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",false,false); parameters.push_back(plist);
  42                 CommandParameter prabund("rabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(prabund);
  43                 CommandParameter psabund("sabund", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(psabund);
  44                 CommandParameter pshared("shared", "InputTypes", "", "", "LRSS", "LRSS", "none",false,false); parameters.push_back(pshared);
  45         CommandParameter 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); 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::getOutputFileNameTag(string type, string inputName=""){
  89         try {
  90         string outputFileName = "";
  91                 map<string, vector<string> >::iterator it;
  92
  93         //is this a type this command creates
  94         it = outputTypes.find(type);
  95         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  96         else {
  97             if (type == "summary")            {   outputFileName =  "summary";   }
  98             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  99         }
 100         return outputFileName;
 101         }
 102         catch(exception& e) {
 103                 m->errorOut(e, "SummaryCommand", "getOutputFileNameTag");
 104                 exit(1);
 105         }
 106 }
 107 //**********************************************************************************************************************
 108 SummaryCommand::SummaryCommand(){
 109         try {
 110                 abort = true; calledHelp = true;
 111                 setParameters();
 112                 vector<string> tempOutNames;
 113                 outputTypes["summary"] = tempOutNames;
 114         }
 115         catch(exception& e) {
 116                 m->errorOut(e, "SummaryCommand", "SummaryCommand");
 117                 exit(1);
 118         }
 119 }
 120 //**********************************************************************************************************************
 121
 122 SummaryCommand::SummaryCommand(string option)  {
 123         try {
 124                 abort = false; calledHelp = false;
 125                 allLines = 1;
 126
 127                 //allow user to run help
 128                 if(option == "help") {  help();  abort = true; calledHelp = true; }
 129                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 130
 131                 else {
 132                         vector<string> myArray = setParameters();
 133
 134                         OptionParser parser(option);
 135                         map<string,string> parameters = parser.getParameters();
 136                         map<string,string>::iterator it;
 137
 138                         ValidParameters validParameter;
 139
 140                         //check to make sure all parameters are valid for command
 141                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 142                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 143                         }
 144
 145                         //initialize outputTypes
 146                         vector<string> tempOutNames;
 147                         outputTypes["summary"] = tempOutNames;
 148
 149                         //if the user changes the input directory command factory will send this info to us in the output parameter
 150                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 151                         if (inputDir == "not found"){   inputDir = "";          }
 152                         else {
 153                                 string path;
 154                                 it = parameters.find("shared");
 155                                 //user has given a template file
 156                                 if(it != parameters.end()){
 157                                         path = m->hasPath(it->second);
 158                                         //if the user has not given a path then, add inputdir. else leave path alone.
 159                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 160                                 }
 161
 162                                 it = parameters.find("rabund");
 163                                 //user has given a template file
 164                                 if(it != parameters.end()){
 165                                         path = m->hasPath(it->second);
 166                                         //if the user has not given a path then, add inputdir. else leave path alone.
 167                                         if (path == "") {       parameters["rabund"] = inputDir + it->second;           }
 168                                 }
 169
 170                                 it = parameters.find("sabund");
 171                                 //user has given a template file
 172                                 if(it != parameters.end()){
 173                                         path = m->hasPath(it->second);
 174                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                         if (path == "") {       parameters["sabund"] = inputDir + it->second;           }
 176                                 }
 177
 178                                 it = parameters.find("list");
 179                                 //user has given a template file
 180                                 if(it != parameters.end()){
 181                                         path = m->hasPath(it->second);
 182                                         //if the user has not given a path then, add inputdir. else leave path alone.
