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 "simpsoneven.h"
  18 #include "invsimpson.h"
  19 #include "npshannon.h"
  20 #include "shannon.h"
  21 #include "smithwilson.h"
  22 #include "heip.h"
  23 #include "shannoneven.h"
  24 #include "jackknife.h"
  25 #include "geom.h"
  26 #include "qstat.h"
  27 #include "logsd.h"
  28 #include "bergerparker.h"
  29 #include "bstick.h"
  30 #include "goodscoverage.h"
  31 #include "efron.h"
  32 #include "boneh.h"
  33 #include "solow.h"
  34 #include "shen.h"
  35 #include "coverage.h"
  36
  37
  38 //**********************************************************************************************************************
  39 CollectCommand::CollectCommand(string option)  {
  40         try {
  41                 globaldata = GlobalData::getInstance();
  42                 abort = false;
  43                 allLines = 1;
  44                 labels.clear();
  45                 Estimators.clear();
  46
  47                 //allow user to run help
  48                 if(option == "help") { validCalculator = new ValidCalculators(); help(); delete validCalculator; abort = true; }
  49
  50                 else {
  51                         //valid paramters for this command
  52                         string Array[] =  {"freq","label","calc","abund","size","outputdir","inputdir"};
  53                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  54
  55                         OptionParser parser(option);
  56                         map<string,string> parameters = parser.getParameters();
  57
  58                         ValidParameters validParameter;
  59
  60                         //check to make sure all parameters are valid for command
  61                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  62                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  63                         }
  64
  65                         //if the user changes the output directory command factory will send this info to us in the output parameter
  66                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  67
  68                         //make sure the user has already run the read.otu command
  69                         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; }
  70
  71                         //check for optional parameter and set defaults
  72                         // ...at some point should added some additional type checking...
  73                         label = validParameter.validFile(parameters, "label", false);
  74                         if (label == "not found") { label = ""; }
  75                         else {
  76                                 if(label != "all") {  splitAtDash(label, labels);  allLines = 0;  }
  77                                 else { allLines = 1;  }
  78                         }
  79
  80                         //if the user has not specified any labels use the ones from read.otu
  81                         if (label == "") {
  82                                 allLines = globaldata->allLines;
  83                                 labels = globaldata->labels;
  84                         }
  85
  86                         calc = validParameter.validFile(parameters, "calc", false);
  87                         if (calc == "not found") { calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  88                         else {
  89                                  if (calc == "default")  {  calc = "sobs-chao-ace-jack-shannon-npshannon-simpson";  }
  90                         }
  91                         splitAtDash(calc, Estimators);
  92
  93                         string temp;
  94                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
  95                         convert(temp, freq);
  96
  97                         temp = validParameter.validFile(parameters, "abund", false);            if (temp == "not found") { temp = "10"; }
  98                         convert(temp, abund);
  99
 100                         temp = validParameter.validFile(parameters, "size", false);                     if (temp == "not found") { temp = "0"; }
 101                         convert(temp, size);
 102                 }
 103
 104         }
 105         catch(exception& e) {
 106                 m->errorOut(e, "CollectCommand", "CollectCommand");
 107                 exit(1);
 108         }
 109 }
 110 //**********************************************************************************************************************
 111
 112 void CollectCommand::help(){
 113         try {
 114                 m->mothurOut("The collect.single command can only be executed after a successful read.otu command. WITH ONE EXECEPTION. \n");
 115                 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");
 116                 m->mothurOut("The collect.single command parameters are label, freq, calc and abund.  No parameters are required. \n");
 117                 m->mothurOut("The collect.single command should be in the following format: \n");
 118                 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");
 119                 m->mothurOut("collect.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n");
 120                 m->mothurOut("Example collect(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-chao-ace-jack).\n");
 121                 m->mothurOut("The default values for freq is 100, and calc are sobs-chao-ace-jack-shannon-npshannon-simpson.\n");
 122                 validCalculator->printCalc("single", cout);
 123                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 124                 m->mothurOut("Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n\n");
 125         }
 126         catch(exception& e) {
 127                 m->errorOut(e, "CollectCommand", "help");
 128                 exit(1);
 129         }
 130 }
 131
 132 //**********************************************************************************************************************
 133
 134 CollectCommand::~CollectCommand(){}
 135
 136 //**********************************************************************************************************************
 137
 138 int CollectCommand::execute(){
 139         try {
 140
 141                 if (abort == true) { return 0; }
 142
 143                 vector<string> outputNames;
 144
 145                 string hadShared = "";
 146                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 147                 else { hadShared = globaldata->getSharedFile(); inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 148
