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 GetRelAbundCommand::GetRelAbundCommand(){
  53         try {
  54                 abort = true; calledHelp = true;
  55                 setParameters();
  56                 vector<string> tempOutNames;
  57                 outputTypes["relabund"] = tempOutNames;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "GetRelAbundCommand", "GetRelAbundCommand");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65 GetRelAbundCommand::GetRelAbundCommand(string option) {
  66         try {
  67                 abort = false; calledHelp = false;
  68                 allLines = 1;
  69
  70                 //allow user to run help
  71                 if(option == "help") { help(); abort = true; calledHelp = true; }
  72                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  73
  74                 else {
  75                         vector<string> myArray = setParameters();
  76
  77                         OptionParser parser(option);
  78                         map<string,string> parameters = parser.getParameters();
  79                         map<string,string>::iterator it;
  80
  81                         ValidParameters validParameter;
  82
  83                         //check to make sure all parameters are valid for command
  84                         for (it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         //initialize outputTypes
  89                         vector<string> tempOutNames;
  90                         outputTypes["relabund"] = tempOutNames;
  91
  92                         //if the user changes the input directory command factory will send this info to us in the output parameter
  93                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  94                         if (inputDir == "not found"){   inputDir = "";          }
  95                         else {
  96                                 string path;
  97                                 it = parameters.find("shared");
  98                                 //user has given a template file
  99                                 if(it != parameters.end()){
 100                                         path = m->hasPath(it->second);
 101                                         //if the user has not given a path then, add inputdir. else leave path alone.
 102                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 103                                 }
 104                         }
 105
 106                         //get shared file
 107                         sharedfile = validParameter.validFile(parameters, "shared", true);
 108                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 109                         else if (sharedfile == "not found") {
 110                                 //if there is a current shared file, use it
 111                                 sharedfile = m->getSharedFile();
 112                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 113                                 else {  m->mothurOut("You have no current sharedfile and the shared parameter is required."); m->mothurOutEndLine(); abort = true; }
 114                         }
 115
 116
 117                         //if the user changes the output directory command factory will send this info to us in the output parameter
 118                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(sharedfile);             }
 119
 120                         //check for optional parameter and set defaults
 121                         // ...at some point should added some additional type checking...
 122                         label = validParameter.validFile(parameters, "label", false);
 123                         if (label == "not found") { label = ""; }
 124                         else {
 125                                 if(label != "all") {  m->splitAtDash(label, labels);  allLines = 0;  }
 126                                 else { allLines = 1;  }
 127                         }
 128
 129                         groups = validParameter.validFile(parameters, "groups", false);
 130                         if (groups == "not found") { groups = ""; pickedGroups = false; }
 131                         else {
 132                                 pickedGroups = true;
 133                                 m->splitAtDash(groups, Groups);
 134                                 m->Groups = Groups;
 135                         }
 136
 137                         scale = validParameter.validFile(parameters, "scale", false);                           if (scale == "not found") { scale = "totalgroup"; }
 138
 139                         if ((scale != "totalgroup") && (scale != "totalotu") && (scale != "averagegroup") && (scale != "averageotu")) {
 140                                 m->mothurOut(scale + " is not a valid scaling option for the get.relabund command. Choices are totalgroup, totalotu, averagegroup, averageotu."); m->mothurOutEndLine(); abort = true;
 141                         }
 142                 }
 143
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "GetRelAbundCommand", "GetRelAbundCommand");
 147                 exit(1);
 148         }
 149 }
 150 //**********************************************************************************************************************
 151
 152 int GetRelAbundCommand::execute(){
 153         try {
 154
 155                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 156
 157                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + "relabund";
 158                 ofstream out;
 159                 m->openOutputFile(outputFileName, out);
 160                 out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 161
 162                 input = new InputData(sharedfile, "sharedfile");
 163                 lookup = input->getSharedRAbundVectors();
 164                 string lastLabel = lookup[0]->getLabel();
 165
 166                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 167                 set<string> processedLabels;
 168                 set<string> userLabels = labels;
 169
 170                 //as long as you are not at the end of the file or done wih the lines you want
 171                 while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
 172
 173                         if (m->control_pressed) {  outputTypes.clear();  for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  } m->Groups.clear(); delete input;  out.close(); remove(outputFileName.c_str()); return 0; }
 174
 175                         if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
 176
 177                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 178                                 getRelAbundance(lookup, out);
 179
 180                                 processedLabels.insert(lookup[0]->getLabel());
