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                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 141                 else {  inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 142
 143                 for (int p = 0; p < inputFileNames.size(); p++) {
 144
 145                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         }  globaldata->Groups.clear(); return 0; }
 146
 147                         if (outputDir == "") { outputDir += hasPath(inputFileNames[p]); }
 148                         string fileNameRoot = outputDir + getRootName(getSimpleName(inputFileNames[p]));
 149                         globaldata->inputFileName = inputFileNames[p];
 150
 151                         if (inputFileNames.size() > 1) {
 152                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 153                         }
 154
 155                         validCalculator = new ValidCalculators();
 156
 157                         for (int i=0; i<Estimators.size(); i++) {
 158                                 if (validCalculator->isValidCalculator("single", Estimators[i]) == true) {
 159                                         if (Estimators[i] == "sobs") {
 160                                                 cDisplays.push_back(new CollectDisplay(new Sobs(), new OneColumnFile(fileNameRoot+"sobs")));
 161                                                 outputNames.push_back(fileNameRoot+"sobs");
 162                                         }else if (Estimators[i] == "chao") {
 163                                                 cDisplays.push_back(new CollectDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"chao")));
 164                                                 outputNames.push_back(fileNameRoot+"chao");
 165                                         }else if (Estimators[i] == "nseqs") {
 166                                                 cDisplays.push_back(new CollectDisplay(new NSeqs(), new OneColumnFile(fileNameRoot+"nseqs")));
 167                                                 outputNames.push_back(fileNameRoot+"nseqs");
 168                                         }else if (Estimators[i] == "coverage") {
 169                                                 cDisplays.push_back(new CollectDisplay(new Coverage(), new OneColumnFile(fileNameRoot+"coverage")));
 170                                                 outputNames.push_back(fileNameRoot+"coverage");
 171                                         }else if (Estimators[i] == "ace") {
 172                                                 cDisplays.push_back(new CollectDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"ace")));
 173                                                 outputNames.push_back(fileNameRoot+"ace");
 174                                         }else if (Estimators[i] == "jack") {
 175                                                 cDisplays.push_back(new CollectDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"jack")));
 176                                                 outputNames.push_back(fileNameRoot+"jack");
 177                                         }else if (Estimators[i] == "shannon") {
 178                                                 cDisplays.push_back(new CollectDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"shannon")));
 179                                                 outputNames.push_back(fileNameRoot+"shannon");
 180                                         }else if (Estimators[i] == "npshannon") {
 181                                                 cDisplays.push_back(new CollectDisplay(new NPShannon(), new OneColumnFile(fileNameRoot+"np_shannon")));
 182                                                 outputNames.push_back(fileNameRoot+"np_shannon");
 183                                         }else if (Estimators[i] == "simpson") {
 184                                                 cDisplays.push_back(new CollectDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"simpson")));
 185                                                 outputNames.push_back(fileNameRoot+"simpson");
 186                                         }else if (Estimators[i] == "bootstrap") {
 187                                                 cDisplays.push_back(new CollectDisplay(new Bootstrap(), new OneColumnFile(fileNameRoot+"bootstrap")));
 188                                                 outputNames.push_back(fileNameRoot+"bootstrap");
 189                                         }else if (Estimators[i] == "geometric") {
 190                                                 cDisplays.push_back(new CollectDisplay(new Geom(), new OneColumnFile(fileNameRoot+"geometric")));
 191                                                 outputNames.push_back(fileNameRoot+"geometric");
 192                                         }else if (Estimators[i] == "qstat") {
 193                                                 cDisplays.push_back(new CollectDisplay(new QStat(), new OneColumnFile(fileNameRoot+"qstat")));
 194                                                 outputNames.push_back(fileNameRoot+"qstat");
 195                                         }else if (Estimators[i] == "logseries") {
 196                                                 cDisplays.push_back(new CollectDisplay(new LogSD(), new OneColumnFile(fileNameRoot+"logseries")));
 197                                                 outputNames.push_back(fileNameRoot+"logseries");
 198                                         }else if (Estimators[i] == "bergerparker") {
 199                                                 cDisplays.push_back(new CollectDisplay(new BergerParker(), new OneColumnFile(fileNameRoot+"bergerparker")));
 200                                                 outputNames.push_back(fileNameRoot+"bergerparker");
 201                                         }else if (Estimators[i] == "bstick") {
 202                                                 cDisplays.push_back(new CollectDisplay(new BStick(), new ThreeColumnFile(fileNameRoot+"bstick")));
 203                                                 outputNames.push_back(fileNameRoot+"bstick");