 149                 for (int p = 0; p < inputFileNames.size(); p++) {
 150
 151                         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; }
 152
 153                         if (outputDir == "") { outputDir += hasPath(inputFileNames[p]); }
 154                         string fileNameRoot = outputDir + getRootName(getSimpleName(inputFileNames[p]));
 155                         globaldata->inputFileName = inputFileNames[p];
 156
 157                         if (inputFileNames.size() > 1) {
 158                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 159                         }
 160
 161                         validCalculator = new ValidCalculators();
 162
 163                         for (int i=0; i<Estimators.size(); i++) {
 164                                 if (validCalculator->isValidCalculator("single", Estimators[i]) == true) {
 165                                         if (Estimators[i] == "sobs") {
 166                                                 cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+"sobs")));
 167                                                 outputNames.push_back(fileNameRoot+"sobs");
 168                                         }else if (Estimators[i] == "chao") {
 169                                                 cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"chao")));
 170                                                 outputNames.push_back(fileNameRoot+"chao");
 171                                         }else if (Estimators[i] == "nseqs") {
 172                                                 cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+"nseqs")));
 173                                                 outputNames.push_back(fileNameRoot+"nseqs");
 174                                         }else if (Estimators[i] == "coverage") {
 175                                                 cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+"coverage")));
 176                                                 outputNames.push_back(fileNameRoot+"coverage");
 177                                         }else if (Estimators[i] == "ace") {
 178                                                 cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"ace")));
 179                                                 outputNames.push_back(fileNameRoot+"ace");
 180                                         }else if (Estimators[i] == "jack") {
 181                                                 cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"jack")));
 182                                                 outputNames.push_back(fileNameRoot+"jack");
 183                                         }else if (Estimators[i] == "shannon") {
 184                                                 cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"shannon")));
 185                                                 outputNames.push_back(fileNameRoot+"shannon");
 186                                         }else if (Estimators[i] == "shannoneven") {
 187                                                 cDisplays.push_back(new CollectDisplay(new ShannonEven(), new OneColumnFile(fileNameRoot+"shannoneven")));
 188                                                 outputNames.push_back(fileNameRoot+"shannoneven");
 189                                         }else if (Estimators[i] == "npshannon") {
 190                                                 cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+"np_shannon")));
 191                                                 outputNames.push_back(fileNameRoot+"np_shannon");
 192                                         }else if (Estimators[i] == "heip") {
 193                                                 cDisplays.push_back(new CollectDisplay(new Heip(), new OneColumnFile(fileNameRoot+"heip")));
 194                                                 outputNames.push_back(fileNameRoot+"heip");
 195                                         }else if (Estimators[i] == "smithwilson") {
 196                                                 cDisplays.push_back(new CollectDisplay(new SmithWilson(), new OneColumnFile(fileNameRoot+"smithwilson")));
 197                                                 outputNames.push_back(fileNameRoot+"smithwilson");
 198                                         }else if (Estimators[i] == "simpson") {
 199                                                 cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"simpson")));
 200                                                 outputNames.push_back(fileNameRoot+"simpson");
 201                                         }else if (Estimators[i] == "simpsoneven") {
 202                                                 cDisplays.push_back(new CollectDisplay(new SimpsonEven(), new OneColumnFile(fileNameRoot+"simpsoneven")));
 203                                                 outputNames.push_back(fileNameRoot+"simpsoneven");
 204                                         }else if (Estimators[i] == "invsimpson") {
 205                                                 cDisplays.push_back(new CollectDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+"invsimpson")));
 206                                                 outputNames.push_back(fileNameRoot+"invsimpson");
 207                                         }else if (Estimators[i] == "bootstrap") {
 208                                                 cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+"bootstrap")));
 209                                                 outputNames.push_back(fileNameRoot+"bootstrap");
 210                                         }else if (Estimators[i] == "geometric") {
 211                                                 cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+"geometric")));
 212                                                 outputNames.push_back(fileNameRoot+"geometric");
 213                                         }else if (Estimators[i] == "qstat") {
 214                                                 cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+"qstat")));
 215                                                 outputNames.push_back(fileNameRoot+"qstat");
 216                                         }else if (Estimators[i] == "logseries") {
 217                                                 cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+"logseries")));
