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
  85                 else {
  86                         vector<string> myArray = setParameters();
  87
  88                         OptionParser parser(option);
  89                         map<string,string> parameters = parser.getParameters();
  90                         map<string,string>::iterator it;
  91
  92                         ValidParameters validParameter;
  93
  94                         //check to make sure all parameters are valid for command
  95                         for (it = parameters.begin(); it != parameters.end(); it++) {
  96                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  97                         }
  98
  99                         //initialize outputTypes
 100                         vector<string> tempOutNames;
 101                         outputTypes["sharedrarefaction"] = tempOutNames;
 102                         outputTypes["sharedr_nseqs"] = tempOutNames;
 103
 104                         //if the user changes the input directory command factory will send this info to us in the output parameter
 105                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 106                         if (inputDir == "not found"){   inputDir = "";          }
 107                         else {
 108                                 string path;
 109                                 it = parameters.find("shared");
 110                                 //user has given a template file
 111                                 if(it != parameters.end()){
 112                                         path = m->hasPath(it->second);
 113                                         //if the user has not given a path then, add inputdir. else leave path alone.
 114                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 115                                 }
 116                         }
 117
 118                         //get shared file
 119                         sharedfile = validParameter.validFile(parameters, "shared", true);
 120                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 121                         else if (sharedfile == "not found") {
 122                                 //if there is a current shared file, use it
 123                                 sharedfile = m->getSharedFile();
 124                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 125                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 126                         }
 127
 128
 129                         //if the user changes the output directory command factory will send this info to us in the output parameter
 130                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 131
 132
 133                         //check for optional parameter and set defaults
 134                         // ...at some point should added some additional type checking...
 135                         label = validParameter.validFile(parameters, "label", false);
 136                         if (label == "not found") { label = ""; }
 137                         else {
 138                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 139                                 else { allLines = 1;  }
 140                         }
 141
 142
 143                         calc = validParameter.validFile(parameters, "calc", false);
 144                         if (calc == "not found") { calc = "sharedobserved";  }
 145                         else {
 146                                  if (calc == "default")  {  calc = "sharedobserved";  }
 147                         }
 148                         m->splitAtDash(calc, Estimators);
 149
 150                         groups = validParameter.validFile(parameters, "groups", false);
 151                         if (groups == "not found") { groups = ""; }
 152                         else {
 153                                 m->splitAtDash(groups, Groups);
 154                         }
 155                         m->Groups = Groups;
 156
 157                         string temp;
 158                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 159                         convert(temp, freq);
 160
 161                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 162                         convert(temp, nIters);
 163
 164                         temp = validParameter.validFile(parameters, "jumble", false);                   if (temp == "not found") { temp = "T"; }
 165                         if (m->isTrue(temp)) { jumble = true; }
 166                         else { jumble = false; }
 167                         m->jumble = jumble;
 168
 169                 }
 170
 171         }
 172         catch(exception& e) {
 173                 m->errorOut(e, "RareFactSharedCommand", "RareFactSharedCommand");
 174                 exit(1);
 175         }
 176 }
 177 //**********************************************************************************************************************
 178
 179 int RareFactSharedCommand::execute(){
 180         try {
 181
 182                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 183
 184                 string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 185
 186                 ValidCalculators validCalculator;
 187
 188                 for (int i=0; i<Estimators.size(); i++) {
 189                         if (validCalculator.isValidCalculator("sharedrarefaction", Estimators[i]) == true) {
 190                                 if (Estimators[i] == "sharedobserved") {
 191                                         rDisplays.push_back(new RareDisplay(new SharedSobs(), new SharedThreeColumnFile(fileNameRoot+"shared.rarefaction", "")));
 192                                         outputNames.push_back(fileNameRoot+"shared.rarefaction"); outputTypes["sharedrarefaction"].push_back(fileNameRoot+"shared.rarefaction");
 193                                 }else if (Estimators[i] == "sharednseqs") {
 194                                         rDisplays.push_back(new RareDisplay(new SharedNSeqs(), new SharedThreeColumnFile(fileNameRoot+"shared.r_nseqs", "")));
 195                                         outputNames.push_back(fileNameRoot+"shared.r_nseqs"); outputTypes["sharedr_nseqs"].push_back(fileNameRoot+"shared.r_nseqs");
 196                                 }
 197                         }
 198                 }
 199
