rarefactcommand.cpp

   1 /*
   2  *  rarefactcommand.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 "rarefactcommand.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 "heip.h"
  19 #include "smithwilson.h"
  20 #include "invsimpson.h"
  21 #include "npshannon.h"
  22 #include "shannoneven.h"
  23 #include "shannon.h"
  24 #include "jackknife.h"
  25 #include "coverage.h"
  26
  27 //**********************************************************************************************************************
  28
  29
  30 RareFactCommand::RareFactCommand(string option)  {
  31         try {
  32                 globaldata = GlobalData::getInstance();
  33                 abort = false;
  34                 allLines = 1;
  35                 labels.clear();
  36                 Estimators.clear();
  37
  38                 //allow user to run help
  39                 if(option == "help") { validCalculator = new ValidCalculators(); help(); delete validCalculator; abort = true; }
  40
  41                 else {
  42                         //valid paramters for this command
  43                         string Array[] =  {"iters","freq","label","calc","abund","processors","outputdir","inputdir"};
  44                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  45
  46                         OptionParser parser(option);
  47                         map<string,string> parameters = parser.getParameters();
  48
  49                         ValidParameters validParameter;
  50
  51                         //check to make sure all parameters are valid for command
  52                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  53                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  54                         }
  55
  56                         //if the user changes the output directory command factory will send this info to us in the output parameter
  57                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
  58                                 outputDir = "";
  59                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
  60                         }
  61
  62                         //make sure the user has already run the read.otu command
  63                         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 rarefact.single command."); m->mothurOutEndLine(); abort = true; }
  64
  65                         //check for optional parameter and set defaults
  66                         // ...at some point should added some additional type checking...
  67                         label = validParameter.validFile(parameters, "label", false);
  68                         if (label == "not found") { label = ""; }
  69                         else {
  70                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
  71                                 else { allLines = 1;  }
  72                         }
  73
  74                         //if the user has not specified any labels use the ones from read.otu
  75                         if(label == "") {
  76                                 allLines = globaldata->allLines;
  77                                 labels = globaldata->labels;
  78                         }
  79
  80                         calc = validParameter.validFile(parameters, "calc", false);
  81                         if (calc == "not found") { calc = "sobs";  }
  82                         else {
  83                                  if (calc == "default")  {  calc = "sobs";  }
  84                         }
  85                         m->splitAtDash(calc, Estimators);
  86
  87                         string temp;
  88                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
  89                         convert(temp, freq);
  90
  91                         temp = validParameter.validFile(parameters, "abund", false);                    if (temp == "not found") { temp = "10"; }
  92                         convert(temp, abund);
  93
  94                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
  95                         convert(temp, nIters);
  96
  97                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = "1";                             }
  98                         convert(temp, processors);
  99                 }
 100
 101         }
 102         catch(exception& e) {
 103                 m->errorOut(e, "RareFactCommand", "RareFactCommand");
 104                 exit(1);
 105         }
 106 }
 107 //**********************************************************************************************************************
 108
 109 void RareFactCommand::help(){
 110         try {
 111                 m->mothurOut("The rarefaction.single command can only be executed after a successful read.otu WTIH ONE EXECEPTION.\n");
 112                 m->mothurOut("The rarefaction.single command can be executed after a successful cluster command.  It will use the .list file from the output of the cluster.\n");
 113                 m->mothurOut("The rarefaction.single command parameters are label, iters, freq, calc, processors and abund.  No parameters are required. \n");
 114                 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");
 115                 m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
 116                 m->mothurOut("The rarefaction.single command should be in the following format: \n");
 117                 m->mothurOut("rarefaction.single(label=yourLabel, iters=yourIters, freq=yourFreq, calc=yourEstimators).\n");
 118                 m->mothurOut("Example rarefaction.single(label=unique-.01-.03, iters=10000, freq=10, calc=sobs-rchao-race-rjack-rbootstrap-rshannon-rnpshannon-rsimpson).\n");
