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                         }
 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
 151                         groups = validParameter.validFile(parameters, "groups", false);
 152                         if (groups == "not found") { groups = ""; }
 153                         else {
 154                                 m->splitAtDash(groups, Groups);
 155                         }
 156                         m->Groups = Groups;
 157
 158                         string temp;
 159                         temp = validParameter.validFile(parameters, "freq", false);                     if (temp == "not found") { temp = "100"; }
 160                         convert(temp, freq);
 161
 162                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 163                         convert(temp, nIters);
 164
 165                         temp = validParameter.validFile(parameters, "jumble", false);                   if (temp == "not found") { temp = "T"; }
 166                         if (m->isTrue(temp)) { jumble = true; }
 167                         else { jumble = false; }
 168                         m->jumble = jumble;
 169
 170                 }
 171
 172         }
 173         catch(exception& e) {
 174                 m->errorOut(e, "RareFactSharedCommand", "RareFactSharedCommand");
 175                 exit(1);
 176         }
 177 }
 178 //**********************************************************************************************************************
 179
 180 int RareFactSharedCommand::execute(){
 181         try {
 182
 183                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 184
 185                 string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(sharedfile));
 186
 187                 ValidCalculators validCalculator;
 188
 189                 for (int i=0; i<Estimators.size(); i++) {
 190                         if (validCalculator.isValidCalculator("sharedrarefaction", Estimators[i]) == true) {
 191                                 if (Estimators[i] == "sharedobserved") {
 192                                         rDisplays.push_back(new RareDisplay(new SharedSobs(), new SharedThreeColumnFile(fileNameRoot+"shared.rarefaction", "")));
 193                                         outputNames.push_back(fileNameRoot+"shared.rarefaction"); outputTypes["sharedrarefaction"].push_back(fileNameRoot+"shared.rarefaction");
 194                                 }else if (Estimators[i] == "sharednseqs") {
 195                                         rDisplays.push_back(new RareDisplay(new SharedNSeqs(), new SharedThreeColumnFile(fileNameRoot+"shared.r_nseqs", "")));
 196                                         outputNames.push_back(fileNameRoot+"shared.r_nseqs"); outputTypes["sharedr_nseqs"].push_back(fileNameRoot+"shared.r_nseqs");
 197                                 }
 198                         }
 199                 }
 200
 201                 //if the users entered no valid calculators don't execute command
 202                 if (rDisplays.size() == 0) { return 0; }
 203
 204                 input = new InputData(sharedfile, "sharedfile");
 205                 lookup = input->getSharedRAbundVectors();
 206                 string lastLabel = lookup[0]->getLabel();
 207
 208                 if (m->control_pressed) {
 209                         m->Groups.clear();
 210                         delete input;
 211                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 212                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 213                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 214                         return 0;
 215                 }
 216
 217
 218                 if (lookup.size() < 2) {
 219                         m->mothurOut("I cannot run the command without at least 2 valid groups.");
 220                         for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
 221                         return 0;
 222                 }
 223
 224                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 225                 set<string> processedLabels;
 226                 set<string> userLabels = labels;
 227
 228                 //as long as you are not at the end of the file or done wih the lines you want
 229                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 230                         if (m->control_pressed) {
 231                                 m->Groups.clear();
 232                                 delete input;
 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                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 240                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 241                                 rCurve = new Rarefact(lookup, rDisplays);
 242                                 rCurve->getSharedCurve(freq, nIters);
 243                                 delete rCurve;
 244
 245                                 processedLabels.insert(lookup[0]->getLabel());
 246                                 userLabels.erase(lookup[0]->getLabel());
 247                         }
 248
 249                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 250                                         string saveLabel = lookup[0]->getLabel();
 251
 252                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 253                                         lookup = input->getSharedRAbundVectors(lastLabel);
 254
 255                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 256                                         rCurve = new Rarefact(lookup, rDisplays);
 257                                         rCurve->getSharedCurve(freq, nIters);
 258                                         delete rCurve;
 259
 260                                         processedLabels.insert(lookup[0]->getLabel());
 261                                         userLabels.erase(lookup[0]->getLabel());
 262
 263                                         //restore real lastlabel to save below
 264                                         lookup[0]->setLabel(saveLabel);
 265                         }
 266
 267
 268                         lastLabel = lookup[0]->getLabel();
 269
 270                         //get next line to process
 271                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 272                         lookup = input->getSharedRAbundVectors();
 273                 }
 274
 275                 if (m->control_pressed) {
 276                         m->Groups.clear();
 277                         delete input;
 278                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 279                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 280                         return 0;
 281                 }
 282
 283                 //output error messages about any remaining user labels
 284                 set<string>::iterator it;
 285                 bool needToRun = false;
 286                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 287                         m->mothurOut("Your file does not include the label " + *it);
 288                         if (processedLabels.count(lastLabel) != 1) {
 289                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 290                                 needToRun = true;
 291                         }else {
 292                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 293                         }
 294                 }
 295
 296                 if (m->control_pressed) {
 297                         m->Groups.clear();
 298                         delete input;
 299                         for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 300                         for (int i = 0; i < outputNames.size(); i++) {  remove(outputNames[i].c_str());         }
 301                         return 0;
 302                 }
 303
 304                 //run last label if you need to
 305                 if (needToRun == true)  {
 306                         for (int i = 0; i < lookup.size(); i++) {  if (lookup[i] != NULL) {     delete lookup[i]; }  }
 307                         lookup = input->getSharedRAbundVectors(lastLabel);
 308
 309                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 310                         rCurve = new Rarefact(lookup, rDisplays);
 311                         rCurve->getSharedCurve(freq, nIters);
 312                         delete rCurve;
 313                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 314                 }
 315
 316                 for(int i=0;i<rDisplays.size();i++){    delete rDisplays[i];    }
 317                 m->Groups.clear();
 318                 delete input;
 319
 320                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 321
 322                 m->mothurOutEndLine();
 323                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 324                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 325                 m->mothurOutEndLine();
 326
 327                 return 0;
 328         }
 329         catch(exception& e) {
 330                 m->errorOut(e, "RareFactSharedCommand", "execute");
 331                 exit(1);
 332         }
 333 }
 334
 335
 336 //**********************************************************************************************************************