rarefactsharedcommand.cpp

   1 /*
   2  *  rarefactsharedcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/6/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "rarefactsharedcommand.h"
  11 #include "sharedsobs.h"
  12 #include "sharednseqs.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> RareFactSharedCommand::getValidParameters(){
  16         try {
  17                 string Array[] =  {"iters","freq","label","calc","groups", "jumble","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "RareFactSharedCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 RareFactSharedCommand::RareFactSharedCommand(){
  28         try {
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["sharedrarefaction"] = tempOutNames;
  32                 outputTypes["sharedr_nseqs"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "RareFactSharedCommand", "RareFactSharedCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> RareFactSharedCommand::getRequiredParameters(){
  41         try {
  42                 vector<string> myArray;
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "RareFactSharedCommand", "getRequiredParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 vector<string> RareFactSharedCommand::getRequiredFiles(){
  52         try {
  53                 string Array[] =  {"shared"};
  54                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "RareFactSharedCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63
  64 RareFactSharedCommand::RareFactSharedCommand(string option)  {
  65         try {
  66                 globaldata = GlobalData::getInstance();
  67
  68                 abort = false;
  69                 allLines = 1;
  70                 labels.clear();
  71                 Estimators.clear();
  72                 Groups.clear();
  73
  74                 //allow user to run help
  75                 if(option == "help") { validCalculator = new ValidCalculators(); help(); abort = true; }
  76
  77                 else {
  78                         //valid paramters for this command
  79                         string Array[] =  {"iters","freq","label","calc","groups", "jumble","outputdir","inputdir"};
  80                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  81
  82                         OptionParser parser(option);
  83                         map<string,string> parameters = parser.getParameters();
  84
  85                         ValidParameters validParameter;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["sharedrarefaction"] = tempOutNames;
  95                         outputTypes["sharedr_nseqs"] = tempOutNames;
  96
  97                         //make sure the user has already run the read.otu command
  98                         if (globaldata->getSharedFile() == "") {
  99                                 if (globaldata->getListFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the collect.shared command."); m->mothurOutEndLine(); abort = true; }
 100                                 else if (globaldata->getGroupFile() == "") { m->mothurOut("You must read a list and a group, or a shared before you can use the collect.shared command."); m->mothurOutEndLine(); abort = true; }
 101                         }
 102
 103                         //if the user changes the output directory command factory will send this info to us in the output parameter
 104                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){
 105                                 outputDir = "";
 106                                 outputDir += m->hasPath(globaldata->inputFileName); //if user entered a file with a path then preserve it
 107                         }
 108
 109
 110                         //check for optional parameter and set defaults
 111                         // ...at some point should added some additional type checking...
 112                         label = validParameter.validFile(parameters, "label", false);
 113                         if (label == "not found") { label = ""; }
 114                         else {
 115                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 116                                 else { allLines = 1;  }
 117                         }
 118
 119                         //if the user has not specified any labels use the ones from read.otu
 120                         if(label == "") {
 121                                 allLines = globaldata->allLines;
 122                                 labels = globaldata->labels;
 123                         }
 124
 125                         calc = validParameter.validFile(parameters, "calc", false);
 126                         if (calc == "not found") { calc = "sharedobserved";  }
 127                         else {
 128                                  if (calc == "default")  {  calc = "sharedobserved";  }
 129                         }
 130                         m->splitAtDash(calc, Estimators);
 131
 132                         groups = validParameter.validFile(parameters, "groups", false);
 133                         if (groups == "not found") { groups = ""; }
 134                         else {
 135                                 m->splitAtDash(groups, Groups);
 136                         }
 137                         globaldata->Groups = Groups;
 138
 139                         string temp;
 140                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 141                         convert(temp, freq);
 142
 143                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 144                         convert(temp, nIters);
 145
 146                         temp = validParameter.validFile(parameters, "jumble", false);                   if (temp == "not found") { temp = "T"; }
 147                         if (m->isTrue(temp)) { jumble = true; }
 148                         else { jumble = false; }
 149                         globaldata->jumble = jumble;
 150
 151                         if (abort == false) {
 152
 153                                 string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(globaldata->inputFileName));
 154 //                              format = globaldata->getFormat();
 155
 156
 157                                 validCalculator = new ValidCalculators();
 158
 159                                 for (int i=0; i<Estimators.size(); i++) {
 160                                         if (validCalculator->isValidCalculator("sharedrarefaction", Estimators[i]) == true) {
 161                                                 if (Estimators[i] == "sharedobserved") {
 162                                                         rDisplays.push_back(new RareDisplay(new SharedSobs(), new SharedThreeColumnFile(fileNameRoot+"shared.rarefaction", "")));
 163                                                         outputNames.push_back(fileNameRoot+"shared.rarefaction"); outputTypes["sharedrarefaction"].push_back(fileNameRoot+"shared.rarefaction");
 164                                                 }else if (Estimators[i] == "sharednseqs") {
 165                                                         rDisplays.push_back(new RareDisplay(new SharedNSeqs(), new SharedThreeColumnFile(fileNameRoot+"shared.r_nseqs", "")));
 166                                                         outputNames.push_back(fileNameRoot+"shared.r_nseqs"); outputTypes["sharedr_nseqs"].push_back(fileNameRoot+"shared.r_nseqs");
 167                                                 }
 168                                         }
 169                                 }
 170                         }
 171
 172                 }
 173
 174         }
 175         catch(exception& e) {
 176                 m->errorOut(e, "RareFactSharedCommand", "RareFactSharedCommand");
 177                 exit(1);
 178         }
 179 }
 180
 181 //**********************************************************************************************************************
 182
 183 void RareFactSharedCommand::help(){
 184         try {
 185                 m->mothurOut("The rarefaction.shared command can only be executed after a successful read.otu command.\n");
 186                 m->mothurOut("The rarefaction.shared command parameters are label, iters, groups, jumble and calc.  No parameters are required.\n");
 187                 m->mothurOut("The rarefaction command should be in the following format: \n");
 188                 m->mothurOut("rarefaction.shared(label=yourLabel, iters=yourIters, calc=yourEstimators, jumble=yourJumble, groups=yourGroups).\n");