 183                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 184                                 }
 185                         }
 186
 187                         //check for required parameters
 188                         listfile = validParameter.validFile(parameters, "list", true);
 189                         if (listfile == "not open") { listfile = ""; abort = true; }
 190                         else if (listfile == "not found") { listfile = ""; }
 191                         else {  format = "list"; inputfile = listfile; m->setListFile(listfile); }
 192
 193                         sabundfile = validParameter.validFile(parameters, "sabund", true);
 194                         if (sabundfile == "not open") { sabundfile = ""; abort = true; }
 195                         else if (sabundfile == "not found") { sabundfile = ""; }
 196                         else {  format = "sabund"; inputfile = sabundfile; m->setSabundFile(sabundfile); }
 197
 198                         rabundfile = validParameter.validFile(parameters, "rabund", true);
 199                         if (rabundfile == "not open") { rabundfile = ""; abort = true; }
 200                         else if (rabundfile == "not found") { rabundfile = ""; }
 201                         else {  format = "rabund"; inputfile = rabundfile; m->setRabundFile(rabundfile); }
 202
 203                         sharedfile = validParameter.validFile(parameters, "shared", true);
 204                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 205                         else if (sharedfile == "not found") { sharedfile = ""; }
 206                         else {  format = "sharedfile"; inputfile = sharedfile; m->setSharedFile(sharedfile); }
 207
 208                         if ((sharedfile == "") && (listfile == "") && (rabundfile == "") && (sabundfile == "")) {
 209                                 //is there are current file available for any of these?
 210                                 //give priority to shared, then list, then rabund, then sabund
 211                                 //if there is a current shared file, use it
 212                                 sharedfile = m->getSharedFile();
 213                                 if (sharedfile != "") { inputfile = sharedfile; format = "sharedfile"; m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 214                                 else {
 215                                         listfile = m->getListFile();
 216                                         if (listfile != "") { inputfile = listfile; format = "list"; m->mothurOut("Using " + listfile + " as input file for the list parameter."); m->mothurOutEndLine(); }
 217                                         else {
 218                                                 rabundfile = m->getRabundFile();
 219                                                 if (rabundfile != "") { inputfile = rabundfile; format = "rabund"; m->mothurOut("Using " + rabundfile + " as input file for the rabund parameter."); m->mothurOutEndLine(); }
 220                                                 else {
 221                                                         sabundfile = m->getSabundFile();
 222                                                         if (sabundfile != "") { inputfile = sabundfile; format = "sabund"; m->mothurOut("Using " + sabundfile + " as input file for the sabund parameter."); m->mothurOutEndLine(); }
 223                                                         else {
 224                                                                 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();
 225                                                                 abort = true;
 226                                                         }
 227                                                 }
 228                                         }
 229                                 }
 230                         }
 231
 232                         //if the user changes the output directory command factory will send this info to us in the output parameter
 233                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(inputfile);              }
 234
 235                         //check for optional parameter and set defaults
 236                         // ...at some point should added some additional type checking...
 237                         label = validParameter.validFile(parameters, "label", false);
 238                         if (label == "not found") { label = ""; }
 239                         else {
 240                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 241                                 else { allLines = 1;  }
 242                         }
 243
 244                         calc = validParameter.validFile(parameters, "calc", false);
 245                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 246                         else {
 247                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
 248                         }
 249                         m->splitAtDash(calc, Estimators);
 250                         if (m->inUsersGroups("citation", Estimators)) {
 251                                 ValidCalculators validCalc; validCalc.printCitations(Estimators);
 252                                 //remove citation from list of calcs
 253                                 for (int i = 0; i < Estimators.size(); i++) { if (Estimators[i] == "citation") {  Estimators.erase(Estimators.begin()+i); break; } }
 254                         }
 255
 256                         string temp;
 257                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
 258                         m->mothurConvert(temp, abund);
 259
 260                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 261                         m->mothurConvert(temp, size);
 262
 263                         temp = validParameter.validFile(parameters, "groupmode", false);                if (temp == "not found") { temp = "T"; }
 264                         groupMode = m->isTrue(temp);
 265
 266             temp = validParameter.validFile(parameters, "iters", false);                        if (temp == "not found") { temp = "1000"; }
 267                         m->mothurConvert(temp, iters);
 268
 269             temp = validParameter.validFile(parameters, "subsample", false);            if (temp == "not found") { temp = "F"; }
 270                         if (m->isNumeric1(temp)) { m->mothurConvert(temp, subsampleSize); subsample = true; }
 271             else {
 272                 if (m->isTrue(temp)) { subsample = true; subsampleSize = -1; }  //we will set it to smallest group later
 273                 else { subsample = false; subsampleSize = -1; }
