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","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, "iters", false);                    if (temp == "not found") { temp = "1000"; }
  88                         convert(temp, nIters);
  89
  90                         temp = validParameter.validFile(parameters, "jumble", false);                   if (temp == "not found") { temp = "T"; }
  91                         if (isTrue(temp)) { jumble = true; }
  92                         else { jumble = false; }
  93                         globaldata->jumble = jumble;
  94
  95                         if (abort == false) {
  96
  97                                 string fileNameRoot = outputDir + getRootName(getSimpleName(globaldata->inputFileName));
  98 //                              format = globaldata->getFormat();
  99
 100
 101                                 validCalculator = new ValidCalculators();
 102
 103                                 for (int i=0; i<Estimators.size(); i++) {
 104                                         if (validCalculator->isValidCalculator("sharedrarefaction", Estimators[i]) == true) {
 105                                                 if (Estimators[i] == "sharedobserved") {
 106                                                         rDisplays.push_back(new RareDisplay(new SharedSobs(), new SharedThreeColumnFile(fileNameRoot+"shared.rarefaction", "")));
 107                                                         outputNames.push_back(fileNameRoot+"shared.rarefaction");
 108                                                 }else if (Estimators[i] == "sharednseqs") {
 109                                                         rDisplays.push_back(new RareDisplay(new SharedNSeqs(), new SharedThreeColumnFile(fileNameRoot+"shared.r_nseqs", "")));
 110                                                         outputNames.push_back(fileNameRoot+"shared.r_nseqs");
 111                                                 }
 112                                         }
 113                                 }
 114                         }
 115
 116                 }
 117
 118         }
 119         catch(exception& e) {
 120                 m->errorOut(e, "RareFactSharedCommand", "RareFactSharedCommand");
 121                 exit(1);
 122         }
 123 }
 124
 125 //**********************************************************************************************************************
 126
 127 void RareFactSharedCommand::help(){
 128         try {
 129                 m->mothurOut("The rarefaction.shared command can only be executed after a successful read.otu command.\n");
 130                 m->mothurOut("The rarefaction.shared command parameters are label, iters, groups, jumble and calc.  No parameters are required.\n");
 131                 m->mothurOut("The rarefaction command should be in the following format: \n");
 132                 m->mothurOut("rarefaction.shared(label=yourLabel, iters=yourIters, calc=yourEstimators, jumble=yourJumble, groups=yourGroups).\n");
 133                 m->mothurOut("Example rarefaction.shared(label=unique-0.01-0.03,  iters=10000, groups=B-C, jumble=T, calc=sharedobserved).\n");
 134                 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");
 135                 m->mothurOut("The default value for groups is all the groups in your groupfile, and jumble is true.\n");
 136                 validCalculator->printCalc("sharedrarefaction", cout);
 137                 m->mothurOut("The label parameter is used to analyze specific labels in your input.\n");
 138                 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");
 139                 m->mothurOut("Note: No spaces between parameter labels (i.e. freq), '=' and parameters (i.e.yourFreq).\n\n");
 140         }
 141         catch(exception& e) {
 142                 m->errorOut(e, "RareFactSharedCommand", "help");
 143                 exit(1);
 144         }
 145 }
 146
 147 //**********************************************************************************************************************
 148
 149 RareFactSharedCommand::~RareFactSharedCommand(){
 150         if (abort == false) {
 151                 delete input;   globaldata->ginput = NULL;
 152                 delete read;
 153                 delete validCalculator;
 154         }
 155 }
 156
 157 //**********************************************************************************************************************
 158
 159 int RareFactSharedCommand::execute(){
 160         try {
 161
 162                 if (abort == true) { return 0; }
 163
 164                 //if the users entered no valid calculators don't execute command
 165                 if (rDisplays.size() == 0) { return 0; }
 166
 167                 read = new ReadOTUFile(globaldata->inputFileName);
 168                 read->read(&*globaldata);
 169
 170                 input = globaldata->ginput;
 171                 lookup = input->getSharedRAbundVectors();
 172                 string lastLabel = lookup[0]->getLabel();
 173
 174                 if (m->control_pressed) {
 175                         globaldata->Groups.clear();
 176                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 177                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 178                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 179                         return 0;
 180                 }
 181
 182
 183                 if (lookup.size() < 2) {
 184                         m->mothurOut("I cannot run the command without at least 2 valid groups.");
 185                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 186                         return 0;
 187                 }
 188
 189                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 190                 set<string> processedLabels;
 191                 set<string> userLabels = labels;
 192
 193                 //as long as you are not at the end of the file or done wih the lines you want
 194                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 195                         if (m->control_pressed) {
 196                                 globaldata->Groups.clear();
 197                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 198                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 199                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 200                                 return 0;
 201                         }
 202
 203                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 204                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 205                                 rCurve = new Rarefact(lookup, rDisplays);
 206                                 rCurve->getSharedCurve(freq, nIters);
 207                                 delete rCurve;
 208
 209                                 processedLabels.insert(lookup[0]->getLabel());
 210                                 userLabels.erase(lookup[0]->getLabel());
 211                         }
 212
 213                         if ((anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 214                                         string saveLabel = lookup[0]->getLabel();
 215
 216                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 217                                         lookup = input->getSharedRAbundVectors(lastLabel);
 218
 219                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 220                                         rCurve = new Rarefact(lookup, rDisplays);
 221                                         rCurve->getSharedCurve(freq, nIters);
 222                                         delete rCurve;
 223
 224                                         processedLabels.insert(lookup[0]->getLabel());
 225                                         userLabels.erase(lookup[0]->getLabel());
 226
 227                                         //restore real lastlabel to save below
 228                                         lookup[0]->setLabel(saveLabel);
 229                         }
 230
 231
 232                         lastLabel = lookup[0]->getLabel();
 233
 234                         //get next line to process
 235                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 236                         lookup = input->getSharedRAbundVectors();
 237                 }
 238
 239                 if (m->control_pressed) {
 240                                 globaldata->Groups.clear();
 241                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 242                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 243                                 return 0;
 244                 }
 245
 246                 //output error messages about any remaining user labels
 247                 set<string>::iterator it;
 248                 bool needToRun = false;
 249                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 250                         m->mothurOut("Your file does not include the label " + *it);
 251                         if (processedLabels.count(lastLabel) != 1) {
 252                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 253                                 needToRun = true;
 254                         }else {
 255                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 256                         }
 257                 }
 258
 259                 if (m->control_pressed) {
 260                                 globaldata->Groups.clear();
 261                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 262                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 263                                 return 0;
 264                 }
 265
 266                 //run last label if you need to
 267                 if (needToRun == true)  {
 268                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i]; }  }
 269                         lookup = input->getSharedRAbundVectors(lastLabel);
 270
 271                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 272                         rCurve = new Rarefact(lookup, rDisplays);
 273                         rCurve->getSharedCurve(freq, nIters);
 274                         delete rCurve;
 275                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 276                 }
 277
 278                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 279
 280                 //reset groups parameter
 281                 globaldata->Groups.clear();
 282
 283                 if (m->control_pressed) {
 284                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 285                                 return 0;
 286                 }
 287
 288                 m->mothurOutEndLine();
 289                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 290                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 291                 m->mothurOutEndLine();
 292
 293                 return 0;
 294         }
 295         catch(exception& e) {
 296                 m->errorOut(e, "RareFactSharedCommand", "execute");
 297                 exit(1);
 298         }
 299 }
 300
 301
 302 //**********************************************************************************************************************