collectcommand.cpp

   1 /*
   2  *  collectcommand.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 "collectcommand.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 "qstat.h"
  22 #include "logsd.h"
  23 #include "bergerparker.h"
  24 #include "bstick.h"
  25 #include "goodscoverage.h"
  26 #include "efron.h"
  27 #include "boneh.h"
  28 #include "solow.h"
  29 #include "shen.h"
  30 #include "coverage.h"
  31
  32
  33 //**********************************************************************************************************************
  34 CollectCommand::CollectCommand(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[] =  {"freq","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                         //if the user changes the output directory command factory will send this info to us in the output parameter
  61                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  62
  63                         //make sure the user has already run the read.otu command
  64                         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 collect.single command."); m->mothurOutEndLine(); abort = true; }
  65
  66                         //check for optional parameter and set defaults
  67                         // ...at some point should added some additional type checking...
  68                         label = validParameter.validFile(parameters, "label", false);
  69                         if (label == "not found") { label = ""; }
  70                         else {
  71                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  72                                 else { allLines = 1;  }
  73                         }
  74
  75                         //if the user has not specified any labels use the ones from read.otu
  76                         if (label == "") {
  77                                 allLines = globaldata->allLines;
  78                                 labels = globaldata->labels;
  79                         }
  80
  81                         calc = validParameter.validFile(parameters, "calc", false);
  82                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  83                         else {
  84                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  85                         }
  86                         splitAtDash(calc, Estimators);
  87
  88                         string temp;
  89                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
  90                         convert(temp, freq);
  91
  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, "CollectCommand", "CollectCommand");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106
 107 void CollectCommand::help(){
 108         try {
 109                 m->mothurOut("The collect.single command can only be executed after a successful read.otu command. WITH ONE EXECEPTION. \n");
 110                 m->mothurOut("The collect.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 collect.single command parameters are label, freq, calc and abund.  No parameters are required. \n");
 112                 m->mothurOut("The collect.single command should be in the following format: \n");
 113                 m->mothurOut("The freq parameter is used indicate when to output your data, by default it is set to 100. But you can set it to a percentage of the number of sequence. For example freq=0.10, means 10%. \n");
 114                 m->mothurOut("collect.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n");
 115                 m->mothurOut("Example collect(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-chao-ace-jack).\n");
 116                 m->mothurOut("The default values for freq is 100, and calc are sobs-chao-ace-jack-shannon-npshannon-simpson.\n");
 117                 validCalculator->printCalc("single", cout);
 118                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 119                 m->mothurOut("Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n\n");
 120         }
 121         catch(exception& e) {
 122                 m->errorOut(e, "CollectCommand", "help");
 123                 exit(1);
 124         }
 125 }
 126
 127 //**********************************************************************************************************************
 128
 129 CollectCommand::~CollectCommand(){}
 130
 131 //**********************************************************************************************************************
 132
 133 int CollectCommand::execute(){
 134         try {
 135
 136                 if (abort == true) { return 0; }
 137
 138                 vector<string> outputNames;
 139
 140                 string hadShared = "";
 141                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 142                 else { hadShared = globaldata->getSharedFile(); inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 143
 144                 for (int p = 0; p < inputFileNames.size(); p++) {
 145
 146                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         }  globaldata->Groups.clear(); if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 147
 148                         if (outputDir == "") { outputDir += hasPath(inputFileNames[p]); }
 149                         string fileNameRoot = outputDir + getRootName(getSimpleName(inputFileNames[p]));
 150                         globaldata->inputFileName = inputFileNames[p];
 151
 152                         if (inputFileNames.size() > 1) {
 153                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 154                         }
 155
 156                         validCalculator = new ValidCalculators();
 157
 158                         for (int i=0; i<Estimators.size(); i++) {
 159                                 if (validCalculator->isValidCalculator("single", Estimators[i]) == true) {
 160                                         if (Estimators[i] == "sobs") {
 161                                                 cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+"sobs")));
 162                                                 outputNames.push_back(fileNameRoot+"sobs");
 163                                         }else if (Estimators[i] == "chao") {
 164                                                 cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"chao")));
 165                                                 outputNames.push_back(fileNameRoot+"chao");
 166                                         }else if (Estimators[i] == "nseqs") {
 167                                                 cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+"nseqs")));
 168                                                 outputNames.push_back(fileNameRoot+"nseqs");
 169                                         }else if (Estimators[i] == "coverage") {
 170                                                 cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+"coverage")));
 171                                                 outputNames.push_back(fileNameRoot+"coverage");
 172                                         }else if (Estimators[i] == "ace") {
 173                                                 cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"ace")));
 174                                                 outputNames.push_back(fileNameRoot+"ace");
 175                                         }else if (Estimators[i] == "jack") {
 176                                                 cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"jack")));
