[j+2]);
 545                     j++; j++;
 546                 }else{ //no lci or hci for this calc.
 547                     thisSet.push_back(theseLabels[j]);
 548                 }
 549                 allLabels.push_back(thisSet);
 550                 thisSet.clear();
 551             }
 552             fileLabels[combineFileName] = allLabels;
 553
 554             map<string, map<string, string> >::iterator itfind = typesFiles.find(extension);
 555             if (itfind != typesFiles.end()) {
 556                 (itfind->second)[outputNames[i]] = file2Group[i];
 557             }else {
 558                 map<string, string> temp;
 559                 temp[outputNames[i]] = file2Group[i];
 560                 typesFiles[extension] = temp;
 561             }
 562             if (!(m->inUsersGroups(file2Group[i], groupNames))) {  groupNames.push_back(file2Group[i]); }
 563                 }
 564
 565                 //for each type create a combo file
 566
 567                 for (map<string, map<string, string> >::iterator it = typesFiles.begin(); it != typesFiles.end(); it++) {
 568
 569                         ofstream out;
 570                         string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + it->first;
 571                         m->openOutputFileAppend(combineFileName, out);
 572                         newFileNames.push_back(combineFileName);
 573                         map<string, string> thisTypesFiles = it->second; //it->second maps filename to group
 574             set<int> numSampledSet;
 575
 576                         //open each type summary file
 577                         map<string, map<int, vector< vector<string> > > > files; //maps file name to lines in file
 578                         int maxLines = 0;
 579                         for (map<string, string>::iterator itFileNameGroup = thisTypesFiles.begin(); itFileNameGroup != thisTypesFiles.end(); itFileNameGroup++) {
 580
 581                 string thisfilename = itFileNameGroup->first;
 582                 string group = itFileNameGroup->second;
 583
 584                                 ifstream temp;
 585                                 m->openInputFile(thisfilename, temp);
 586
 587                                 //read through first line - labels
 588                                 m->getline(temp);       m->gobble(temp);
 589
 590                                 map<int, vector< vector<string> > > thisFilesLines;
 591                                 while (!temp.eof()){
 592                     int numSampled = 0;
 593                     temp >> numSampled; m->gobble(temp);
 594
 595                     vector< vector<string> > theseReads;
 596                     vector<string> thisSet; thisSet.push_back(toString(numSampled)); theseReads.push_back(thisSet); thisSet.clear();
 597                     for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //output thing like 0.03-A lci-A hci-A
 598                         vector<string> reads;
 599                         string next = "";
 600                         for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) { //output modified labels
 601                             temp >> next; m->gobble(temp);
 602                             reads.push_back(next);
 603                         }
 604                         theseReads.push_back(reads);
 605                     }
 606                     thisFilesLines[numSampled] = theseReads;
 607                     m->gobble(temp);
 608
 609                     numSampledSet.insert(numSampled);
 610                                 }
 611
 612                                 files[group] = thisFilesLines;
 613
 614                                 //save longest file for below
 615                                 if (maxLines < thisFilesLines.size()) { maxLines = thisFilesLines.size(); }
 616
 617                                 temp.close();
 618                                 m->mothurRemove(thisfilename);
 619                         }
 620
 621             //output new labels line
 622             out << fileLabels[combineFileName][0][0] << '\t';
 623             for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //output thing like 0.03-A lci-A hci-A
 624                 for (int n = 0; n < groupNames.size(); n++) { // for each group
 625                     for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) { //output modified labels
 626                         out << fileLabels[combineFileName][k][l] << '-' << groupNames[n] << '\t';
 627                     }
 628                 }
 629             }
 630                         out << endl;
 631
 632                         //for each label
 633                         for (set<int>::iterator itNumSampled = numSampledSet.begin(); itNumSampled != numSampledSet.end(); itNumSampled++) {
 634
 635                 out << (*itNumSampled) << '\t';
 636
 637                 if (m->control_pressed) { break; }
 638
 639                 for (int k = 1; k < fileLabels[combineFileName].size(); k++) { //each chunk
 640                                     //grab data for each group
 641                     for (map<string, map<int, vector< vector<string> > > >::iterator itFileNameGroup = files.begin(); itFileNameGroup != files.end(); itFileNameGroup++) {
 642
 643                         string group = itFileNameGroup->first;
 644
 645                         map<int, vector< vector<string> > >::iterator itLine = files[group].find(*itNumSampled);
 646                         if (itLine != files[group].end()) {
 647                             for (int l = 0; l < (itLine->second)[k].size(); l++) {
 648                                 out << (itLine->second)[k][l] << '\t';
 649
 650                             }
 651                         }else {
 652                             for (int l = 0; l < fileLabels[combineFileName][k].size(); l++) {
 653                                 out << "NA" << '\t';
 654                             }
 655                         }
 656                     }
 657                 }
 658                 out << endl;
 659                         }
 660                         out.close();
 661                 }
 662
 663                 //return combine file name
 664                 return newFileNames;
 665
 666         }
 667         catch(exception& e) {
 668                 m->errorOut(e, "RareFactCommand", "createGroupFile");
 669                 exit(1);
 670         }
 671 }
 672 //**********************************************************************************************************************
 673 vector<string> RareFactCommand::parseSharedFile(string filename, map<string, set<int> >& label2Ends) {
 674         try {
 675                 vector<string> filenames;
 676
 677                 map<string, ofstream*> filehandles;
 678                 map<string, ofstream*>::iterator it3;
 679
 680                 input = new InputData(filename, "sharedfile");
 681                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 682
 683                 string sharedFileRoot = m->getRootName(filename);
 684
 685                 //clears file before we start to write to it below
 686                 for (int i=0; i<lookup.size(); i++) {
 687                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 688                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 689                 }
 690
 691                 ofstream* temp;
 692                 for (int i=0; i<lookup.size(); i++) {
 693                         temp = new ofstream;
 694                         filehandles[lookup[i]->getGroup()] = temp;
 695                         groups.push_back(lookup[i]->getGroup());
 696                 }
 697
 698                 while(lookup[0] != NULL) {
 699
 700                         for (int i = 0; i < lookup.size(); i++) {
 701                                 RAbundVector rav = lookup[i]->getRAbundVector();
 702                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 703                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 704                                 (*(filehandles[lookup[i]->getGroup()])).close();
 705                 label2Ends[lookup[i]->getLabel()].insert(rav.getNumSeqs());
 706                         }
 707
 708                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 709                         lookup = input->getSharedRAbundVectors();
 710                 }
 711
 712                 //free memory
 713                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 714                         delete it3->second;
 715                 }
 716
 717                 delete input;
 718                 m->clearGroups();
 719
 720                 return filenames;
 721         }
 722         catch(exception& e) {
 723                 m->errorOut(e, "RareFactCommand", "parseSharedFile");
 724                 exit(1);
 725         }
 726 }
 727 //**********************************************************************************************************************
 728
 729
 730