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                 string hadShared = "";
 140                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 141                 else { hadShared = globaldata->getSharedFile(); inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 142
 143                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 144
 145                 for (int p = 0; p < inputFileNames.size(); p++) {
 146
 147                         string fileNameRoot = outputDir + getRootName(getSimpleName(inputFileNames[p])) + "summary";
 148                         globaldata->inputFileName = inputFileNames[p];
 149                         outputNames.push_back(fileNameRoot);
 150
 151                         if (inputFileNames.size() > 1) {
 152                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 153                         }
 154
 155                         sumCalculators.clear();
 156
 157                         validCalculator = new ValidCalculators();
 158
 159                         for (int i=0; i<Estimators.size(); i++) {
 160                                 if (validCalculator->isValidCalculator("summary", Estimators[i]) == true) {
 161                                         if(Estimators[i] == "sobs"){
 162                                                 sumCalculators.push_back(new Sobs());
 163                                         }else if(Estimators[i] == "chao"){
 164                                                 sumCalculators.push_back(new Chao1());
 165                                         }else if(Estimators[i] == "coverage"){
 166                                                 sumCalculators.push_back(new Coverage());
 167                                         }else if(Estimators[i] == "geometric"){
 168                                                 sumCalculators.push_back(new Geom());
 169                                         }else if(Estimators[i] == "logseries"){
 170                                                 sumCalculators.push_back(new LogSD());
 171                                         }else if(Estimators[i] == "qstat"){
 172                                                 sumCalculators.push_back(new QStat());
 173                                         }else if(Estimators[i] == "bergerparker"){
 174                                                 sumCalculators.push_back(new BergerParker());
 175                                         }else if(Estimators[i] == "bstick"){
 176                                                 sumCalculators.push_back(new BStick());
 177                                         }else if(Estimators[i] == "ace"){
 178                                                 if(abund < 5)
 179                                                         abund = 10;
 180                                                 sumCalculators.push_back(new Ace(abund));
 181                                         }else if(Estimators[i] == "jack"){
 182                                                 sumCalculators.push_back(new Jackknife());
 183                                         }else if(Estimators[i] == "shannon"){
 184                                                 sumCalculators.push_back(new Shannon());
 185                                         }else if(Estimators[i] == "npshannon"){
 186                                                 sumCalculators.push_back(new NPShannon());
 187                                         }else if(Estimators[i] == "simpson"){
 188                                                 sumCalculators.push_back(new Simpson());
 189                                         }else if(Estimators[i] == "bootstrap"){
 190                                                 sumCalculators.push_back(new Bootstrap());
 191                                         }else if (Estimators[i] == "nseqs") {
 192                                                 sumCalculators.push_back(new NSeqs());
 193                                         }else if (Estimators[i] == "goodscoverage") {
 194                                                 sumCalculators.push_back(new GoodsCoverage());
 195                                         }else if (Estimators[i] == "efron") {
 196                                                 sumCalculators.push_back(new Efron(size));
 197                                         }else if (Estimators[i] == "boneh") {
 198                                                 sumCalculators.push_back(new Boneh(size));
 199                                         }else if (Estimators[i] == "solow") {
 200                                                 sumCalculators.push_back(new Solow(size));
 201                                         }else if (Estimators[i] == "shen") {
 202                                                 sumCalculators.push_back(new Shen(size, abund));
 203                                         }
 204                                 }
 205                         }
 206
 207                         //if the users entered no valid calculators don't execute command
 208                         if (sumCalculators.size() == 0) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 209
 210                         ofstream outputFileHandle;
 211                         openOutputFile(fileNameRoot, outputFileHandle);
 212                         outputFileHandle << "label";
 213
 214                         read = new ReadOTUFile(globaldata->inputFileName);
 215                         read->read(&*globaldata);
 216
 217                         sabund = globaldata->sabund;
 218                         string lastLabel = sabund->getLabel();
 219                         input = globaldata->ginput;
 220
 221                         for(int i=0;i<sumCalculators.size();i++){
 222                                 if(sumCalculators[i]->getCols() == 1){
 223                                         outputFileHandle << '\t' << sumCalculators[i]->getName();
 224                                 }
 225                                 else{
 226                                         outputFileHandle << '\t' << sumCalculators[i]->getName() << "\t" << sumCalculators[i]->getName() << "_lci\t" << sumCalculators[i]->getName() << "_hci";
 227                                 }
 228                         }
 229                         outputFileHandle << endl;
 230
 231                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 232                         set<string> processedLabels;
 233                         set<string> userLabels = labels;
 234
 235                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }  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;  }
 236
 237                         while((sabund != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 238
 239                                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 240
 241                                 if(allLines == 1 || labels.count(sabund->getLabel()) == 1){
 242
 243                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 244                                         processedLabels.insert(sabund->getLabel());
 245                                         userLabels.erase(sabund->getLabel());
 246
 247                                         outputFileHandle << sabund->getLabel();
 248                                         for(int i=0;i<sumCalculators.size();i++){
 249                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 250
 251                                                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 252
 253                                                 outputFileHandle << '\t';
