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