summarycommand.cpp

   1 /*
   2  *  summarycommand.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 "summarycommand.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 "heip.h"
  22 #include "smithwilson.h"
  23 #include "shannoneven.h"
  24 #include "jackknife.h"
  25 #include "geom.h"
  26 #include "logsd.h"
  27 #include "qstat.h"
  28 #include "bergerparker.h"
  29 #include "bstick.h"
  30 #include "goodscoverage.h"
  31 #include "coverage.h"
  32 #include "efron.h"
  33 #include "boneh.h"
  34 #include "solow.h"
  35 #include "shen.h"
  36 #include "subsample.h"
  37
  38 //**********************************************************************************************************************
  39 vector<string> SummaryCommand::setParameters(){
  40         try {
  41                 CommandParameter plist("list", "InputTypes", "", "", "LRSS", "LRSS", "none","summary",false,false,true); parameters.push_back(plist);
  42                 CommandParameter prabund("rabund", "InputTypes", "", "", "LRSS", "LRSS", "none","summary",false,false); parameters.push_back(prabund);
  43                 CommandParameter psabund("sabund", "InputTypes", "", "", "LRSS", "LRSS", "none","summary",false,false); parameters.push_back(psabund);
  44                 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none","summary",false,false,true); parameters.push_back(pshared);
  45         CommandParameter psubsample("subsample", "String", "", "", "", "", "","",false,false); parameters.push_back(psubsample);
  46         CommandParameter piters("iters", "Number", "", "1000", "", "", "","",false,false); parameters.push_back(piters);
  47                 CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
  48                 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,true); parameters.push_back(pcalc);
  49                 CommandParameter pabund("abund", "Number", "", "10", "", "", "","",false,false); parameters.push_back(pabund);
  50                 CommandParameter psize("size", "Number", "", "0", "", "", "","",false,false); parameters.push_back(psize);
  51                 CommandParameter pgroupmode("groupmode", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pgroupmode);
  52                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  53                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  54
  55                 vector<string> myArray;
  56                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "SummaryCommand", "setParameters");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 string SummaryCommand::getHelpString(){
  66         try {
  67                 string helpString = "";
  68                 ValidCalculators validCalculator;
  69                 helpString += "The summary.single command parameters are list, sabund, rabund, shared, subsample, iters, label, calc, abund and groupmode.  list, sabund, rabund or shared is required unless you have a valid current file.\n";
  70                 helpString += "The summary.single command should be in the following format: \n";
  71                 helpString += "summary.single(label=yourLabel, calc=yourEstimators).\n";
  72                 helpString += "Example summary.single(label=unique-.01-.03, calc=sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson).\n";
  73                 helpString += validCalculator.printCalc("summary");
  74         helpString += "The subsample parameter allows you to enter the size of the sample or you can set subsample=T and mothur will use the size of your smallest group in the case of a shared file. With a list, sabund or rabund file you must provide a subsample size.\n";
  75         helpString += "The iters parameter allows you to choose the number of times you would like to run the subsample.\n";
  76                 helpString += "The default value calc is sobs-chao-ace-jack-shannon-npshannon-simpson\n";
  77                 helpString += "If you are running summary.single with a shared file and would like your summary results collated in one file, set groupmode=t. (Default=true).\n";
  78                 helpString += "The label parameter is used to analyze specific labels in your input.\n";
  79                 helpString += "Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabels).\n";
  80                 return helpString;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "SummaryCommand", "getHelpString");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 string SummaryCommand::getOutputPattern(string type) {
  89     try {
  90         string pattern = "";
  91
  92         if (type == "summary") {  pattern = "[filename],summary-[filename],[tag],summary"; }
  93         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  94
  95         return pattern;
  96     }
  97     catch(exception& e) {
  98         m->errorOut(e, "SummaryCommand", "getOutputPattern");
  99         exit(1);
 100     }
 101 }
 102 //**********************************************************************************************************************
 103 SummaryCommand::SummaryCommand(){
 104         try {
 105                 abort = true; calledHelp = true;
 106                 setParameters();
 107                 vector<string> tempOutNames;
 108                 outputTypes["summary"] = tempOutNames;
 109         }
 110         catch(exception& e) {
 111                 m->errorOut(e, "SummaryCommand", "SummaryCommand");
