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