 189                 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");
 190                 m->mothurOut("Example rarefaction.shared(label=unique-0.01-0.03,  iters=10000, groups=B-C, jumble=T, calc=sharedobserved).\n");
 191                 m->mothurOut("The default values for iters is 1000, freq is 100, and calc is sharedobserved which calculates the shared rarefaction curve for the observed richness.\n");
 192                 m->mothurOut("The default value for groups is all the groups in your groupfile, and jumble is true.\n");
 193                 validCalculator->printCalc("sharedrarefaction", cout);
 194                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 195                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like analyzed.  You must enter at least 2 valid groups.\n");
 196                 m->mothurOut("Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n\n");
 197         }
 198         catch(exception& e) {
 199                 m->errorOut(e, "RareFactSharedCommand", "help");
 200                 exit(1);
 201         }
 202 }
 203
 204 //**********************************************************************************************************************
 205
 206 RareFactSharedCommand::~RareFactSharedCommand(){
 207         if (abort == false) {
 208                 delete input;   globaldata->ginput = NULL;
 209                 delete read;
 210                 delete validCalculator;
 211         }
 212 }
 213
 214 //**********************************************************************************************************************
 215
 216 int RareFactSharedCommand::execute(){
 217         try {
 218
 219                 if (abort == true) { return 0; }
 220
 221                 //if the users entered no valid calculators don't execute command
 222                 if (rDisplays.size() == 0) { return 0; }
 223
 224                 read = new ReadOTUFile(globaldata->inputFileName);
 225                 read->read(&*globaldata);
 226
 227                 input = globaldata->ginput;
 228                 lookup = input->getSharedRAbundVectors();
 229                 string lastLabel = lookup[0]->getLabel();
 230
 231                 if (m->control_pressed) {
 232                         globaldata->Groups.clear();
 233                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 234                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 235                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 236                         return 0;
 237                 }
 238
 239
 240                 if (lookup.size() < 2) {
 241                         m->mothurOut("I cannot run the command without at least 2 valid groups.");
 242                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 243                         return 0;
 244                 }
 245
 246                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 247                 set<string> processedLabels;
 248                 set<string> userLabels = labels;
 249
 250                 //as long as you are not at the end of the file or done wih the lines you want
 251                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 252                         if (m->control_pressed) {
 253                                 globaldata->Groups.clear();
 254                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 255                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 256                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 257                                 return 0;
 258                         }
 259
 260                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 261                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 262                                 rCurve = new Rarefact(lookup, rDisplays);
 263                                 rCurve->getSharedCurve(freq, nIters);
 264                                 delete rCurve;
 265
 266                                 processedLabels.insert(lookup[0]->getLabel());
 267                                 userLabels.erase(lookup[0]->getLabel());
 268                         }
 269
 270                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 271                                         string saveLabel = lookup[0]->getLabel();
 272
 273                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 274                                         lookup = input->getSharedRAbundVectors(lastLabel);
 275
 276                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 277                                         rCurve = new Rarefact(lookup, rDisplays);
 278                                         rCurve->getSharedCurve(freq, nIters);
 279                                         delete rCurve;
 280
 281                                         processedLabels.insert(lookup[0]->getLabel());
 282                                         userLabels.erase(lookup[0]->getLabel());
 283
 284                                         //restore real lastlabel to save below
 285                                         lookup[0]->setLabel(saveLabel);
 286                         }
 287
 288
 289                         lastLabel = lookup[0]->getLabel();
 290
 291                         //get next line to process
 292                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 293                         lookup = input->getSharedRAbundVectors();
 294                 }
 295
 296                 if (m->control_pressed) {
 297                                 globaldata->Groups.clear();
 298                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 299                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 300                                 return 0;
 301                 }
 302
 303                 //output error messages about any remaining user labels
 304                 set<string>::iterator it;
 305                 bool needToRun = false;
 306                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 307                         m->mothurOut("Your file does not include the label " + *it);
 308                         if (processedLabels.count(lastLabel) != 1) {
 309                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 310                                 needToRun = true;
 311                         }else {
 312                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 313                         }
 314                 }
 315
 316                 if (m->control_pressed) {
 317                                 globaldata->Groups.clear();
 318                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 319                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 320                                 return 0;
 321                 }
 322
 323                 //run last label if you need to
 324                 if (needToRun == true)  {
 325                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i]; }  }
 326                         lookup = input->getSharedRAbundVectors(lastLabel);
 327
 328                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 329                         rCurve = new Rarefact(lookup, rDisplays);
 330                         rCurve->getSharedCurve(freq, nIters);
 331                         delete rCurve;
 332                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 333                 }
 334
 335                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 336
 337                 //reset groups parameter
 338                 globaldata->Groups.clear();
 339
 340                 if (m->control_pressed) {
 341                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 342                                 return 0;
 343                 }
 344
 345                 m->mothurOutEndLine();
 346                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 347                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 348                 m->mothurOutEndLine();
 349
 350                 return 0;
 351         }
 352         catch(exception& e) {
 353                 m->errorOut(e, "RareFactSharedCommand", "execute");
 354                 exit(1);
 355         }
 356 }
 357
 358
 359 //**********************************************************************************************************************