 112                 exit(1);
 113         }
 114 }
 115 //**********************************************************************************************************************
 116
 117 SummaryCommand::SummaryCommand(string option)  {
 118         try {
 119                 abort = false; calledHelp = false;
 120                 allLines = 1;
 121
 122                 //allow user to run help
 123                 if(option == "help") {  help();  abort = true; calledHelp = true; }
 124                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 125
 126                 else {
 127                         vector<string> myArray = setParameters();
 128
 129                         OptionParser parser(option);
 130                         map<string,string> parameters = parser.getParameters();
 131                         map<string,string>::iterator it;
 132
 133                         ValidParameters validParameter;
 134
 135                         //check to make sure all parameters are valid for command
 136                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 137                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 138                         }
 139
 140                         //initialize outputTypes
 141                         vector<string> tempOutNames;
 142                         outputTypes["summary"] = tempOutNames;
 143
 144                         //if the user changes the input directory command factory will send this info to us in the output parameter
 145                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 146                         if (inputDir == "not found"){   inputDir = "";          }
 147                         else {
 148                                 string path;
 149                                 it = parameters.find("shared");
 150                                 //user has given a template file
 151                                 if(it != parameters.end()){
 152                                         path = m->hasPath(it->second);
 153                                         //if the user has not given a path then, add inputdir. else leave path alone.
 154                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 155                                 }
 156
 157                                 it = parameters.find("rabund");
 158                                 //user has given a template file
 159                                 if(it != parameters.end()){
 160                                         path = m->hasPath(it->second);
 161                                         //if the user has not given a path then, add inputdir. else leave path alone.
 162                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 163                                 }
 164
 165                                 it = parameters.find("sabund");
 166                                 //user has given a template file
 167                                 if(it != parameters.end()){
 168                                         path = m->hasPath(it->second);
 169                                         //if the user has not given a path then, add inputdir. else leave path alone.
 170                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 171                                 }
 172
 173                                 it = parameters.find("list");
 174                                 //user has given a template file
 175                                 if(it != parameters.end()){
 176                                         path = m->hasPath(it->second);
 177                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 179                                 }
 180                         }
 181
 182                         //check for required parameters
 183                         listfile = validParameter.validFile(parameters, "list", true);
 184                         if (listfile == "not open") { listfile = ""; abort = true; }
 185                         else if (listfile == "not found") { listfile = ""; }
 186                         else {  format = "list"; inputfile = listfile; m->setListFile(listfile); }
 187
 188                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 189                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 190                         else if (sabundfile == "not found") { sabundfile = ""; }
 191                         else {  format = "sabund"; inputfile = sabundfile; m->setSabundFile(sabundfile); }
 192
 193                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 194                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 195                         else if (rabundfile == "not found") { rabundfile = ""; }
 196                         else {  format = "rabund"; inputfile = rabundfile; m->setRabundFile(rabundfile); }
 197
 198                         sharedfile = validParameter.validFile(parameters, "shared", true);
 199                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 200                         else if (sharedfile == "not found") { sharedfile = ""; }
 201                         else {  format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
 202
 203                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 204                                 //is there are current file available for any of these?