 119                 m->mothurOut("The default values for iters is 1000, freq is 100, and calc is rarefaction which calculates the rarefaction curve for the observed richness.\n");
 120                 validCalculator->printCalc("rarefaction", cout);
 121                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 122                 m->mothurOut("Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n\n");
 123         }
 124         catch(exception& e) {
 125                 m->errorOut(e, "RareFactCommand", "help");
 126                 exit(1);
 127         }
 128 }
 129
 130 //**********************************************************************************************************************
 131
 132 RareFactCommand::~RareFactCommand(){}
 133
 134 //**********************************************************************************************************************
 135
 136 int RareFactCommand::execute(){
 137         try {
 138
 139                 if (abort == true) { return 0; }
 140
 141                 vector<string> outputNames;
 142
 143                 string hadShared = "";
 144                 if ((globaldata->getFormat() != "sharedfile")) { inputFileNames.push_back(globaldata->inputFileName);  }
 145                 else { hadShared = globaldata->getSharedFile(); inputFileNames = parseSharedFile(globaldata->getSharedFile());  globaldata->setFormat("rabund");  }
 146
 147                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 148
 149                 for (int p = 0; p < inputFileNames.size(); p++) {
 150
 151                         string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(inputFileNames[p]));
 152                         globaldata->inputFileName = inputFileNames[p];
 153
 154                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } return 0; }
 155
 156                         if (inputFileNames.size() > 1) {
 157                                 m->mothurOutEndLine(); m->mothurOut("Processing group " + groups[p]); m->mothurOutEndLine(); m->mothurOutEndLine();
 158                         }
 159                         int i;
 160                         validCalculator = new ValidCalculators();
 161
 162
 163                         for (i=0; i<Estimators.size(); i++) {
 164                                 if (validCalculator->isValidCalculator("rarefaction", Estimators[i]) == true) {
 165                                         if (Estimators[i] == "sobs") {
 166                                                 rDisplays.push_back(new RareDisplay(new Sobs(), new ThreeColumnFile(fileNameRoot+"rarefaction")));
 167                                                 outputNames.push_back(fileNameRoot+"rarefaction");
 168                                         }else if (Estimators[i] == "chao") {
 169                                                 rDisplays.push_back(new RareDisplay(new Chao1(), new ThreeColumnFile(fileNameRoot+"r_chao")));
 170                                                 outputNames.push_back(fileNameRoot+"r_chao");
 171                                         }else if (Estimators[i] == "ace") {
 172                                                 if(abund < 5)
 173                                                         abund = 10;
 174                                                 rDisplays.push_back(new RareDisplay(new Ace(abund), new ThreeColumnFile(fileNameRoot+"r_ace")));
 175                                                 outputNames.push_back(fileNameRoot+"r_ace");
 176                                         }else if (Estimators[i] == "jack") {
 177                                                 rDisplays.push_back(new RareDisplay(new Jackknife(), new ThreeColumnFile(fileNameRoot+"r_jack")));
 178                                                 outputNames.push_back(fileNameRoot+"r_jack");
 179                                         }else if (Estimators[i] == "shannon") {
 180                                                 rDisplays.push_back(new RareDisplay(new Shannon(), new ThreeColumnFile(fileNameRoot+"r_shannon")));
 181                                                 outputNames.push_back(fileNameRoot+"r_shannon");
 182                                         }else if (Estimators[i] == "shannoneven") {
 183                                                 rDisplays.push_back(new RareDisplay(new ShannonEven(), new ThreeColumnFile(fileNameRoot+"r_shannoneven")));
 184                                                 outputNames.push_back(fileNameRoot+"r_shannoneven");
 185                                         }else if (Estimators[i] == "heip") {
 186                                                 rDisplays.push_back(new RareDisplay(new Heip(), new ThreeColumnFile(fileNameRoot+"r_heip")));
 187                                                 outputNames.push_back(fileNameRoot+"r_heip");
 188                                         }else if (Estimators[i] == "smithwilson") {
 189                                                 rDisplays.push_back(new RareDisplay(new SmithWilson(), new ThreeColumnFile(fileNameRoot+"r_smithwilson")));
 190                                                 outputNames.push_back(fileNameRoot+"r_smithwilson");
 191                                         }else if (Estimators[i] == "npshannon") {
 192                                                 rDisplays.push_back(new RareDisplay(new NPShannon(), new ThreeColumnFile(fileNameRoot+"r_npshannon")));
 193                                                 outputNames.push_back(fileNameRoot+"r_npshannon");
 194                                         }else if (Estimators[i] == "simpson") {
