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 "npshannon.h"
  18 #include "shannon.h"
  19 #include "jackknife.h"
  20 #include "geom.h"
  21 #include "logsd.h"
  22 #include "qstat.h"
  23 #include "bergerparker.h"
  24 #include "bstick.h"
  25 #include "goodscoverage.h"
  26 #include "coverage.h"
  27 #include "efron.h"
  28 #include "boneh.h"
  29 #include "solow.h"
  30 #include "shen.h"
  31
  32 //**********************************************************************************************************************
  33
  34 SummaryCommand::SummaryCommand(string option)  {
  35         try {
  36                 globaldata = GlobalData::getInstance();
  37                 abort = false;
  38                 allLines = 1;
  39                 labels.clear();
  40                 Estimators.clear();
  41
  42                 //allow user to run help
  43                 if(option == "help") { validCalculator = new ValidCalculators(); help(); delete validCalculator; abort = true; }
  44
  45                 else {
  46                         //valid paramters for this command
  47                         string Array[] =  {"label","calc","abund","size","outputdir","inputdir"};
  48                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49
  50                         OptionParser parser(option);
  51                         map<string,string> parameters = parser.getParameters();
  52
  53                         ValidParameters validParameter;
  54
  55                         //check to make sure all parameters are valid for command
  56                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  57                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  58                         }
  59
  60                         //make sure the user has already run the read.otu command
  61                         if ((globaldata->getSharedFile() == "") && (globaldata->getListFile() == "") && (globaldata->getRabundFile() == "") && (globaldata->getSabundFile() == "")) { m->mothurOut("You must read a list, sabund, rabund or shared file before you can use the summary.single command."); m->mothurOutEndLine(); abort = true; }
  62
  63                         //if the user changes the output directory command factory will send this info to us in the output parameter
  64                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  65                                 outputDir = "";
  66                                 outputDir += hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
  67                         }
  68
  69                         //check for optional parameter and set defaults
  70                         // ...at some point should added some additional type checking...
  71                         label = validParameter.validFile(parameters, "label", false);
  72                         if (label == "not found") { label = ""; }
  73                         else {
  74                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  75                                 else { allLines = 1;  }
  76                         }
  77
  78                         //if the user has not specified any labels use the ones from read.otu
  79                         if(label == "") {
  80                                 allLines = globaldata->allLines;
  81                                 labels = globaldata->labels;
  82                         }
  83
  84                         calc = validParameter.validFile(parameters, "calc", false);
  85                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  86                         else {
  87                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  88                         }
  89                         splitAtDash(calc, Estimators);
  90
  91                         string temp;
  92                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
  93                         convert(temp, abund);
  94
  95                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
  96                         convert(temp, size);
  97
  98                 }
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "SummaryCommand", "SummaryCommand");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106
 107 void SummaryCommand::help(){
 108         try {
 109                 m->mothurOut("The summary.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION.\n");
 110                 m->mothurOut("The summary.single command can be executed after a successful cluster command.  It will use the .list file from the output of the cluster.\n");
 111                 m->mothurOut("The summary.single command parameters are label, calc, abund.  No parameters are required.\n");
 112                 m->mothurOut("The summary.single command should be in the following format: \n");
 113                 m->mothurOut("summary.single(label=yourLabel, calc=yourEstimators).\n");
 114                 m->mothurOut("Example summary.single(label=unique-.01-.03, calc=sobs-chao-ace-jack-bootstrap-shannon-npshannon-simpson).\n");
 115                 validCalculator->printCalc("summary", cout);
 116                 m->mothurOut("The default value calc is sobs-chao-ace-jack-shannon-npshannon-simpson\n");
 117                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 118                 m->mothurOut("Note: No spaces between parameter labels (i.e. label), '=' and parameters (i.e.yourLabels).\n\n");
 119         }
 120         catch(exception& e) {
 121                 m->errorOut(e, "SummaryCommand", "help");
 122                 exit(1);
 123         }
 124 }
 125
 126 //**********************************************************************************************************************
 127
 128 SummaryCommand::~SummaryCommand(){}
 129
 130 //**********************************************************************************************************************
 131
 132 int SummaryCommand::execute(){
 133         try {
 134
 135                 if (abort == true) { return 0; }
 136
 137                 vector<string> outputNames;
 138
 139                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 140                 else {  inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 141
 142                 if (m->control_pressed) { return 0; }
 143
 144                 for (int p = 0; p < inputFileNames.size(); p++) {
 145
 146                         string fileNameRoot = outputDir + getRootName(getSimpleName(inputFileNames[p])) + "summary";
 147                         globaldata->inputFileName = inputFileNames[p];
 148                         outputNames.push_back(fileNameRoot);
 149
 150                         if (inputFileNames.size() > 1) {
 151                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 152                         }
 153
 154                         sumCalculators.clear();
 155
 156                         validCalculator = new ValidCalculators();
 157
 158                         for (int i=0; i<Estimators.size(); i++) {
 159                                 if (validCalculator->isValidCalculator("summary", Estimators[i]) == true) {
 160                                         if(Estimators[i] == "sobs"){