 181                                 userLabels.erase(lookup[0]->getLabel());
 182                         }
 183
 184                         if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
 185                                 string saveLabel = lookup[0]->getLabel();
 186
 187                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 188                                 lookup = input->getSharedRAbundVectors(lastLabel);
 189                                 m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 190
 191                                 getRelAbundance(lookup, out);
 192
 193                                 processedLabels.insert(lookup[0]->getLabel());
 194                                 userLabels.erase(lookup[0]->getLabel());
 195
 196                                 //restore real lastlabel to save below
 197                                 lookup[0]->setLabel(saveLabel);
 198                         }
 199
 200                         lastLabel = lookup[0]->getLabel();
 201                         //prevent memory leak
 202                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i]; lookup[i] = NULL; }
 203
 204                         if (m->control_pressed) {  outputTypes.clear();  m->Groups.clear(); delete input;  out.close(); remove(outputFileName.c_str()); return 0; }
 205
 206                         //get next line to process
 207                         lookup = input->getSharedRAbundVectors();
 208                 }
 209
 210                 if (m->control_pressed) { outputTypes.clear(); m->Groups.clear(); delete input;  out.close(); remove(outputFileName.c_str());  return 0; }
 211
 212                 //output error messages about any remaining user labels
 213                 set<string>::iterator it;
 214                 bool needToRun = false;
 215                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 216                         m->mothurOut("Your file does not include the label " + *it);
 217                         if (processedLabels.count(lastLabel) != 1) {
 218                                 m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
 219                                 needToRun = true;
 220                         }else {
 221                                 m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
 222                         }
 223                 }
 224
 225                 //run last label if you need to
 226                 if (needToRun == true)  {
 227                         for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
 228                         lookup = input->getSharedRAbundVectors(lastLabel);
 229
 230                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 231
 232                         getRelAbundance(lookup, out);
 233
 234                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 235                 }
 236
 237                 //reset groups parameter
 238                 m->Groups.clear();
 239                 delete input;
 240                 out.close();
 241
 242                 if (m->control_pressed) { outputTypes.clear(); remove(outputFileName.c_str()); return 0;}
 243
 244                 m->mothurOutEndLine();
 245                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 246                 m->mothurOut(outputFileName); m->mothurOutEndLine(); outputNames.push_back(outputFileName); outputTypes["relabund"].push_back(outputFileName);
 247                 m->mothurOutEndLine();
 248
 249                 //set relabund file as new current relabundfile
 250                 string current = "";
 251                 itTypes = outputTypes.find("relabund");
 252                 if (itTypes != outputTypes.end()) {
 253                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setRelAbundFile(current); }
 254                 }
 255
 256                 return 0;
 257         }
 258         catch(exception& e) {
 259                 m->errorOut(e, "GetRelAbundCommand", "execute");
 260                 exit(1);
 261         }
 262 }
 263 //**********************************************************************************************************************
 264
 265 int GetRelAbundCommand::getRelAbundance(vector<SharedRAbundVector*>& thisLookUp, ofstream& out){
 266         try {
 267
 268                  for (int i = 0; i < thisLookUp.size(); i++) {
 269                         out << thisLookUp[i]->getLabel() << '\t' << thisLookUp[i]->getGroup() << '\t' << thisLookUp[i]->getNumBins() << '\t';
 270
 271                         for (int j = 0; j < thisLookUp[i]->getNumBins(); j++) {
 272
 273                                 if (m->control_pressed) { return 0; }
 274
 275                                 int abund = thisLookUp[i]->getAbundance(j);
 276
 277                                 float relabund = 0.0;
 278
 279                                 if (scale == "totalgroup") {
 280                                         relabund = abund / (float) thisLookUp[i]->getNumSeqs();
 281                                 }else if (scale == "totalotu") {
 282                                         //calc the total in this otu
 283                                         int totalOtu = 0;
 284                                         for (int l = 0; l < thisLookUp.size(); l++) {  totalOtu += thisLookUp[l]->getAbundance(j); }
 285                                         relabund = abund / (float) totalOtu;
 286                                 }else if (scale == "averagegroup") {
 287                                         relabund = abund / (float) (thisLookUp[i]->getNumSeqs() / (float) thisLookUp[i]->getNumBins());
 288                                 }else if (scale == "averageotu") {
 289                                         //calc the total in this otu
 290                                         int totalOtu = 0;
 291                                         for (int l = 0; l < thisLookUp.size(); l++) {  totalOtu += thisLookUp[l]->getAbundance(j); }
 292                                         float averageOtu = totalOtu / (float) thisLookUp.size();
 293
 294                                         relabund = abund / (float) averageOtu;
 295                                 }else{ m->mothurOut(scale + " is not a valid scaling option."); m->mothurOutEndLine(); m->control_pressed = true; return 0; }
 296
 297                                 out << relabund << '\t';
 298                         }
 299                         out << endl;
 300                  }
 301
 302                  return 0;
 303         }
 304         catch(exception& e) {
 305                 m->errorOut(e, "GetRelAbundCommand", "getRelAbundance");
 306                 exit(1);
 307         }
 308 }
 309 //**********************************************************************************************************************
 310
 311