 204                                         }else if (Estimators[i] == "goodscoverage") {
 205                                                 cDisplays.push_back(new CollectDisplay(new GoodsCoverage(), new OneColumnFile(fileNameRoot+"goodscoverage")));
 206                                                 outputNames.push_back(fileNameRoot+"goodscoverage");
 207                                         }else if (Estimators[i] == "efron") {
 208                                                 cDisplays.push_back(new CollectDisplay(new Efron(size), new OneColumnFile(fileNameRoot+"efron")));
 209                                                 outputNames.push_back(fileNameRoot+"efron");
 210                                         }else if (Estimators[i] == "boneh") {
 211                                                 cDisplays.push_back(new CollectDisplay(new Boneh(size), new OneColumnFile(fileNameRoot+"boneh")));
 212                                                 outputNames.push_back(fileNameRoot+"boneh");
 213                                         }else if (Estimators[i] == "solow") {
 214                                                 cDisplays.push_back(new CollectDisplay(new Solow(size), new OneColumnFile(fileNameRoot+"solow")));
 215                                                 outputNames.push_back(fileNameRoot+"solow");
 216                                         }else if (Estimators[i] == "shen") {
 217                                                 cDisplays.push_back(new CollectDisplay(new Shen(size, abund), new OneColumnFile(fileNameRoot+"shen")));
 218                                                 outputNames.push_back(fileNameRoot+"shen");
 219                                         }
 220                                 }
 221                         }
 222
 223                         //if the users entered no valid calculators don't execute command
 224                         if (cDisplays.size() == 0) { return 0; }
 225
 226                         read = new ReadOTUFile(inputFileNames[p]);
 227                         read->read(&*globaldata);
 228
 229                         order = globaldata->gorder;
 230                         string lastLabel = order->getLabel();
 231                         input = globaldata->ginput;
 232
 233                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 234                         set<string> processedLabels;
 235                         set<string> userLabels = labels;
 236
 237                         if (m->control_pressed) {
 238                                 for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 239                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 240                                 delete input;  globaldata->ginput = NULL;
 241                                 delete read;
 242                                 delete order; globaldata->gorder = NULL;
 243                                 delete validCalculator;
 244                                 globaldata->Groups.clear();
 245                                 return 0;
 246                         }
 247
 248
 249                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 250
 251                                 if (m->control_pressed) {
 252                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 253                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 254                                         delete input;  globaldata->ginput = NULL;
 255                                         delete read;
 256                                         delete order; globaldata->gorder = NULL;
 257                                         delete validCalculator;
 258                                         globaldata->Groups.clear();
 259                                         return 0;
 260                                 }
 261
 262
 263                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 264
 265                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 266                                         cCurve = new Collect(order, cDisplays);
 267                                         cCurve->getCurve(freq);
 268                                         delete cCurve;
 269
 270                                         processedLabels.insert(order->getLabel());
 271                                         userLabels.erase(order->getLabel());
 272
 273
 274                                 }
 275                                 //you have a label the user want that is smaller than this label and the last label has not already been processed
 276                                 if ((anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 277                                         string saveLabel = order->getLabel();
 278
 279                                         delete order;
 280                                         order = (input->getOrderVector(lastLabel));
 281
 282                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 283                                         cCurve = new Collect(order, cDisplays);
 284                                         cCurve->getCurve(freq);
 285                                         delete cCurve;
 286
 287
 288                                         processedLabels.insert(order->getLabel());
 289                                         userLabels.erase(order->getLabel());
 290
 291                                         //restore real lastlabel to save below
 292                                         order->setLabel(saveLabel);
 293                                 }
 294
 295                                 lastLabel = order->getLabel();
 296
 297                                 delete order;
 298                                 order = (input->getOrderVector());
 299                         }
 300
 301
 302                         if (m->control_pressed) {
 303                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 304                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 305                                         delete input;  globaldata->ginput = NULL;