 274             }
 275
 276             if (subsample == false) { iters = 1; }
 277             else {
 278                 //if you did not set a samplesize and are not using a sharedfile
 279                 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; }
 280             }
 281
 282                 }
 283         }
 284         catch(exception& e) {
 285                 m->errorOut(e, "SummaryCommand", "SummaryCommand");
 286                 exit(1);
 287         }
 288 }
 289 //**********************************************************************************************************************
 290
 291 int SummaryCommand::execute(){
 292         try {
 293
 294                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 295
 296                 if ((format != "sharedfile")) { inputFileNames.push_back(inputfile);  }
 297                 else {  inputFileNames = parseSharedFile(sharedfile);  format = "rabund"; }
 298
 299                 if (m->control_pressed) { return 0; }
 300
 301                 int numLines = 0;
 302                 int numCols = 0;
 303                 map<string, string> groupIndex;
 304
 305                 for (int p = 0; p < inputFileNames.size(); p++) {
 306
 307                         numLines = 0;
 308                         numCols = 0;
 309
 310                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p])) + getOutputFileNameTag("summary");
 311             string fileNameAve = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p])) + "ave-std." + getOutputFileNameTag("summary");
 312             outputNames.push_back(fileNameRoot); outputTypes["summary"].push_back(fileNameRoot);
 313
 314                         if (inputFileNames.size() > 1) {
 315                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 316                 groupIndex[fileNameRoot] = groups[p];
 317                         }
 318
 319                         sumCalculators.clear();
 320
 321                         ValidCalculators validCalculator;
 322
 323                         for (int i=0; i<Estimators.size(); i++) {
 324                                 if (validCalculator.isValidCalculator("summary", Estimators[i]) == true) {
 325                                         if(Estimators[i] == "sobs"){
 326                                                 sumCalculators.push_back(new Sobs());
 327                                         }else if(Estimators[i] == "chao"){
 328                                                 sumCalculators.push_back(new Chao1());
 329                                         }else if(Estimators[i] == "coverage"){
 330                                                 sumCalculators.push_back(new Coverage());
 331                                         }else if(Estimators[i] == "geometric"){
 332                                                 sumCalculators.push_back(new Geom());
 333                                         }else if(Estimators[i] == "logseries"){
 334                                                 sumCalculators.push_back(new LogSD());
 335                                         }else if(Estimators[i] == "qstat"){
 336                                                 sumCalculators.push_back(new QStat());
 337                                         }else if(Estimators[i] == "bergerparker"){
 338                                                 sumCalculators.push_back(new BergerParker());
 339                                         }else if(Estimators[i] == "bstick"){
 340                                                 sumCalculators.push_back(new BStick());
 341                                         }else if(Estimators[i] == "ace"){
 342                                                 if(abund < 5)
 343                                                         abund = 10;
 344                                                 sumCalculators.push_back(new Ace(abund));
 345                                         }else if(Estimators[i] == "jack"){
 346                                                 sumCalculators.push_back(new Jackknife());
 347                                         }else if(Estimators[i] == "shannon"){
 348                                                 sumCalculators.push_back(new Shannon());
 349                                         }else if(Estimators[i] == "shannoneven"){
 350                                                 sumCalculators.push_back(new ShannonEven());
 351                                         }else if(Estimators[i] == "npshannon"){
 352                                                 sumCalculators.push_back(new NPShannon());
 353                                         }else if(Estimators[i] == "heip"){
 354                                                 sumCalculators.push_back(new Heip());
 355                                         }else if(Estimators[i] == "smithwilson"){
 356                                                 sumCalculators.push_back(new SmithWilson());
 357                                         }else if(Estimators[i] == "simpson"){
 358                                                 sumCalculators.push_back(new Simpson());
 359                                         }else if(Estimators[i] == "simpsoneven"){
 360                                                 sumCalculators.push_back(new SimpsonEven());
 361                                         }else if(Estimators[i] == "invsimpson"){
 362                                                 sumCalculators.push_back(new InvSimpson());
 363                                         }else if(Estimators[i] == "bootstrap"){
 364                                                 sumCalculators.push_back(new Bootstrap());
 365                                         }else if (Estimators[i] == "nseqs") {
 366                                                 sumCalculators.push_back(new NSeqs());
 367                                         }else if (Estimators[i] == "goodscoverage") {
 368                                                 sumCalculators.push_back(new GoodsCoverage());
 369                                         }else if (Estimators[i] == "efron") {
 370                                                 sumCalculators.push_back(new Efron(size));
 371                                         }else if (Estimators[i] == "boneh") {
 372                                                 sumCalculators.push_back(new Boneh(size));
 373                                         }else if (Estimators[i] == "solow") {
 374                                                 sumCalculators.push_back(new Solow(size));
 375                                         }else if (Estimators[i] == "shen") {