 254                                                 sumCalculators[i]->print(outputFileHandle);
 255                                         }
 256                                         outputFileHandle << endl;
 257                                 }
 258
 259                                 if ((anyLabelsToProcess(sabund->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 260                                         string saveLabel = sabund->getLabel();
 261
 262                                         delete sabund;
 263                                         sabund = input->getSAbundVector(lastLabel);
 264
 265                                         m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 266                                         processedLabels.insert(sabund->getLabel());
 267                                         userLabels.erase(sabund->getLabel());
 268
 269                                         outputFileHandle << sabund->getLabel();
 270                                         for(int i=0;i<sumCalculators.size();i++){
 271                                                 vector<double> data = sumCalculators[i]->getValues(sabund);
 272
 273                                                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 274
 275                                                 outputFileHandle << '\t';
 276                                                 sumCalculators[i]->print(outputFileHandle);
 277                                         }
 278                                         outputFileHandle << endl;
 279
 280                                         //restore real lastlabel to save below
 281                                         sabund->setLabel(saveLabel);
 282                                 }
 283
 284                                 lastLabel = sabund->getLabel();
 285
 286                                 delete sabund;
 287                                 sabund = input->getSAbundVector();
 288                         }
 289
 290                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 291
 292                         //output error messages about any remaining user labels
 293                         set<string>::iterator it;
 294                         bool needToRun = false;
 295                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 296                                 m->mothurOut("Your file does not include the label " + *it);
 297                                 if (processedLabels.count(lastLabel) != 1) {
 298                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 299                                         needToRun = true;
 300                                 }else {
 301                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 302                                 }
 303                         }
 304
 305                         //run last label if you need to
 306                         if (needToRun == true)  {
 307                                 if (sabund != NULL) {   delete sabund;  }
 308                                 sabund = input->getSAbundVector(lastLabel);
 309
 310                                 m->mothurOut(sabund->getLabel()); m->mothurOutEndLine();
 311                                 outputFileHandle << sabund->getLabel();
 312                                 for(int i=0;i<sumCalculators.size();i++){
 313                                         vector<double> data = sumCalculators[i]->getValues(sabund);
 314
 315                                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 316
 317                                         outputFileHandle << '\t';
 318                                         sumCalculators[i]->print(outputFileHandle);
 319                                 }
 320                                 outputFileHandle << endl;
 321                                 delete sabund;
 322                         }
 323
 324                         outputFileHandle.close();
 325
 326                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } 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;  }
 327
 328
 329                         delete input;  globaldata->ginput = NULL;
 330                         delete read;
 331                         delete validCalculator;
 332                         globaldata->sabund = NULL;
 333                         for(int i=0;i<sumCalculators.size();i++){  delete sumCalculators[i]; }
 334                 }
 335
 336                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 337
 338                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  }  return 0;  }
 339
 340                 m->mothurOutEndLine();
 341                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 342                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 343                 m->mothurOutEndLine();
 344
 345                 return 0;
 346         }
 347         catch(exception& e) {
 348                 m->errorOut(e, "SummaryCommand", "execute");
 349                 exit(1);
 350         }
 351 }
 352 //**********************************************************************************************************************
 353 vector<string> SummaryCommand::parseSharedFile(string filename) {
 354         try {
 355                 vector<string> filenames;
 356
 357                 map<string, ofstream*> filehandles;
 358                 map<string, ofstream*>::iterator it3;
 359
 360
 361                 //read first line
 362                 read = new ReadOTUFile(filename);
 363                 read->read(&*globaldata);
 364
 365                 input = globaldata->ginput;
 366                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 367
 368                 string sharedFileRoot = getRootName(filename);
 369
 370                 //clears file before we start to write to it below
 371                 for (int i=0; i<lookup.size(); i++) {
 372                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 373                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 374                 }
 375
 376                 ofstream* temp;
 377                 for (int i=0; i<lookup.size(); i++) {
 378                         temp = new ofstream;
 379                         filehandles[lookup[i]->getGroup()] = temp;
 380                         groups.push_back(lookup[i]->getGroup());
 381                 }
 382
 383                 while(lookup[0] != NULL) {
 384
 385                         for (int i = 0; i < lookup.size(); i++) {
 386                                 RAbundVector rav = lookup[i]->getRAbundVector();
 387                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 388                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 389                                 (*(filehandles[lookup[i]->getGroup()])).close();
 390                         }
 391
 392                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 393                         lookup = input->getSharedRAbundVectors();
 394                 }
 395
 396                 //free memory
 397                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 398                         delete it3->second;
 399                 }
 400                 delete read;
 401                 delete input;
 402                 globaldata->ginput = NULL;
 403
 404                 return filenames;
 405         }
 406         catch(exception& e) {
 407                 m->errorOut(e, "SummaryCommand", "parseSharedFile");
 408                 exit(1);
 409         }
 410 }
 411 //**********************************************************************************************************************