 205                                 //give priority to shared, then list, then rabund, then sabund
 206                                 //if there is a current shared file, use it
 207                                 sharedfile = m->getSharedFile();
 208                                 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 209                                 else {
 210                                         listfile = m->getListFile();
 211                                         if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 212                                         else {
 213                                                 rabundfile = m->getRabundFile();
 214                                                 if (rabundfile != "") { inputfile = rabundfile; format = "rabund"; m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 215                                                 else {
 216                                                         sabundfile = m->getSabundFile();
 217                                                         if (sabundfile != "") { inputfile = sabundfile; format = "sabund"; m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 218                                                         else {
 219                                                                 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();
 220                                                                 abort = true;
 221                                                         }
 222                                                 }
 223                                         }
 224                                 }
 225                         }
 226
 227                         //if the user changes the output directory command factory will send this info to us in the output parameter
 228                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputfile);              }
 229
 230                         //check for optional parameter and set defaults
 231                         // ...at some point should added some additional type checking...
 232                         label = validParameter.validFile(parameters, "label", false);
 233                         if (label == "not found") { label = ""; }
 234                         else {
 235                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 236                                 else { allLines = 1;  }
 237                         }
 238
 239                         calc = validParameter.validFile(parameters, "calc", false);
 240                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 241                         else {
 242                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 243                         }
 244                         m->splitAtDash(calc, Estimators);
 245                         if (m->inUsersGroups("citation", Estimators)) {
 246                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 247                                 //remove citation from list of calcs
 248                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 249                         }
 250
 251                         string temp;
 252                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
 253                         m->mothurConvert(temp, abund);
 254
 255                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 256                         m->mothurConvert(temp, size);
 257
 258                         temp = validParameter.validFile(parameters, "groupmode", false);                if (temp == "not found") { temp = "T"; }
 259                         groupMode = m->isTrue(temp);
 260
 261             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 262                         m->mothurConvert(temp, iters);
 263
 264             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 265                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 266             else {
 267                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 268                 else { subsample = false; subsampleSize = -1; }
 269             }
 270
 271             if (subsample == false) { iters = 0; }
 272             else {
 273                 //if you did not set a samplesize and are not using a sharedfile
 274                 if ((subsampleSize == -1) && (format != "sharedfile"))  { m->mothurOut("[ERROR]: If you want to subsample with a list, rabund or sabund file, you must provide the sample size.  You can do this by setting subsample=yourSampleSize.\n");  abort=true; }
 275             }
 276
 277                 }
 278         }
 279         catch(exception& e) {
 280                 m->errorOut(e, "SummaryCommand", "SummaryCommand");
 281                 exit(1);
 282         }
 283 }
 284 //**********************************************************************************************************************
 285
 286 int SummaryCommand::execute(){
 287         try {
 288
 289                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 290
 291                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 292                 else {  inputFileNames = parseSharedFile(sharedfile);  format = "rabund"; }
 293
 294                 if (m->control_pressed) { return 0; }
 295
 296                 int numLines = 0;
 297                 int numCols = 0;
 298                 map<string, string> groupIndex;
 299
 300                 for (int p = 0; p < inputFileNames.size(); p++) {
 301
 302                         numLines = 0;
 303                         numCols = 0;
 304
 305             map<string, string> variables;
 306             variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 307                         string fileNameRoot = getOutputFileName("summary",variables);
 308             variables["[tag]"] = "ave-std";
 309             string fileNameAve = getOutputFileName("summary",variables);
 310             outputNames.push_back(fileNameRoot); outputTypes["summary"].push_back(fileNameRoot);
 311
 312                         if (inputFileNames.size() > 1) {
 313                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 314                 groupIndex[fileNameRoot] = groups[p];