 376                                                 sumCalculators.push_back(new Shen(size, abund));
 377                                         }
 378                                 }
 379                         }
 380
 381                         //if the users entered no valid calculators don't execute command
 382                         if (sumCalculators.size() == 0) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);  } return 0; }
 383
 384                         ofstream outputFileHandle;
 385                         m->openOutputFile(fileNameRoot, outputFileHandle);
 386                         outputFileHandle << "label";
 387
 388             ofstream outAve;
 389             if (subsample) {
 390                 m->openOutputFile(fileNameAve, outAve);
 391                 outputNames.push_back(fileNameAve); outputTypes["summary"].push_back(fileNameAve);
 392                 outAve << "label\tmethod";
 393                 outAve.setf(ios::fixed, ios::floatfield); outAve.setf(ios::showpoint);
 394                 if (inputFileNames.size() > 1) {
 395                     groupIndex[fileNameAve] = groups[p];
 396                 }
 397             }
 398
 399                         input = new InputData(inputFileNames[p], format);
 400                         sabund = input->getSAbundVector();
 401                         string lastLabel = sabund->getLabel();
 402
 403                         for(int i=0;i<sumCalculators.size();i++){
 404                                 if(sumCalculators[i]->getCols() == 1){
 405                                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 406                     if (subsample) { outAve << '\t' << sumCalculators[i]->getName();  }
 407                                         numCols++;
 408                                 }
 409                                 else{
 410                                         outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 411                     if (subsample) { outAve << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";  }
 412                                         numCols += 3;
 413                                 }
 414                         }
 415                         outputFileHandle << endl;
 416             if (subsample) {  outAve << endl; }
 417
 418                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 419                         set<string> processedLabels;
 420                         set<string> userLabels = labels;
 421
 422
 423
 424                         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;  }
 425
 426                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 427
 428                                 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;  }
 429
 430                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 431
 432                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 433                                         processedLabels.insert(sabund->getLabel());
 434                                         userLabels.erase(sabund->getLabel());
 435
 436                     process(sabund, outputFileHandle, outAve);
 437
 438                     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;  }
 439                                         numLines++;
 440                                 }
 441
 442                                 if ((m->anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 443                                         string saveLabel = sabund->getLabel();
 444
 445                                         delete sabund;
 446                                         sabund = input->getSAbundVector(lastLabel);
 447
 448                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 449                                         processedLabels.insert(sabund->getLabel());
 450                                         userLabels.erase(sabund->getLabel());
 451
 452                     process(sabund, outputFileHandle, outAve);
 453
 454                     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;  }
 455                                         numLines++;
 456
 457                                         //restore real lastlabel to save below
 458                                         sabund->setLabel(saveLabel);
 459                                 }
 460
 461                                 lastLabel = sabund->getLabel();
 462
 463                                 delete sabund;
 464                                 sabund = input->getSAbundVector();
 465                         }
 466
 467                         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;  }
 468
 469                         //output error messages about any remaining user labels
 470                         set<string>::iterator it;
 471                         bool needToRun = false;
 472                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 473                                 m->mothurOut("Your file does not include the label " + *it);
 474                                 if (processedLabels.count(lastLabel) != 1) {
 475                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 476                                         needToRun = true;
 477                                 }else {
 478                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 479                                 }
 480                         }
 481
 482                         //run last label if you need to
 483                         if (needToRun == true)  {
 484                                 if (sabund != NULL) {   delete sabund;  }
 485                                 sabund = input->getSAbundVector(lastLabel);
 486
 487                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 488                 process(sabund, outputFileHandle, outAve);
 489
 490                 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;  }
 491                                 numLines++;
 492                                 delete sabund;
 493                         }
 494
 495                         outputFileHandle.close();
 496             if (subsample) { outAve.close(); }
 497
 498                         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;  }
 499
 500
 501                         delete input;
 502                         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 503                 }
 504
 505                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  }  return 0;  }
 506
 507                 //create summary file containing all the groups data for each label - this function just combines the info from the files already created.
