getrelabundcommand.cpp

   1 /*
   2  *  getrelabundcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 6/21/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getrelabundcommand.h"
  11
  12 //**********************************************************************************************************************
  13 vector<string> GetRelAbundCommand::setParameters(){
  14         try {
  15                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pshared);
  16                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  17                 CommandParameter pscale("scale", "Multiple", "totalgroup-totalotu-averagegroup-averageotu", "totalgroup", "", "", "",false,false); parameters.push_back(pscale);
  18                 CommandParameter plabel("label", "String", "", "", "", "", "",false,false); parameters.push_back(plabel);
  19                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  20                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  21
  22                 vector<string> myArray;
  23                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  24                 return myArray;
  25         }
  26         catch(exception& e) {
  27                 m->errorOut(e, "GetRelAbundCommand", "setParameters");
  28                 exit(1);
  29         }
  30 }
  31 //**********************************************************************************************************************
  32 string GetRelAbundCommand::getHelpString(){
  33         try {
  34                 string helpString = "";
  35                 helpString += "The get.relabund command parameters are shared, groups, scale and label.  shared is required, unless you have a valid current file.\n";
  36                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like included. The group names are separated by dashes.\n";
  37                 helpString += "The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n";
  38                 helpString += "The scale parameter allows you to select what scale you would like to use. Choices are totalgroup, totalotu, averagegroup, averageotu, default is totalgroup.\n";
  39                 helpString += "The get.relabund command should be in the following format: get.relabund(groups=yourGroups, label=yourLabels).\n";
  40                 helpString += "Example get.relabund(groups=A-B-C, scale=averagegroup).\n";
  41                 helpString += "The default value for groups is all the groups in your groupfile, and all labels in your inputfile will be used.\n";
  42                 helpString += "The get.relabund command outputs a .relabund file.\n";
  43                 helpString += "Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n";
  44                 return helpString;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "GetRelAbundCommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 string GetRelAbundCommand::getOutputFileNameTag(string type, string inputName=""){
  53         try {
  54         string outputFileName = "";
  55                 map<string, vector<string> >::iterator it;
  56
  57         //is this a type this command creates
  58         it = outputTypes.find(type);
  59         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  60         else {
  61             if (type == "relabund")         {   outputFileName = "relabund" ;  }
  62             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  63         }
  64         return outputFileName;
  65         }
  66         catch(exception& e) {
  67                 m->errorOut(e, "GetRelAbundCommand", "getOutputFileNameTag");
  68                 exit(1);
  69         }
  70 }
  71 //**********************************************************************************************************************
  72 GetRelAbundCommand::GetRelAbundCommand(){
  73         try {
  74                 abort = true; calledHelp = true;
  75                 setParameters();
  76                 vector<string> tempOutNames;
  77                 outputTypes["relabund"] = tempOutNames;
  78         }
  79         catch(exception& e) {
  80                 m->errorOut(e, "GetRelAbundCommand", "GetRelAbundCommand");
  81                 exit(1);
  82         }
  83 }
  84 //**********************************************************************************************************************
  85 GetRelAbundCommand::GetRelAbundCommand(string option) {
  86         try {
  87                 abort = false; calledHelp = false;
  88                 allLines = 1;
  89
  90                 //allow user to run help
  91                 if(option == "help") { help(); abort = true; calledHelp = true; }
  92                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  93
  94                 else {
  95                         vector<string> myArray = setParameters();
  96
  97                         OptionParser parser(option);
  98                         map<string,string> parameters = parser.getParameters();
  99                         map<string,string>::iterator it;
 100
 101                         ValidParameters validParameter;
 102
 103                         //check to make sure all parameters are valid for command
 104                         for (it = parameters.begin(); it != parameters.end(); it++) {
 105                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 106                         }
 107
 108                         //initialize outputTypes
 109                         vector<string> tempOutNames;
 110                         outputTypes["relabund"] = tempOutNames;
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("shared");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 123                                 }
 124                         }
 125
 126                         //get shared file
 127                         sharedfile = validParameter.validFile(parameters, "shared", true);
 128                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 129                         else if (sharedfile == "not found") {
 130                                 //if there is a current shared file, use it
 131                                 sharedfile = m->getSharedFile();
 132                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 133                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 134                         }else { m->setSharedFile(sharedfile); }
 135
 136
 137                         //if the user changes the output directory command factory will send this info to us in the output parameter
 138                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 139
 140                         //check for optional parameter and set defaults
 141                         // ...at some point should added some additional type checking...