 315                         }
 316
 317                         sumCalculators.clear();
 318
 319                         ValidCalculators validCalculator;
 320
 321                         for (int i=0; i<Estimators.size(); i++) {
 322                                 if (validCalculator.isValidCalculator("summary", Estimators[i]) == true) {
 323                                         if(Estimators[i] == "sobs"){
 324                                                 sumCalculators.push_back(new Sobs());
 325                                         }else if(Estimators[i] == "chao"){
 326                                                 sumCalculators.push_back(new Chao1());
 327                                         }else if(Estimators[i] == "coverage"){
 328                                                 sumCalculators.push_back(new Coverage());
 329                                         }else if(Estimators[i] == "geometric"){
 330                                                 sumCalculators.push_back(new Geom());
 331                                         }else if(Estimators[i] == "logseries"){
 332                                                 sumCalculators.push_back(new LogSD());
 333                                         }else if(Estimators[i] == "qstat"){
 334                                                 sumCalculators.push_back(new QStat());
 335                                         }else if(Estimators[i] == "bergerparker"){
 336                                                 sumCalculators.push_back(new BergerParker());
 337                                         }else if(Estimators[i] == "bstick"){
 338                                                 sumCalculators.push_back(new BStick());
 339                                         }else if(Estimators[i] == "ace"){
 340                                                 if(abund < 5)
 341                                                         abund = 10;
 342                                                 sumCalculators.push_back(new Ace(abund));
 343                                         }else if(Estimators[i] == "jack"){
 344                                                 sumCalculators.push_back(new Jackknife());
 345                                         }else if(Estimators[i] == "shannon"){
 346                                                 sumCalculators.push_back(new Shannon());
 347                                         }else if(Estimators[i] == "shannoneven"){
 348                                                 sumCalculators.push_back(new ShannonEven());
 349                                         }else if(Estimators[i] == "npshannon"){
 350                                                 sumCalculators.push_back(new NPShannon());
 351                                         }else if(Estimators[i] == "heip"){
 352                                                 sumCalculators.push_back(new Heip());
 353                                         }else if(Estimators[i] == "smithwilson"){
 354                                                 sumCalculators.push_back(new SmithWilson());
 355                                         }else if(Estimators[i] == "simpson"){
 356                                                 sumCalculators.push_back(new Simpson());
 357                                         }else if(Estimators[i] == "simpsoneven"){
 358                                                 sumCalculators.push_back(new SimpsonEven());
 359                                         }else if(Estimators[i] == "invsimpson"){
 360                                                 sumCalculators.push_back(new InvSimpson());
 361                                         }else if(Estimators[i] == "bootstrap"){
 362                                                 sumCalculators.push_back(new Bootstrap());
 363                                         }else if (Estimators[i] == "nseqs") {
 364                                                 sumCalculators.push_back(new NSeqs());
 365                                         }else if (Estimators[i] == "goodscoverage") {
 366                                                 sumCalculators.push_back(new GoodsCoverage());
 367                                         }else if (Estimators[i] == "efron") {
 368                                                 sumCalculators.push_back(new Efron(size));
 369                                         }else if (Estimators[i] == "boneh") {
 370                                                 sumCalculators.push_back(new Boneh(size));
 371                                         }else if (Estimators[i] == "solow") {
 372                                                 sumCalculators.push_back(new Solow(size));
 373                                         }else if (Estimators[i] == "shen") {
 374                                                 sumCalculators.push_back(new Shen(size, abund));
 375                                         }
 376                                 }
 377                         }
 378
 379                         //if the users entered no valid calculators don't execute command
 380                         if (sumCalculators.size() == 0) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);  } return 0; }
 381
 382                         ofstream outputFileHandle;
 383                         m->openOutputFile(fileNameRoot, outputFileHandle);
 384                         outputFileHandle << "label";
 385
 386             ofstream outAve;
 387             if (subsample) {
 388                 m->openOutputFile(fileNameAve, outAve);
 389                 outputNames.push_back(fileNameAve); outputTypes["summary"].push_back(fileNameAve);
 390                 outAve << "label\tmethod";
 391                 outAve.setf(ios::fixed, ios::floatfield); outAve.setf(ios::showpoint);
 392                 if (inputFileNames.size() > 1) {
 393                     groupIndex[fileNameAve] = groups[p];
 394                 }
 395             }
 396
 397                         input = new InputData(inputFileNames[p], format);
 398                         sabund = input->getSAbundVector();
 399                         string lastLabel = sabund->getLabel();