 177                                                 outputNames.push_back(fileNameRoot+"jack");
 178                                         }else if (Estimators[i] == "shannon") {
 179                                                 cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"shannon")));
 180                                                 outputNames.push_back(fileNameRoot+"shannon");
 181                                         }else if (Estimators[i] == "npshannon") {
 182                                                 cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+"np_shannon")));
 183                                                 outputNames.push_back(fileNameRoot+"np_shannon");
 184                                         }else if (Estimators[i] == "simpson") {
 185                                                 cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"simpson")));
 186                                                 outputNames.push_back(fileNameRoot+"simpson");
 187                                         }else if (Estimators[i] == "bootstrap") {
 188                                                 cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+"bootstrap")));
 189                                                 outputNames.push_back(fileNameRoot+"bootstrap");
 190                                         }else if (Estimators[i] == "geometric") {
 191                                                 cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+"geometric")));
 192                                                 outputNames.push_back(fileNameRoot+"geometric");
 193                                         }else if (Estimators[i] == "qstat") {
 194                                                 cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+"qstat")));
 195                                                 outputNames.push_back(fileNameRoot+"qstat");
 196                                         }else if (Estimators[i] == "logseries") {
 197                                                 cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+"logseries")));
 198                                                 outputNames.push_back(fileNameRoot+"logseries");
 199                                         }else if (Estimators[i] == "bergerparker") {
 200                                                 cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+"bergerparker")));
 201                                                 outputNames.push_back(fileNameRoot+"bergerparker");
 202                                         }else if (Estimators[i] == "bstick") {
 203                                                 cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+"bstick")));
 204                                                 outputNames.push_back(fileNameRoot+"bstick");
 205                                         }else if (Estimators[i] == "goodscoverage") {
 206                                                 cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+"goodscoverage")));
 207                                                 outputNames.push_back(fileNameRoot+"goodscoverage");
 208                                         }else if (Estimators[i] == "efron") {
 209                                                 cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+"efron")));
 210                                                 outputNames.push_back(fileNameRoot+"efron");
 211                                         }else if (Estimators[i] == "boneh") {
 212                                                 cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+"boneh")));
 213                                                 outputNames.push_back(fileNameRoot+"boneh");
 214                                         }else if (Estimators[i] == "solow") {
 215                                                 cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+"solow")));
 216                                                 outputNames.push_back(fileNameRoot+"solow");
 217                                         }else if (Estimators[i] == "shen") {
 218                                                 cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+"shen")));
 219                                                 outputNames.push_back(fileNameRoot+"shen");
 220                                         }
 221                                 }
 222                         }
 223
 224                         //if the users entered no valid calculators don't execute command
 225                         if (cDisplays.size() == 0) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 226
 227                         read = new ReadOTUFile(inputFileNames[p]);
 228                         read->read(&*globaldata);
 229
 230                         order = globaldata->gorder;
 231                         string lastLabel = order->getLabel();
 232                         input = globaldata->ginput;
 233
 234                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 235                         set<string> processedLabels;
 236                         set<string> userLabels = labels;
 237
 238                         if (m->control_pressed) {
 239                                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 240                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 241                                 delete input;  globaldata->ginput = NULL;
 242                                 delete read;
 243                                 delete order; globaldata->gorder = NULL;
 244                                 delete validCalculator;
 245                                 globaldata->Groups.clear();
 246                                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 247                                 return 0;
 248                         }
 249
 250
 251                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 252
 253                                 if (m->control_pressed) {
 254                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 255                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 256                                         delete input;  globaldata->ginput = NULL;
 257                                         delete read;
 258                                         delete order; globaldata->gorder = NULL;
 259                                         delete validCalculator;
 260                                         globaldata->Groups.clear();
 261                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 262                                         return 0;
 263                                 }
 264
 265
 266                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 267
 268                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 269                                         cCurve = new Collect(order, cDisplays);
 270                                         cCurve->getCurve(freq);
 271                                         delete cCurve;
 272
 273                                         processedLabels.insert(order->getLabel());
 274                                         userLabels.erase(order->getLabel());
 275
 276
 277                                 }
 278                                 //you have a label the user want that is smaller than this label and the last label has not already been processed
 279                                 if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 280                                         string saveLabel = order->getLabel();