 508                 if ((sharedfile != "") && (groupMode)) {   vector<string> comboNames = createGroupSummaryFile(numLines, numCols, outputNames, groupIndex);  for (int i = 0; i < comboNames.size(); i++) { outputNames.push_back(comboNames[i]); } }
 509
 510                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);  }  return 0;  }
 511
 512                 m->mothurOutEndLine();
 513                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 514                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 515                 m->mothurOutEndLine();
 516
 517                 return 0;
 518         }
 519         catch(exception& e) {
 520                 m->errorOut(e, "SummaryCommand", "execute");
 521                 exit(1);
 522         }
 523 }
 524 //**********************************************************************************************************************
 525 int SummaryCommand::process(SAbundVector*& sabund, ofstream& outputFileHandle, ofstream& outAve) {
 526     try {
 527
 528         //calculator -> data -> values
 529         vector< vector< vector<double> > >  results; results.resize(sumCalculators.size());
 530
 531         outputFileHandle << sabund->getLabel();
 532
 533         SubSample sample;
 534         for (int thisIter = 0; thisIter < iters+1; thisIter++) {
 535
 536             SAbundVector* thisIterSabund = sabund;
 537
 538             //we want the summary results for the whole dataset, then the subsampling
 539             if ((thisIter > 0) && subsample) { //subsample sabund and run it
 540                 //copy sabund since getSample destroys it
 541                 RAbundVector rabund = sabund->getRAbundVector();
 542                 SAbundVector* newSabund = new SAbundVector();
 543                 *newSabund = rabund.getSAbundVector();
 544
 545                 sample.getSample(newSabund, subsampleSize);
 546                 thisIterSabund = newSabund;
 547             }
 548
 549             for(int i=0;i<sumCalculators.size();i++){
 550                 vector<double> data = sumCalculators[i]->getValues(thisIterSabund);
 551
 552                 if (m->control_pressed) {  return 0;  }
 553
 554                 if (thisIter == 0) {
 555                     outputFileHandle << '\t';
 556                     sumCalculators[i]->print(outputFileHandle);
 557                 }else {
 558                     //some of the calc have hci and lci need to make room for that
 559                     if (results[i].size() == 0) {  results[i].resize(data.size());  }
 560                     //save results for ave and std.
 561                     for (int j = 0; j < data.size(); j++) {
 562                         if (m->control_pressed) {  return 0;  }
 563                         results[i][j].push_back(data[j]);
 564                     }
 565                 }
 566             }
 567
 568             //cleanup memory
 569             if ((thisIter > 0) && subsample) { delete thisIterSabund; }
 570         }
 571         outputFileHandle << endl;
 572
 573         if (subsample) {
 574             outAve << sabund->getLabel() << '\t' << "ave\t";
 575             //find ave and std for this label and output
 576             //will need to modify the createGroupSummary to combine results and not mess with the .summary file.
 577
 578             //calcs -> values
 579             vector< vector<double> >  calcAverages; calcAverages.resize(sumCalculators.size());
 580             for (int i = 0; i < calcAverages.size(); i++) {  calcAverages[i].resize(results[i].size(), 0);  }
 581
 582             for (int thisIter = 0; thisIter < iters; thisIter++) { //sum all groups dists for each calculator
 583                 for (int i = 0; i < calcAverages.size(); i++) {  //initialize sums to zero.
 584                     for (int j = 0; j < calcAverages[i].size(); j++) {
 585                         calcAverages[i][j] += results[i][j][thisIter];
 586                     }
 587                 }
 588             }
 589
 590             for (int i = 0; i < calcAverages.size(); i++) {  //finds average.