 400
 401                         for(int i=0;i<sumCalculators.size();i++){
 402                                 if(sumCalculators[i]->getCols() == 1){
 403                                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 404                     if (subsample) { outAve << '\t' << sumCalculators[i]->getName();  }
 405                                         numCols++;
 406                                 }
 407                                 else{
 408                                         outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 409                     if (subsample) { outAve << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";  }
 410                                         numCols += 3;
 411                                 }
 412                         }
 413                         outputFileHandle << endl;
 414             if (subsample) {  outAve << endl; }
 415
 416                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 417                         set<string> processedLabels;
 418                         set<string> userLabels = labels;
 419
 420
 421
 422                         if (m->control_pressed) {  outputFileHandle.close(); outAve.close(); for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }  delete sabund;  delete input;  return 0;  }
 423
 424                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 425
 426                                 if (m->control_pressed) { outputFileHandle.close(); outAve.close();  for (int i = 0; i < outputNames.size(); i++) {     m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }  delete sabund;  delete input;  return 0;  }
 427
 428                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 429
 430                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 431                                         processedLabels.insert(sabund->getLabel());
 432                                         userLabels.erase(sabund->getLabel());
 433
 434                     process(sabund, outputFileHandle, outAve);
 435
 436                     if (m->control_pressed) { outputFileHandle.close(); outAve.close();  for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }  delete sabund;  delete input;  return 0;  }
 437                                         numLines++;
 438                                 }
 439
 440                                 if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 441                                         string saveLabel = sabund->getLabel();
 442
 443                                         delete sabund;
 444                                         sabund = input->getSAbundVector(lastLabel);
 445
 446                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 447                                         processedLabels.insert(sabund->getLabel());
 448                                         userLabels.erase(sabund->getLabel());
 449
 450                     process(sabund, outputFileHandle, outAve);
 451
 452                     if (m->control_pressed) { outputFileHandle.close(); outAve.close(); for (int i = 0; i < outputNames.size(); i++) {  m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }  delete sabund;  delete input;  return 0;  }
 453                                         numLines++;
 454
 455                                         //restore real lastlabel to save below
 456                                         sabund->setLabel(saveLabel);
 457                                 }
 458
 459                                 lastLabel = sabund->getLabel();
 460
 461                                 delete sabund;
 462                                 sabund = input->getSAbundVector();
 463                         }
 464
 465                         if (m->control_pressed) {  outputFileHandle.close(); outAve.close();  for (int i = 0; i < outputNames.size(); i++) {    m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }   delete input;  return 0;  }
 466
 467                         //output error messages about any remaining user labels
 468                         set<string>::iterator it;
 469                         bool needToRun = false;
 470                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 471                                 m->mothurOut("Your file does not include the label " + *it);
 472                                 if (processedLabels.count(lastLabel) != 1) {
 473                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 474                                         needToRun = true;
 475                                 }else {
 476                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 477                                 }
 478                         }
 479
 480                         //run last label if you need to
 481                         if (needToRun == true)  {
 482                                 if (sabund != NULL) {   delete sabund;  }
 483                                 sabund = input->getSAbundVector(lastLabel);
 484
 485                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 486                 process(sabund, outputFileHandle, outAve);
 487
 488                 if (m->control_pressed) { outputFileHandle.close(); outAve.close(); for (int i = 0; i < outputNames.size(); i++) {      m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }  delete sabund;  delete input;  return 0;  }
 489                                 numLines++;
 490                                 delete sabund;
 491                         }
 492
 493                         outputFileHandle.close();
 494             if (subsample) { outAve.close(); }
 495
 496                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }   delete input;  return 0;  }
 497
 498
 499                         delete input;
 500                         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 501                 }
 502
 503                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  }  return 0;  }
 504
 505                 //create summary file containing all the groups data for each label - this function just combines the info from the files already created.