 195                                                 rDisplays.push_back(new RareDisplay(new Simpson(), new ThreeColumnFile(fileNameRoot+"r_simpson")));
 196                                                 outputNames.push_back(fileNameRoot+"r_simpson");
 197                                         }else if (Estimators[i] == "simpsoneven") {
 198                                                 rDisplays.push_back(new RareDisplay(new SimpsonEven(), new ThreeColumnFile(fileNameRoot+"r_simpsoneven")));
 199                                                 outputNames.push_back(fileNameRoot+"r_simpsoneven");
 200                                         }else if (Estimators[i] == "invsimpson") {
 201                                                 rDisplays.push_back(new RareDisplay(new InvSimpson(), new ThreeColumnFile(fileNameRoot+"r_invsimpson")));
 202                                                 outputNames.push_back(fileNameRoot+"r_invsimpson");
 203                                         }else if (Estimators[i] == "bootstrap") {
 204                                                 rDisplays.push_back(new RareDisplay(new Bootstrap(), new ThreeColumnFile(fileNameRoot+"r_bootstrap")));
 205                                                 outputNames.push_back(fileNameRoot+"r_bootstrap");
 206                                         }else if (Estimators[i] == "coverage") {
 207                                                 rDisplays.push_back(new RareDisplay(new Coverage(), new ThreeColumnFile(fileNameRoot+"r_coverage")));
 208                                                 outputNames.push_back(fileNameRoot+"r_coverage");
 209                                         }else if (Estimators[i] == "nseqs") {
 210                                                 rDisplays.push_back(new RareDisplay(new NSeqs(), new ThreeColumnFile(fileNameRoot+"r_nseqs")));
 211                                                 outputNames.push_back(fileNameRoot+"r_nseqs");
 212                                         }
 213                                 }
 214                         }
 215
 216
 217                         //if the users entered no valid calculators don't execute command
 218                         if (rDisplays.size() == 0) { for(int i=0;i<rDisplays.size();i++){       delete rDisplays[i];    } delete validCalculator; return 0; }
 219
 220                         read = new ReadOTUFile(globaldata->inputFileName);
 221                         read->read(&*globaldata);
 222
 223                         order = globaldata->gorder;
 224                         string lastLabel = order->getLabel();
 225                         input = globaldata->ginput;
 226
 227                         //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 228                         set<string> processedLabels;
 229                         set<string> userLabels = labels;
 230
 231                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    } delete validCalculator; delete read; delete input; globaldata->ginput = NULL; delete order; globaldata->gorder = NULL; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 232
 233                         //as long as you are not at the end of the file or done wih the lines you want
 234                         while((order != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 235
 236                                 if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    } delete validCalculator; delete read; delete input; globaldata->ginput = NULL; delete order; globaldata->gorder = NULL; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 237
 238
 239                                 if(allLines == 1 || labels.count(order->getLabel()) == 1){
 240
 241                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 242                                         rCurve = new Rarefact(order, rDisplays, processors);
 243                                         rCurve->getCurve(freq, nIters);
 244                                         delete rCurve;
 245
 246                                         processedLabels.insert(order->getLabel());
 247                                         userLabels.erase(order->getLabel());
 248                                 }
 249
 250                                 if ((m->anyLabelsToProcess(order->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 251                                         string saveLabel = order->getLabel();
 252
 253                                         delete order;
 254                                         order = (input->getOrderVector(lastLabel));
 255
 256                                         m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 257                                         rCurve = new Rarefact(order, rDisplays, processors);
 258                                         rCurve->getCurve(freq, nIters);
 259                                         delete rCurve;
 260
 261                                         processedLabels.insert(order->getLabel());
 262                                         userLabels.erase(order->getLabel());
 263
 264                                         //restore real lastlabel to save below
 265                                         order->setLabel(saveLabel);
 266                                 }
 267
 268                                 lastLabel = order->getLabel();
 269
 270                                 delete order;
 271                                 order = (input->getOrderVector());
 272                         }
 273
 274                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    } delete validCalculator; delete read; delete input; globaldata->ginput = NULL; for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str()); }  return 0; }