 142                         label = validParameter.validFile(parameters, "label", false);
 143                         if (label == "not found") { label = ""; }
 144                         else {
 145                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 146                                 else { allLines = 1;  }
 147                         }
 148
 149                         groups = validParameter.validFile(parameters, "groups", false);
 150                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 151                         else {
 152                                 pickedGroups = true;
 153                                 m->splitAtDash(groups, Groups);
 154                                 m->setGroups(Groups);
 155                         }
 156
 157                         scale = validParameter.validFile(parameters, "scale", false);                           if (scale == "not found") { scale = "totalgroup"; }
 158
 159                         if ((scale != "totalgroup") && (scale != "totalotu") && (scale != "averagegroup") && (scale != "averageotu")) {
 160                                 m->mothurOut(scale + " is not a valid scaling option for the get.relabund command. Choices are totalgroup, totalotu, averagegroup, averageotu."); m->mothurOutEndLine(); abort = true;
 161                         }
 162                 }
 163
 164         }
 165         catch(exception& e) {
 166                 m->errorOut(e, "GetRelAbundCommand", "GetRelAbundCommand");
 167                 exit(1);
 168         }
 169 }
 170 //**********************************************************************************************************************
 171
 172 int GetRelAbundCommand::execute(){
 173         try {
 174
 175                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 176
 177                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + getOutputFileNameTag("relabund");
 178                 ofstream out;
 179                 m->openOutputFile(outputFileName, out);
 180                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 181
 182                 input = new InputData(sharedfile, "sharedfile");
 183                 lookup = input->getSharedRAbundVectors();
 184                 string lastLabel = lookup[0]->getLabel();
 185
 186                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 187                 set<string> processedLabels;
 188                 set<string> userLabels = labels;
 189
 190                 //as long as you are not at the end of the file or done wih the lines you want
 191                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 192
 193                         if (m->control_pressed) {  outputTypes.clear();  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->clearGroups(); delete input;  out.close(); m->mothurRemove(outputFileName); return 0; }
 194
 195                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 196
 197                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 198                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 199                                 getRelAbundance(lookup, out);
 200
 201                                 processedLabels.insert(lookup[0]->getLabel());
 202                                 userLabels.erase(lookup[0]->getLabel());
 203                         }
 204
 205                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 206                                 string saveLabel = lookup[0]->getLabel();
 207
 208                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 209                                 lookup = input->getSharedRAbundVectors(lastLabel);
 210                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 211                                 if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 212                                 getRelAbundance(lookup, out);
 213
 214                                 processedLabels.insert(lookup[0]->getLabel());
 215                                 userLabels.erase(lookup[0]->getLabel());
 216
 217                                 //restore real lastlabel to save below
 218                                 lookup[0]->setLabel(saveLabel);
 219                         }
 220
 221                         lastLabel = lookup[0]->getLabel();
 222                         //prevent memory leak
 223                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 224
 225                         if (m->control_pressed) {  outputTypes.clear();  m->clearGroups(); delete input;  out.close(); m->mothurRemove(outputFileName); return 0; }
 226
 227                         //get next line to process
 228                         lookup = input->getSharedRAbundVectors();
 229                 }
 230
 231                 if (m->control_pressed) { outputTypes.clear(); m->clearGroups(); delete input;  out.close(); m->mothurRemove(outputFileName);  return 0; }
 232
 233                 //output error messages about any remaining user labels
 234                 set<string>::iterator it;