 218                                                 outputNames.push_back(fileNameRoot+"logseries");
 219                                         }else if (Estimators[i] == "bergerparker") {
 220                                                 cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+"bergerparker")));
 221                                                 outputNames.push_back(fileNameRoot+"bergerparker");
 222                                         }else if (Estimators[i] == "bstick") {
 223                                                 cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+"bstick")));
 224                                                 outputNames.push_back(fileNameRoot+"bstick");
 225                                         }else if (Estimators[i] == "goodscoverage") {
 226                                                 cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+"goodscoverage")));
 227                                                 outputNames.push_back(fileNameRoot+"goodscoverage");
 228                                         }else if (Estimators[i] == "efron") {
 229                                                 cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+"efron")));
 230                                                 outputNames.push_back(fileNameRoot+"efron");
 231                                         }else if (Estimators[i] == "boneh") {
 232                                                 cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+"boneh")));
 233                                                 outputNames.push_back(fileNameRoot+"boneh");
 234                                         }else if (Estimators[i] == "solow") {
 235                                                 cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+"solow")));
 236                                                 outputNames.push_back(fileNameRoot+"solow");
 237                                         }else if (Estimators[i] == "shen") {
 238                                                 cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+"shen")));
 239                                                 outputNames.push_back(fileNameRoot+"shen");
 240                                         }
 241                                 }
 242                         }
 243
 244                         //if the users entered no valid calculators don't execute command
 245                         if (cDisplays.size() == 0) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 246
 247                         read = new ReadOTUFile(inputFileNames[p]);
 248                         read->read(&*globaldata);
 249
 250                         order = globaldata->gorder;
 251                         string lastLabel = order->getLabel();
 252                         input = globaldata->ginput;
 253
 254                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 255                         set<string> processedLabels;
 256                         set<string> userLabels = labels;
 257
 258                         if (m->control_pressed) {
 259                                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 260                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 261                                 delete input;  globaldata->ginput = NULL;
 262                                 delete read;
 263                                 delete order; globaldata->gorder = NULL;
 264                                 delete validCalculator;
 265                                 globaldata->Groups.clear();
 266                                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 267                                 return 0;
 268                         }
 269
 270
 271                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 272
 273                                 if (m->control_pressed) {
 274                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 275                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 276                                         delete input;  globaldata->ginput = NULL;
 277                                         delete read;
 278                                         delete order; globaldata->gorder = NULL;
 279                                         delete validCalculator;
 280                                         globaldata->Groups.clear();
 281                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 282                                         return 0;
 283                                 }
 284
 285
 286                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 287
 288                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 289                                         cCurve = new Collect(order, cDisplays);
 290                                         cCurve->getCurve(freq);
 291                                         delete cCurve;
 292
 293                                         processedLabels.insert(order->getLabel());
 294                                         userLabels.erase(order->getLabel());
 295
 296
 297                                 }
 298                                 //you have a label the user want that is smaller than this label and the last label has not already been processed
 299                                 if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 300                                         string saveLabel = order->getLabel();
 301
 302                                         delete order;
 303                                         order = (input->getOrderVector(lastLabel));
 304
 305                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 306                                         cCurve = new Collect(order, cDisplays);
 307                                         cCurve->getCurve(freq);
 308                                         delete cCurve;
 309
 310
 311                                         processedLabels.insert(order->getLabel());
 312                                         userLabels.erase(order->getLabel());
 313
 314                                         //restore real lastlabel to save below
 315                                         order->setLabel(saveLabel);
 316                                 }
 317
 318                                 lastLabel = order->getLabel();
 319
 320                                 delete order;
 321                                 order = (input->getOrderVector());