 306                                         delete read;
 307                                         delete validCalculator;
 308                                         globaldata->Groups.clear();
 309                                         return 0;
 310                         }
 311
 312                         //output error messages about any remaining user labels
 313                         set<string>::iterator it;
 314                         bool needToRun = false;
 315                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 316                                 m->mothurOut("Your file does not include the label " + *it);
 317                                 if (processedLabels.count(lastLabel) != 1) {
 318                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 319                                         needToRun = true;
 320                                 }else {
 321                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 322                                 }
 323                         }
 324
 325                         //run last label if you need to
 326                         if (needToRun == true)  {
 327                                 if (order != NULL) {    delete order;   }
 328                                 order = (input->getOrderVector(lastLabel));
 329
 330                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 331
 332                                 cCurve = new Collect(order, cDisplays);
 333                                 cCurve->getCurve(freq);
 334                                 delete cCurve;
 335
 336                                 if (m->control_pressed) {
 337                                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 338                                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 339                                         delete input;  globaldata->ginput = NULL;
 340                                         delete read;
 341                                         delete order; globaldata->gorder = NULL;
 342                                         delete validCalculator;
 343                                         globaldata->Groups.clear();
 344                                         return 0;
 345                                 }
 346                                 delete order;
 347                         }
 348
 349                         for(int i=0;i<cDisplays.size();i++){    delete cDisplays[i];    }
 350                         cDisplays.clear();
 351                         delete input;  globaldata->ginput = NULL;
 352                         delete read;
 353                         globaldata->gorder = NULL;
 354                         delete validCalculator;
 355                 }
 356
 357                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());         } return 0; }
 358
 359                 m->mothurOutEndLine();
 360                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 361                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 362                 m->mothurOutEndLine();
 363
 364
 365                 return 0;
 366         }
 367         catch(exception& e) {
 368                 m->errorOut(e, "CollectCommand", "execute");
 369                 exit(1);
 370         }
 371 }
 372
 373 //**********************************************************************************************************************
 374 vector<string> CollectCommand::parseSharedFile(string filename) {
 375         try {
 376                 vector<string> filenames;
 377
 378                 map<string, ofstream*> filehandles;
 379                 map<string, ofstream*>::iterator it3;
 380
 381
 382                 //read first line
 383                 read = new ReadOTUFile(filename);
 384                 read->read(&*globaldata);
 385
 386                 input = globaldata->ginput;
 387                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 388
 389                 string sharedFileRoot = getRootName(filename);
 390
 391                 //clears file before we start to write to it below
 392                 for (int i=0; i<lookup.size(); i++) {
 393                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 394                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 395                 }
 396
 397                 ofstream* temp;
 398                 for (int i=0; i<lookup.size(); i++) {
 399                         temp = new ofstream;
 400                         filehandles[lookup[i]->getGroup()] = temp;
 401                         groups.push_back(lookup[i]->getGroup());
 402                 }
 403
 404                 while(lookup[0] != NULL) {
 405
 406                         for (int i = 0; i < lookup.size(); i++) {
 407                                 RAbundVector rav = lookup[i]->getRAbundVector();
 408                                 openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 409                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 410                                 (*(filehandles[lookup[i]->getGroup()])).close();
 411                         }
 412
 413                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 414                         lookup = input->getSharedRAbundVectors();
 415                 }
 416
 417                 //free memory
 418                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 419                         delete it3->second;
 420                 }
 421                 delete read;
 422                 delete input;
 423                 globaldata->ginput = NULL;
 424
 425                 return filenames;
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "CollectCommand", "parseSharedFile");
 429                 exit(1);
 430         }
 431 }
 432 //**********************************************************************************************************************
 433