 275
 276                         //output error messages about any remaining user labels
 277                         set<string>::iterator it;
 278                         bool needToRun = false;
 279                         for (it = userLabels.begin(); it != userLabels.end(); it++) {
 280                                 m->mothurOut("Your file does not include the label " + *it);
 281                                 if (processedLabels.count(lastLabel) != 1) {
 282                                         m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 283                                         needToRun = true;
 284                                 }else {
 285                                         m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 286                                 }
 287                         }
 288
 289                         if (m->control_pressed) { if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  } for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    } delete validCalculator; delete read; delete input; globaldata->ginput = NULL;  for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
 290
 291                         //run last label if you need to
 292                         if (needToRun == true)  {
 293                                 if (order != NULL) {    delete order;   }
 294                                 order = (input->getOrderVector(lastLabel));
 295
 296                                 m->mothurOut(order->getLabel()); m->mothurOutEndLine();
 297                                 rCurve = new Rarefact(order, rDisplays, processors);
 298                                 rCurve->getCurve(freq, nIters);
 299                                 delete rCurve;
 300
 301                                 delete order;
 302                         }
 303
 304
 305                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 306                         rDisplays.clear();
 307                         globaldata->gorder = NULL;
 308                         delete input;  globaldata->ginput = NULL;
 309                         delete read;
 310                         delete validCalculator;
 311
 312                 }
 313
 314                 if (hadShared != "") {  globaldata->setSharedFile(hadShared); globaldata->setFormat("sharedfile");  }
 315
 316                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str()); } return 0; }
 317
 318                 m->mothurOutEndLine();
 319                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 320                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 321                 m->mothurOutEndLine();
 322
 323                 return 0;
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "RareFactCommand", "execute");
 327                 exit(1);
 328         }
 329 }
 330 //**********************************************************************************************************************
 331 vector<string> RareFactCommand::parseSharedFile(string filename) {
 332         try {
 333                 vector<string> filenames;
 334
 335                 map<string, ofstream*> filehandles;
 336                 map<string, ofstream*>::iterator it3;
 337
 338
 339                 //read first line
 340                 read = new ReadOTUFile(filename);
 341                 read->read(&*globaldata);
 342
 343                 input = globaldata->ginput;
 344                 vector<SharedRAbundVector*> lookup = input->getSharedRAbundVectors();
 345
 346                 string sharedFileRoot = m->getRootName(filename);
 347
 348                 //clears file before we start to write to it below
 349                 for (int i=0; i<lookup.size(); i++) {
 350                         remove((sharedFileRoot + lookup[i]->getGroup() + ".rabund").c_str());
 351                         filenames.push_back((sharedFileRoot + lookup[i]->getGroup() + ".rabund"));
 352                 }
 353
 354                 ofstream* temp;
 355                 for (int i=0; i<lookup.size(); i++) {
 356                         temp = new ofstream;
 357                         filehandles[lookup[i]->getGroup()] = temp;
 358                         groups.push_back(lookup[i]->getGroup());
 359                 }
 360
 361                 while(lookup[0] != NULL) {
 362
 363                         for (int i = 0; i < lookup.size(); i++) {
 364                                 RAbundVector rav = lookup[i]->getRAbundVector();
 365                                 m->openOutputFileAppend(sharedFileRoot + lookup[i]->getGroup() + ".rabund", *(filehandles[lookup[i]->getGroup()]));
 366                                 rav.print(*(filehandles[lookup[i]->getGroup()]));
 367                                 (*(filehandles[lookup[i]->getGroup()])).close();
 368                         }
 369
 370                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 371                         lookup = input->getSharedRAbundVectors();
 372                 }
 373
 374                 //free memory
 375                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 376                         delete it3->second;
 377                 }
 378                 delete read;
 379                 delete input;
 380                 globaldata->ginput = NULL;
 381
 382                 return filenames;
 383         }
 384         catch(exception& e) {
 385                 m->errorOut(e, "RareFactCommand", "parseSharedFile");
 386                 exit(1);
 387         }
 388 }
 389 //**********************************************************************************************************************
 390
 391
 392