 235                 bool needToRun = false;
 236                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 237                         m->mothurOut("Your file does not include the label " + *it);
 238                         if (processedLabels.count(lastLabel) != 1) {
 239                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 240                                 needToRun = true;
 241                         }else {
 242                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 243                         }
 244                 }
 245
 246                 //run last label if you need to
 247                 if (needToRun == true)  {
 248                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 249                         lookup = input->getSharedRAbundVectors(lastLabel);
 250
 251                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 252                         if (!m->printedHeaders) { lookup[0]->printHeaders(out); }
 253                         getRelAbundance(lookup, out);
 254
 255                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 256                 }
 257
 258                 //reset groups parameter
 259                 m->clearGroups();
 260                 delete input;
 261                 out.close();
 262
 263                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(outputFileName); return 0;}
 264
 265                 m->mothurOutEndLine();
 266                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 267                 m->mothurOut(outputFileName); m->mothurOutEndLine(); outputNames.push_back(outputFileName); outputTypes["relabund"].push_back(outputFileName);
 268                 m->mothurOutEndLine();
 269
 270                 //set relabund file as new current relabundfile
 271                 string current = "";
 272                 itTypes = outputTypes.find("relabund");
 273                 if (itTypes != outputTypes.end()) {
 274                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRelAbundFile(current); }
 275                 }
 276
 277                 return 0;
 278         }
 279         catch(exception& e) {
 280                 m->errorOut(e, "GetRelAbundCommand", "execute");
 281                 exit(1);
 282         }
 283 }
 284 //**********************************************************************************************************************
 285
 286 int GetRelAbundCommand::getRelAbundance(vector<SharedRAbundVector*>& thisLookUp, ofstream& out){
 287         try {
 288
 289                  for (int i = 0; i < thisLookUp.size(); i++) {
 290                         out << thisLookUp[i]->getLabel() << '\t' << thisLookUp[i]->getGroup() << '\t' << thisLookUp[i]->getNumBins() << '\t';
 291
 292                         for (int j = 0; j < thisLookUp[i]->getNumBins(); j++) {
 293
 294                                 if (m->control_pressed) { return 0; }
 295
 296                                 int abund = thisLookUp[i]->getAbundance(j);
 297
 298                                 float relabund = 0.0;
 299
 300                                 if (scale == "totalgroup") {
 301                                         relabund = abund / (float) thisLookUp[i]->getNumSeqs();
 302                                 }else if (scale == "totalotu") {
 303                                         //calc the total in this otu
 304                                         int totalOtu = 0;
 305                                         for (int l = 0; l < thisLookUp.size(); l++) {  totalOtu += thisLookUp[l]->getAbundance(j); }
 306                                         relabund = abund / (float) totalOtu;
 307                                 }else if (scale == "averagegroup") {
 308                                         relabund = abund / (float) (thisLookUp[i]->getNumSeqs() / (float) thisLookUp[i]->getNumBins());
 309                                 }else if (scale == "averageotu") {
 310                                         //calc the total in this otu
 311                                         int totalOtu = 0;
 312                                         for (int l = 0; l < thisLookUp.size(); l++) {  totalOtu += thisLookUp[l]->getAbundance(j); }
 313                                         float averageOtu = totalOtu / (float) thisLookUp.size();
 314
 315                                         relabund = abund / (float) averageOtu;
 316                                 }else{ m->mothurOut(scale + " is not a valid scaling option."); m->mothurOutEndLine(); m->control_pressed = true; return 0; }
 317
 318                                 out << relabund << '\t';
 319                         }
 320                         out << endl;
 321                  }
 322
 323                  return 0;
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "GetRelAbundCommand", "getRelAbundance");
 327                 exit(1);
 328         }
 329 }
 330 //**********************************************************************************************************************
 331
 332