 506                 if ((sharedfile != "") && (groupMode)) {   vector<string> comboNames = createGroupSummaryFile(numLines, numCols, outputNames, groupIndex);  for (int i = 0; i < comboNames.size(); i++) { outputNames.push_back(comboNames[i]); } }
 507
 508                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  }  return 0;  }
 509
 510                 m->mothurOutEndLine();
 511                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 512                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 513                 m->mothurOutEndLine();
 514
 515                 return 0;
 516         }
 517         catch(exception& e) {
 518                 m->errorOut(e, "SummaryCommand", "execute");
 519                 exit(1);
 520         }
 521 }
 522 //**********************************************************************************************************************
 523 int SummaryCommand::process(SAbundVector*& sabund, ofstream& outputFileHandle, ofstream& outAve) {
 524     try {
 525
 526         //calculator -> data -> values
 527         vector< vector< vector<double> > >  results; results.resize(sumCalculators.size());
 528
 529         outputFileHandle << sabund->getLabel();
 530
 531         SubSample sample;
 532         for (int thisIter = 0; thisIter < iters+1; thisIter++) {
 533
 534             SAbundVector* thisIterSabund = sabund;
 535
 536             //we want the summary results for the whole dataset, then the subsampling
 537             if ((thisIter > 0) && subsample) { //subsample sabund and run it
 538                 //copy sabund since getSample destroys it
 539                 RAbundVector rabund = sabund->getRAbundVector();
 540                 SAbundVector* newSabund = new SAbundVector();
 541                 *newSabund = rabund.getSAbundVector();
 542
 543                 sample.getSample(newSabund, subsampleSize);
 544                 thisIterSabund = newSabund;
 545             }
 546
 547             for(int i=0;i<sumCalculators.size();i++){
 548                 vector<double> data = sumCalculators[i]->getValues(thisIterSabund);
 549
 550                 if (m->control_pressed) {  return 0;  }
 551
 552                 if (thisIter == 0) {
 553                     outputFileHandle << '\t';
 554                     sumCalculators[i]->print(outputFileHandle);
 555                 }else {
 556                     //some of the calc have hci and lci need to make room for that
 557                     if (results[i].size() == 0) {  results[i].resize(data.size());  }
 558                     //save results for ave and std.
 559                     for (int j = 0; j < data.size(); j++) {
 560                         if (m->control_pressed) {  return 0;  }
 561                         results[i][j].push_back(data[j]);
 562                     }
 563                 }
 564             }
 565
 566             //cleanup memory
 567             if ((thisIter > 0) && subsample) { delete thisIterSabund; }
 568         }
 569         outputFileHandle << endl;
 570
 571         if (subsample) {
 572             outAve << sabund->getLabel() << '\t' << "ave\t";
 573             //find ave and std for this label and output
 574             //will need to modify the createGroupSummary to combine results and not mess with the .summary file.
 575
 576             //calcs -> values
 577             vector< vector<double> >  calcAverages; calcAverages.resize(sumCalculators.size());
 578             for (int i = 0; i < calcAverages.size(); i++) {  calcAverages[i].resize(results[i].size(), 0);  }
 579
 580             for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 581                 for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 582                     for (int j = 0; j < calcAverages[i].size(); j++) {
 583                         calcAverages[i][j] += results[i][j][thisIter];
 584                     }
 585                 }
 586             }
 587
 588             for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 589                 for (int j = 0; j < calcAverages[i].size(); j++) {
 590                     calcAverages[i][j] /= (float) iters;
 591                     outAve << calcAverages[i][j] << '\t';
 592                 }
 593             }
 594
 595             //find standard deviation
 596             vector< vector<double>  > stdDev; stdDev.resize(sumCalculators.size());
 597             for (int i = 0; i < stdDev.size(); i++) {  stdDev[i].resize(results[i].size(), 0);  }
 598
 599             for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
 600                 for (int i = 0; i < stdDev.size(); i++) {
 601                     for (int j = 0; j < stdDev[i].size(); j++) {
 602                         stdDev[i][j] += ((results[i][j][thisIter] - calcAverages[i][j]) * (results[i][j][thisIter] - calcAverages[i][j]));
 603                     }
 604                 }
 605             }
 606
 607             outAve << endl << sabund->getLabel() << '\t' << "std\t";
 608             for (int i = 0; i < stdDev.size(); i++) {  //finds average.