 161                                                 sumCalculators.push_back(new Sobs());
 162                                         }else if(Estimators[i] == "chao"){
 163                                                 sumCalculators.push_back(new Chao1());
 164                                         }else if(Estimators[i] == "coverage"){
 165                                                 sumCalculators.push_back(new Coverage());
 166                                         }else if(Estimators[i] == "geometric"){
 167                                                 sumCalculators.push_back(new Geom());
 168                                         }else if(Estimators[i] == "logseries"){
 169                                                 sumCalculators.push_back(new LogSD());
 170                                         }else if(Estimators[i] == "qstat"){
 171                                                 sumCalculators.push_back(new QStat());
 172                                         }else if(Estimators[i] == "bergerparker"){
 173                                                 sumCalculators.push_back(new BergerParker());
 174                                         }else if(Estimators[i] == "bstick"){
 175                                                 sumCalculators.push_back(new BStick());
 176                                         }else if(Estimators[i] == "ace"){
 177                                                 if(abund < 5)
 178                                                         abund = 10;
 179                                                 sumCalculators.push_back(new Ace(abund));
 180                                         }else if(Estimators[i] == "jack"){
 181                                                 sumCalculators.push_back(new Jackknife());
 182                                         }else if(Estimators[i] == "shannon"){
 183                                                 sumCalculators.push_back(new Shannon());
 184                                         }else if(Estimators[i] == "npshannon"){
 185                                                 sumCalculators.push_back(new NPShannon());
 186                                         }else if(Estimators[i] == "simpson"){
 187                                                 sumCalculators.push_back(new Simpson());
 188                                         }else if(Estimators[i] == "bootstrap"){
 189                                                 sumCalculators.push_back(new Bootstrap());
 190                                         }else if (Estimators[i] == "nseqs") {
 191                                                 sumCalculators.push_back(new NSeqs());
 192                                         }else if (Estimators[i] == "goodscoverage") {
 193                                                 sumCalculators.push_back(new GoodsCoverage());
 194                                         }else if (Estimators[i] == "efron") {
 195                                                 sumCalculators.push_back(new Efron(size));
 196                                         }else if (Estimators[i] == "boneh") {
 197                                                 sumCalculators.push_back(new Boneh(size));
 198                                         }else if (Estimators[i] == "solow") {
 199                                                 sumCalculators.push_back(new Solow(size));
 200                                         }else if (Estimators[i] == "shen") {
 201                                                 sumCalculators.push_back(new Shen(size, abund));
 202                                         }
 203                                 }
 204                         }
 205
 206                         //if the users entered no valid calculators don't execute command
 207                         if (sumCalculators.size() == 0) { return 0; }
 208
 209                         ofstream outputFileHandle;
 210                         openOutputFile(fileNameRoot, outputFileHandle);
 211                         outputFileHandle << "label";
 212
 213                         read = new ReadOTUFile(globaldata->inputFileName);
 214                         read->read(&*globaldata);
 215
 216                         sabund = globaldata->sabund;
 217                         string lastLabel = sabund->getLabel();
 218                         input = globaldata->ginput;
 219
 220                         for(int i=0;i<sumCalculators.size();i++){
 221                                 if(sumCalculators[i]->getCols() == 1){
 222                                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 223                                 }
 224                                 else{
 225                                         outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 226                                 }
 227                         }
 228                         outputFileHandle << endl;
 229
 230                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 231                         set<string> processedLabels;
 232                         set<string> userLabels = labels;
 233
 234                         if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read; delete sabund; globaldata->sabund = NULL; delete input; globaldata->ginput = NULL; return 0;  }
 235
 236                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 237
 238                                 if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read; delete sabund; globaldata->sabund = NULL; delete input; globaldata->ginput = NULL; return 0;  }
 239
 240                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 241
 242                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 243                                         processedLabels.insert(sabund->getLabel());
 244                                         userLabels.erase(sabund->getLabel());
 245
 246                                         outputFileHandle << sabund->getLabel();
 247                                         for(int i=0;i<sumCalculators.size();i++){
 248                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 249
 250                                                 if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read; delete sabund; globaldata->sabund = NULL; delete input; globaldata->ginput = NULL; return 0;  }
 251
 252                                                 outputFileHandle << '\t';
 253                                                 sumCalculators[i]->print(outputFileHandle);
 254                                         }
 255                                         outputFileHandle << endl;
 256                                 }
 257
 258                                 if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 259                                         string saveLabel = sabund->getLabel();
 260
 261                                         delete sabund;
 262                                         sabund = input->getSAbundVector(lastLabel);
 263
 264                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 265                                         processedLabels.insert(sabund->getLabel());