 200                 //if the users entered no valid calculators don't execute command
 201                 if (rDisplays.size() == 0) { return 0; }
 202
 203                 input = new InputData(sharedfile, "sharedfile");
 204                 lookup = input->getSharedRAbundVectors();
 205                 string lastLabel = lookup[0]->getLabel();
 206
 207                 if (m->control_pressed) {
 208                         m->Groups.clear();
 209                         delete input;
 210                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 211                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 212                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 213                         return 0;
 214                 }
 215
 216
 217                 if (lookup.size() < 2) {
 218                         m->mothurOut("I cannot run the command without at least 2 valid groups.");
 219                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 220                         return 0;
 221                 }
 222
 223                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 224                 set<string> processedLabels;
 225                 set<string> userLabels = labels;
 226
 227                 //as long as you are not at the end of the file or done wih the lines you want
 228                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 229                         if (m->control_pressed) {
 230                                 m->Groups.clear();
 231                                 delete input;
 232                                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 233                                 for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 234                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 235                                 return 0;
 236                         }
 237
 238                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 239                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 240                                 rCurve = new Rarefact(lookup, rDisplays);
 241                                 rCurve->getSharedCurve(freq, nIters);
 242                                 delete rCurve;
 243
 244                                 processedLabels.insert(lookup[0]->getLabel());
 245                                 userLabels.erase(lookup[0]->getLabel());
 246                         }
 247
 248                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 249                                         string saveLabel = lookup[0]->getLabel();
 250
 251                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 252                                         lookup = input->getSharedRAbundVectors(lastLabel);
 253
 254                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 255                                         rCurve = new Rarefact(lookup, rDisplays);
 256                                         rCurve->getSharedCurve(freq, nIters);
 257                                         delete rCurve;
 258
 259                                         processedLabels.insert(lookup[0]->getLabel());
 260                                         userLabels.erase(lookup[0]->getLabel());
 261
 262                                         //restore real lastlabel to save below
 263                                         lookup[0]->setLabel(saveLabel);
 264                         }
 265
 266
 267                         lastLabel = lookup[0]->getLabel();
 268
 269                         //get next line to process
 270                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 271                         lookup = input->getSharedRAbundVectors();
 272                 }
 273
 274                 if (m->control_pressed) {
 275                         m->Groups.clear();
 276                         delete input;
 277                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 278                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 279                         return 0;
 280                 }
 281
 282                 //output error messages about any remaining user labels
 283                 set<string>::iterator it;
 284                 bool needToRun = false;
 285                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 286                         m->mothurOut("Your file does not include the label " + *it);
 287                         if (processedLabels.count(lastLabel) != 1) {
 288                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 289                                 needToRun = true;
 290                         }else {
 291                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 292                         }
 293                 }
 294
 295                 if (m->control_pressed) {
 296                         m->Groups.clear();
 297                         delete input;
 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                 //run last label if you need to
 304                 if (needToRun == true)  {
 305                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i]; }  }
 306                         lookup = input->getSharedRAbundVectors(lastLabel);
 307
 308                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 309                         rCurve = new Rarefact(lookup, rDisplays);
 310                         rCurve->getSharedCurve(freq, nIters);
 311                         delete rCurve;
 312                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 313                 }
 314
 315                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 316                 m->Groups.clear();
 317                 delete input;
 318
 319                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 320
 321                 m->mothurOutEndLine();
 322                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 323                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 324                 m->mothurOutEndLine();
 325
 326                 return 0;
 327         }
 328         catch(exception& e) {
 329                 m->errorOut(e, "RareFactSharedCommand", "execute");
 330                 exit(1);
 331         }
 332 }
 333
 334
 335 //**********************************************************************************************************************