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