 281
 282                                         delete order;
 283                                         order = (input->getOrderVector(lastLabel));
 284
 285                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 286                                         cCurve = new Collect(order, cDisplays);
 287                                         cCurve->getCurve(freq);
 288                                         delete cCurve;
 289
 290
 291                                         processedLabels.insert(order->getLabel());
 292                                         userLabels.erase(order->getLabel());
 293
 294                                         //restore real lastlabel to save below
 295                                         order->setLabel(saveLabel);
 296                                 }
 297
 298                                 lastLabel = order->getLabel();
 299
 300                                 delete order;
 301                                 order = (input->getOrderVector());
 302                         }
 303
 304
 305                         if (m->control_pressed) {
 306                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 307                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 308                                         delete input;  globaldata->ginput = NULL;
 309                                         delete read;
 310                                         delete validCalculator;
 311                                         globaldata->Groups.clear();
 312                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 313                                         return 0;
 314                         }
 315
 316                         //output error messages about any remaining user labels
 317                         set<string>::iterator it;
 318                         bool needToRun = false;
 319                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 320                                 m->mothurOut("Your file does not include the label " + *it);
 321                                 if (processedLabels.count(lastLabel) != 1) {
 322                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 323                                         needToRun = true;
 324                                 }else {
 325                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 326                                 }
 327                         }
 328
 329                         //run last label if you need to
 330                         if (needToRun == true)  {
 331                                 if (order != NULL) {    delete order;   }
 332                                 order = (input->getOrderVector(lastLabel));
 333
 334                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 335
 336                                 cCurve = new Collect(order, cDisplays);
 337                                 cCurve->getCurve(freq);
 338                                 delete cCurve;
 339
 340                                 if (m->control_pressed) {
 341                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 342                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 343                                         delete input;  globaldata->ginput = NULL;
 344                                         delete read;
 345                                         delete order; globaldata->gorder = NULL;
 346                                         delete validCalculator;
 347                                         globaldata->Groups.clear();
 348                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 349                                         return 0;
 350                                 }
 351                                 delete order;
 352                         }
 353
 354                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 355                         cDisplays.clear();
 356                         delete input;  globaldata->ginput = NULL;
 357                         delete read;
 358                         globaldata->gorder = NULL;
 359                         delete validCalculator;
 360                 }
 361
 362                 //return to shared mode if you changed above
 363                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 364
 365                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 366
 367                 m->mothurOutEndLine();
 368                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 369                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 370                 m->mothurOutEndLine();
 371
 372
 373                 return 0;
 374         }
 375         catch(exception& e) {
 376                 m->errorOut(e, "CollectCommand", "execute");
 377                 exit(1);
 378         }
 379 }
 380
 381 //**********************************************************************************************************************
 382 vector<string> CollectCommand::parseSharedFile(string filename) {
 383         try {
 384                 vector<string> filenames;
 385
 386                 map<string, ofstream*> filehandles;
 387                 map<string, ofstream*>::iterator it3;
 388
 389
 390                 //read first line
 391                 read = new ReadOTUFile(filename);
 392                 read->read(&*globaldata);
 393
 394                 input = globaldata->ginput;
 395                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 396
 397                 string sharedFileRoot = getRootName(filename);
 398
 399                 //clears file before we start to write to it below
 400                 for (int i=0; i<lookup.size(); i++) {
 401                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 402                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 403                 }
 404
 405                 ofstream* temp;
 406                 for (int i=0; i<lookup.size(); i++) {
 407                         temp = new ofstream;
 408                         filehandles[lookup[i]->getGroup()] = temp;
 409                         groups.push_back(lookup[i]->getGroup());
 410                 }
 411
 412                 while(lookup[0] != NULL) {
 413
 414                         for (int i = 0; i < lookup.size(); i++) {
 415                                 RAbundVector rav = lookup[i]->getRAbundVector();
 416                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 417                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 418                                 (*(filehandles[lookup[i]->getGroup()])).close();
 419                         }
 420
 421                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 422                         lookup = input->getSharedRAbundVectors();
 423                 }
 424
 425                 //free memory
 426                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 427                         delete it3->second;
 428                 }
 429                 delete read;
 430                 delete input;
 431                 globaldata->ginput = NULL;
 432
 433                 return filenames;
 434         }
 435         catch(exception& e) {
 436                 m->errorOut(e, "CollectCommand", "parseSharedFile");
 437                 exit(1);
 438         }
 439 }
 440 //**********************************************************************************************************************
 441