 609                 for (int j = 0; j < stdDev[i].size(); j++) {
 610                     stdDev[i][j] /= (float) iters;
 611                     stdDev[i][j] = sqrt(stdDev[i][j]);
 612                     outAve << stdDev[i][j] << '\t';
 613                 }
 614             }
 615             outAve << endl;
 616         }
 617
 618         return 0;
 619     }
 620     catch(exception& e) {
 621         m->errorOut(e, "SummaryCommand", "process");
 622         exit(1);
 623     }
 624 }
 625 //**********************************************************************************************************************
 626 vector<string> SummaryCommand::parseSharedFile(string filename) {
 627         try {
 628                 vector<string> filenames;
 629
 630                 map<string, ofstream*> filehandles;
 631                 map<string, ofstream*>::iterator it3;
 632
 633                 input = new InputData(filename, "sharedfile");
 634                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 635
 636                 string sharedFileRoot = m->getRootName(filename);
 637
 638         /******************************************************/
 639         if (subsample) {
 640             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 641                 subsampleSize = lookup[0]->getNumSeqs();
 642                 for (int i = 1; i < lookup.size(); i++) {
 643                     int thisSize = lookup[i]->getNumSeqs();
 644
 645                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 646                 }
 647             }else {
 648                 m->clearGroups();
 649                 vector<string> Groups;
 650                 vector<SharedRAbundVector*> temp;
 651                 for (int i = 0; i < lookup.size(); i++) {
 652                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 653                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 654                         delete lookup[i];
 655                     }else {
 656                         Groups.push_back(lookup[i]->getGroup());
 657                         temp.push_back(lookup[i]);
 658                     }
 659                 }
 660                 lookup = temp;
 661                 m->setGroups(Groups);
 662             }
 663
 664             if (lookup.size() < 1) { m->mothurOut("You have not provided enough valid groups.  I cannot run the command."); m->mothurOutEndLine(); m->control_pressed = true; delete input; return filenames; }
 665         }
 666
 667
 668                 /******************************************************/
 669
 670         //clears file before we start to write to it below
 671                 for (int i=0; i<lookup.size(); i++) {
 672                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 673                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 674                 }
 675
 676                 ofstream* temp;
 677                 for (int i=0; i<lookup.size(); i++) {
 678                         temp = new ofstream;
 679                         filehandles[lookup[i]->getGroup()] = temp;
 680                         groups.push_back(lookup[i]->getGroup());
 681                 }
 682
 683                 while(lookup[0] != NULL) {
 684
 685                         for (int i = 0; i < lookup.size(); i++) {
 686                                 RAbundVector rav = lookup[i]->getRAbundVector();
 687                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 688                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 689                                 (*(filehandles[lookup[i]->getGroup()])).close();
 690                         }
 691
 692                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 693                         lookup = input->getSharedRAbundVectors();
 694                 }
 695
 696                 //free memory
 697                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 698                         delete it3->second;
 699                 }
 700
 701                 delete input;
 702
 703                 return filenames;
 704         }
 705         catch(exception& e) {
 706                 m->errorOut(e, "SummaryCommand", "parseSharedFile");
 707                 exit(1);
 708         }
 709 }
 710 //**********************************************************************************************************************
 711 vector<string> SummaryCommand::createGroupSummaryFile(int numLines, int numCols, vector<string>& outputNames, map<string, string> groupIndex) {
 712         try {
 713
 714                 //open each groups summary file