 266                                         userLabels.erase(sabund->getLabel());
 267
 268                                         outputFileHandle << sabund->getLabel();
 269                                         for(int i=0;i<sumCalculators.size();i++){
 270                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 271
 272                                                 if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read; delete sabund; globaldata->sabund = NULL; delete input; globaldata->ginput = NULL; return 0;  }
 273
 274                                                 outputFileHandle << '\t';
 275                                                 sumCalculators[i]->print(outputFileHandle);
 276                                         }
 277                                         outputFileHandle << endl;
 278
 279                                         //restore real lastlabel to save below
 280                                         sabund->setLabel(saveLabel);
 281                                 }
 282
 283                                 lastLabel = sabund->getLabel();
 284
 285                                 delete sabund;
 286                                 sabund = input->getSAbundVector();
 287                         }
 288
 289                         if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read;  delete input; globaldata->ginput = NULL; return 0;  }
 290
 291                         //output error messages about any remaining user labels
 292                         set<string>::iterator it;
 293                         bool needToRun = false;
 294                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 295                                 m->mothurOut("Your file does not include the label " + *it);
 296                                 if (processedLabels.count(lastLabel) != 1) {
 297                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 298                                         needToRun = true;
 299                                 }else {
 300                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 301                                 }
 302                         }
 303
 304                         //run last label if you need to
 305                         if (needToRun == true)  {
 306                                 if (sabund != NULL) {   delete sabund;  }
 307                                 sabund = input->getSAbundVector(lastLabel);
 308
 309                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 310                                 outputFileHandle << sabund->getLabel();
 311                                 for(int i=0;i<sumCalculators.size();i++){
 312                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 313
 314                                         if (m->control_pressed) { outputFileHandle.close(); for (int i = 0; i < outputNames.size(); i++) {      remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read; delete sabund; globaldata->sabund = NULL; delete input; globaldata->ginput = NULL; return 0;  }
 315
 316                                         outputFileHandle << '\t';
 317                                         sumCalculators[i]->print(outputFileHandle);
 318                                 }
 319                                 outputFileHandle << endl;
 320                                 delete sabund;
 321                         }
 322
 323                         outputFileHandle.close();
 324
 325                         if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; } delete validCalculator; delete read;  delete input; globaldata->ginput = NULL; return 0;  }
 326
 327
 328                         delete input;  globaldata->ginput = NULL;
 329                         delete read;
 330                         delete validCalculator;
 331                         globaldata->sabund = NULL;
 332                         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 333                 }
 334
 335                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }  return 0;  }
 336
 337                 m->mothurOutEndLine();
 338                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 339                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 340                 m->mothurOutEndLine();
 341
 342                 return 0;
 343         }
 344         catch(exception& e) {
 345                 m->errorOut(e, "SummaryCommand", "execute");
 346                 exit(1);
 347         }
 348 }
 349 //**********************************************************************************************************************
 350 vector<string> SummaryCommand::parseSharedFile(string filename) {
 351         try {
 352                 vector<string> filenames;
 353
 354                 map<string, ofstream*> filehandles;
 355                 map<string, ofstream*>::iterator it3;
 356
 357
 358                 //read first line
 359                 read = new ReadOTUFile(filename);
 360                 read->read(&*globaldata);
 361
 362                 input = globaldata->ginput;
 363                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 364
 365                 string sharedFileRoot = getRootName(filename);
 366
 367                 //clears file before we start to write to it below
 368                 for (int i=0; i<lookup.size(); i++) {
 369                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 370                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 371                 }
 372
 373                 ofstream* temp;
 374                 for (int i=0; i<lookup.size(); i++) {
 375                         temp = new ofstream;
 376                         filehandles[lookup[i]->getGroup()] = temp;
 377                         groups.push_back(lookup[i]->getGroup());
 378                 }
 379
 380                 while(lookup[0] != NULL) {
 381
 382                         for (int i = 0; i < lookup.size(); i++) {
 383                                 RAbundVector rav = lookup[i]->getRAbundVector();
 384                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 385                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 386                                 (*(filehandles[lookup[i]->getGroup()])).close();
 387                         }
 388
 389                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 390                         lookup = input->getSharedRAbundVectors();
 391                 }
 392
 393                 //free memory
 394                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 395                         delete it3->second;
 396                 }
 397                 delete read;
 398                 delete input;
 399                 globaldata->ginput = NULL;
 400
 401                 return filenames;
 402         }
 403         catch(exception& e) {
 404                 m->errorOut(e, "SummaryCommand", "parseSharedFile");
 405                 exit(1);
 406         }
 407 }
 408 //**********************************************************************************************************************