 322                         }
 323
 324
 325                         if (m->control_pressed) {
 326                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 327                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 328                                         delete input;  globaldata->ginput = NULL;
 329                                         delete read;
 330                                         delete validCalculator;
 331                                         globaldata->Groups.clear();
 332                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 333                                         return 0;
 334                         }
 335
 336                         //output error messages about any remaining user labels
 337                         set<string>::iterator it;
 338                         bool needToRun = false;
 339                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 340                                 m->mothurOut("Your file does not include the label " + *it);
 341                                 if (processedLabels.count(lastLabel) != 1) {
 342                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 343                                         needToRun = true;
 344                                 }else {
 345                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 346                                 }
 347                         }
 348
 349                         //run last label if you need to
 350                         if (needToRun == true)  {
 351                                 if (order != NULL) {    delete order;   }
 352                                 order = (input->getOrderVector(lastLabel));
 353
 354                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 355
 356                                 cCurve = new Collect(order, cDisplays);
 357                                 cCurve->getCurve(freq);
 358                                 delete cCurve;
 359
 360                                 if (m->control_pressed) {
 361                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 362                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 363                                         delete input;  globaldata->ginput = NULL;
 364                                         delete read;
 365                                         delete order; globaldata->gorder = NULL;
 366                                         delete validCalculator;
 367                                         globaldata->Groups.clear();
 368                                         if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 369                                         return 0;
 370                                 }
 371                                 delete order;
 372                         }
 373
 374                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 375                         cDisplays.clear();
 376                         delete input;  globaldata->ginput = NULL;
 377                         delete read;
 378                         globaldata->gorder = NULL;
 379                         delete validCalculator;
 380                 }
 381
 382                 //return to shared mode if you changed above
 383                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 384
 385                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 386
 387                 m->mothurOutEndLine();
 388                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 389                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 390                 m->mothurOutEndLine();
 391
 392
 393                 return 0;
 394         }
 395         catch(exception& e) {
 396                 m->errorOut(e, "CollectCommand", "execute");
 397                 exit(1);
 398         }
 399 }
 400
 401 //**********************************************************************************************************************
 402 vector<string> CollectCommand::parseSharedFile(string filename) {
 403         try {
 404                 vector<string> filenames;
 405
 406                 map<string, ofstream*> filehandles;
 407                 map<string, ofstream*>::iterator it3;
 408
 409
 410                 //read first line
 411                 read = new ReadOTUFile(filename);
 412                 read->read(&*globaldata);
 413
 414                 input = globaldata->ginput;
 415                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 416
 417                 string sharedFileRoot = getRootName(filename);
 418
 419                 //clears file before we start to write to it below
 420                 for (int i=0; i<lookup.size(); i++) {
 421                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 422                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 423                 }
 424
 425                 ofstream* temp;
 426                 for (int i=0; i<lookup.size(); i++) {
 427                         temp = new ofstream;
 428                         filehandles[lookup[i]->getGroup()] = temp;
 429                         groups.push_back(lookup[i]->getGroup());
 430                 }
 431
 432                 while(lookup[0] != NULL) {
 433
 434                         for (int i = 0; i < lookup.size(); i++) {
 435                                 RAbundVector rav = lookup[i]->getRAbundVector();
 436                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 437                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 438                                 (*(filehandles[lookup[i]->getGroup()])).close();
 439                         }
 440
 441                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 442                         lookup = input->getSharedRAbundVectors();
 443                 }
 444
 445                 //free memory
 446                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 447                         delete it3->second;
 448                 }
 449                 delete read;
 450                 delete input;
 451                 globaldata->ginput = NULL;
 452
 453                 return filenames;
 454         }
 455         catch(exception& e) {
 456                 m->errorOut(e, "CollectCommand", "parseSharedFile");
 457                 exit(1);
 458         }
 459 }
 460 //**********************************************************************************************************************
 461