 715         vector<string> newComboNames;
 716
 717                 map<string, map<string, vector<string> > > files;
 718         map<string, string> filesTypesLabels;
 719         map<string, int> filesTypesNumLines;
 720                 for (int i=0; i<outputNames.size(); i++) {
 721                         vector<string> thisFilesLines;
 722
 723                         ifstream temp;
 724                         m->openInputFile(outputNames[i], temp);
 725
 726                         //read through first line - labels
 727             string labelsLine = m->getline(temp);
 728             vector<string> theseLabels = m->splitWhiteSpace(labelsLine);
 729
 730             string newLabel = "";
 731             for (int j = 0; j < theseLabels.size(); j++) {
 732                  if (j == 1) {  newLabel += "group\t" + theseLabels[j] + '\t';
 733                 }else{  newLabel += theseLabels[j] + '\t';      }
 734             }
 735
 736                         m->gobble(temp);
 737
 738             int stop = numLines;
 739             if (theseLabels.size() != numCols+1) {  stop = numLines*2; }
 740                         //for each label
 741                         for (int k = 0; k < stop; k++) {
 742
 743                                 string thisLine = "";
 744                                 string tempLabel;
 745
 746                                 for (int j = 0; j < theseLabels.size(); j++) {
 747                                         temp >> tempLabel;
 748
 749                                         //save for later
 750                                         if (j == 1) { thisLine += groupIndex[outputNames[i]] + "\t" + tempLabel + "\t"; }
 751                                         else{  thisLine += tempLabel + "\t";    }
 752                                 }
 753
 754                                 thisLine += "\n";
 755
 756                                 thisFilesLines.push_back(thisLine);
 757
 758                                 m->gobble(temp);
 759                         }
 760
 761             string extension = m->getExtension(outputNames[i]);
 762             if (theseLabels.size() != numCols+1) { extension = ".ave-std" + extension;  }
 763             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 764                         m->mothurRemove(combineFileName); //remove old file
 765             filesTypesLabels[extension] = newLabel;
 766             filesTypesNumLines[extension] = stop;
 767
 768             map<string, map<string, vector<string> > >::iterator itFiles = files.find(extension);
 769             if (itFiles != files.end()) { //add new files info to existing type
 770                 files[extension][outputNames[i]] = thisFilesLines;
 771             }else {
 772                 map<string, vector<string> > thisFile;
 773                 thisFile[outputNames[i]] = thisFilesLines;
 774                 files[extension] = thisFile;
 775             }
 776
 777                         temp.close();
 778                         m->mothurRemove(outputNames[i]);
 779                 }
 780
 781
 782         for (map<string, map<string, vector<string> > >::iterator itFiles = files.begin(); itFiles != files.end(); itFiles++) {
 783
 784             if (m->control_pressed) { break; }
 785
 786             string extension = itFiles->first;
 787             map<string, vector<string> > thisType = itFiles->second;
 788             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 789             newComboNames.push_back(combineFileName);
 790             //open combined file
 791             ofstream out;
 792             m->openOutputFile(combineFileName, out);
 793
 794             //output label line to new file
 795             out <<  filesTypesLabels[extension] << endl;
 796
 797             //for each label
 798             for (int k = 0; k < filesTypesNumLines[extension]; k++) {
 799
 800                 //grab summary data for each group
 801                 for (map<string, vector<string> >::iterator itType = thisType.begin(); itType != thisType.end(); itType++) {
 802                     out << (itType->second)[k];
 803                 }
 804             }
 805
 806             outputNames.clear();
 807
 808             out.close();
 809         }
 810
 811                 //return combine file name
 812                 return newComboNames;
 813
 814         }
 815         catch(exception& e) {
 816                 m->errorOut(e, "SummaryCommand", "createGroupSummaryFile");
 817                 exit(1);
 818         }
 819 }
 820 //**********************************************************************************************************************