 591                 for (int j = 0; j < calcAverages[i].size(); j++) {
 592                     calcAverages[i][j] /= (float) iters;
 593                     outAve << calcAverages[i][j] << '\t';
 594                 }
 595             }
 596
 597             //find standard deviation
 598             vector< vector<double>  > stdDev; stdDev.resize(sumCalculators.size());
 599             for (int i = 0; i < stdDev.size(); i++) {  stdDev[i].resize(results[i].size(), 0);  }
 600
 601             for (int thisIter = 0; thisIter < iters; thisIter++) { //compute the difference of each dist from the mean, and square the result of each
 602                 for (int i = 0; i < stdDev.size(); i++) {
 603                     for (int j = 0; j < stdDev[i].size(); j++) {
 604                         stdDev[i][j] += ((results[i][j][thisIter] - calcAverages[i][j]) * (results[i][j][thisIter] - calcAverages[i][j]));
 605                     }
 606                 }
 607             }
 608
 609             outAve << endl << sabund->getLabel() << '\t' << "std\t";
 610             for (int i = 0; i < stdDev.size(); i++) {  //finds average.
 611                 for (int j = 0; j < stdDev[i].size(); j++) {
 612                     stdDev[i][j] /= (float) iters;
 613                     stdDev[i][j] = sqrt(stdDev[i][j]);
 614                     outAve << stdDev[i][j] << '\t';
 615                 }
 616             }
 617             outAve << endl;
 618         }
 619
 620         return 0;
 621     }
 622     catch(exception& e) {
 623         m->errorOut(e, "SummaryCommand", "process");
 624         exit(1);
 625     }
 626 }
 627 //**********************************************************************************************************************
 628 vector<string> SummaryCommand::parseSharedFile(string filename) {
 629         try {
 630                 vector<string> filenames;
 631
 632                 map<string, ofstream*> filehandles;
 633                 map<string, ofstream*>::iterator it3;
 634
 635                 input = new InputData(filename, "sharedfile");
 636                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 637
 638                 string sharedFileRoot = m->getRootName(filename);
 639
 640         /******************************************************/
 641         if (subsample) {
 642             if (subsampleSize == -1) { //user has not set size, set size = smallest samples size
 643                 subsampleSize = lookup[0]->getNumSeqs();
 644                 for (int i = 1; i < lookup.size(); i++) {
 645                     int thisSize = lookup[i]->getNumSeqs();
 646
 647                     if (thisSize < subsampleSize) {     subsampleSize = thisSize;       }
 648                 }
 649             }else {
 650                 m->clearGroups();
 651                 vector<string> Groups;
 652                 vector<SharedRAbundVector*> temp;
 653                 for (int i = 0; i < lookup.size(); i++) {
 654                     if (lookup[i]->getNumSeqs() < subsampleSize) {
 655                         m->mothurOut(lookup[i]->getGroup() + " contains " + toString(lookup[i]->getNumSeqs()) + ". Eliminating."); m->mothurOutEndLine();
 656                         delete lookup[i];
 657                     }else {
 658                         Groups.push_back(lookup[i]->getGroup());
 659                         temp.push_back(lookup[i]);
 660                     }
 661                 }
 662                 lookup = temp;
 663                 m->setGroups(Groups);
 664             }
 665
 666             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; }
 667         }
 668
 669
 670                 /******************************************************/
 671
 672         //clears file before we start to write to it below
 673                 for (int i=0; i<lookup.size(); i++) {
 674                         m->mothurRemove((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 675                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 676                 }
 677
 678                 ofstream* temp;
 679                 for (int i=0; i<lookup.size(); i++) {
 680                         temp = new ofstream;
 681                         filehandles[lookup[i]->getGroup()] = temp;
 682                         groups.push_back(lookup[i]->getGroup());
 683                 }
 684
 685                 while(lookup[0] != NULL) {
 686
 687                         for (int i = 0; i < lookup.size(); i++) {
 688                                 RAbundVector rav = lookup[i]->getRAbundVector();
 689                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 690                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 691                                 (*(filehandles[lookup[i]->getGroup()])).close();
 692                         }
 693
 694                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 695                         lookup = input->getSharedRAbundVectors();
 696                 }
 697
 698                 //free memory
 699                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 700                         delete it3->second;
 701                 }
 702
 703                 delete input;
 704
 705                 return filenames;
 706         }
 707         catch(exception& e) {
 708                 m->errorOut(e, "SummaryCommand", "parseSharedFile");
 709                 exit(1);
 710         }
 711 }
 712 //**********************************************************************************************************************
 713 vector<string> SummaryCommand::createGroupSummaryFile(int numLines, int numCols, vector<string>& outputNames, map<string, string> groupIndex) {
 714         try {
 715
 716                 //open each groups summary file
 717         vector<string> newComboNames;
 718
 719                 map<string, map<string, vector<string> > > files;
 720         map<string, string> filesTypesLabels;
 721         map<string, int> filesTypesNumLines;
 722                 for (int i=0; i<outputNames.size(); i++) {
 723                         vector<string> thisFilesLines;
 724
 725                         ifstream temp;
 726                         m->openInputFile(outputNames[i], temp);
 727
 728                         //read through first line - labels
 729             string labelsLine = m->getline(temp);
 730             vector<string> theseLabels = m->splitWhiteSpace(labelsLine);
 731
 732             string newLabel = "";
 733             for (int j = 0; j < theseLabels.size(); j++) {
 734                  if (j == 1) {  newLabel += "group\t" + theseLabels[j] + '\t';
 735                 }else{  newLabel += theseLabels[j] + '\t';      }
 736             }
 737
 738                         m->gobble(temp);
 739
 740             int stop = numLines;
 741             if (theseLabels.size() != numCols+1) {  stop = numLines*2; }
 742                         //for each label
 743                         for (int k = 0; k < stop; k++) {
 744
 745                                 string thisLine = "";
 746                                 string tempLabel;
 747
 748                                 for (int j = 0; j < theseLabels.size(); j++) {
 749                                         temp >> tempLabel;
 750
 751                                         //save for later
 752                                         if (j == 1) { thisLine += groupIndex[outputNames[i]] + "\t" + tempLabel + "\t"; }
 753                                         else{  thisLine += tempLabel + "\t";    }
 754                                 }
 755
 756                                 thisLine += "\n";
 757
 758                                 thisFilesLines.push_back(thisLine);
 759
 760                                 m->gobble(temp);
 761                         }
 762
 763             string extension = m->getExtension(outputNames[i]);
 764             if (theseLabels.size() != numCols+1) { extension = ".ave-std" + extension;  }
 765             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 766                         m->mothurRemove(combineFileName); //remove old file
 767             filesTypesLabels[extension] = newLabel;
 768             filesTypesNumLines[extension] = stop;
 769
 770             map<string, map<string, vector<string> > >::iterator itFiles = files.find(extension);
 771             if (itFiles != files.end()) { //add new files info to existing type
 772                 files[extension][outputNames[i]] = thisFilesLines;
 773             }else {
 774                 map<string, vector<string> > thisFile;
 775                 thisFile[outputNames[i]] = thisFilesLines;
 776                 files[extension] = thisFile;
 777             }
 778
 779                         temp.close();
 780                         m->mothurRemove(outputNames[i]);
 781                 }
 782
 783
 784         for (map<string, map<string, vector<string> > >::iterator itFiles = files.begin(); itFiles != files.end(); itFiles++) {
 785
 786             if (m->control_pressed) { break; }
 787
 788             string extension = itFiles->first;
 789             map<string, vector<string> > thisType = itFiles->second;
 790             string combineFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "groups" + extension;
 791             newComboNames.push_back(combineFileName);
 792             //open combined file
 793             ofstream out;
 794             m->openOutputFile(combineFileName, out);
 795
 796             //output label line to new file
 797             out <<  filesTypesLabels[extension] << endl;
 798
 799             //for each label
 800             for (int k = 0; k < filesTypesNumLines[extension]; k++) {
 801
 802                 //grab summary data for each group
 803                 for (map<string, vector<string> >::iterator itType = thisType.begin(); itType != thisType.end(); itType++) {
 804                     out << (itType->second)[k];
 805                 }
 806             }
 807
 808             outputNames.clear();
 809
 810             out.close();
 811         }
 812
 813                 //return combine file name
 814                 return newComboNames;
 815
 816         }
 817         catch(exception& e) {
 818                 m->errorOut(e, "SummaryCommand", "createGroupSummaryFile");
 819                 exit(1);
 820         }
 821 }
 822 //**********************************************************************************************************************