sharedcommand.cpp

   1 /*
   2  *  sharedcommand.cpp
   3  *  Dotur
   4  *
   5  *  Created by Sarah Westcott on 1/2/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "sharedcommand.h"
  11
  12 //**********************************************************************************************************************
  13
  14 SharedCommand::SharedCommand(){
  15         try {
  16                 globaldata = GlobalData::getInstance();
  17
  18                 //getting output filename
  19                 filename = globaldata->inputFileName;
  20                 filename = getRootName(filename);
  21                 filename = filename + "shared";
  22                 openOutputFile(filename, out);
  23
  24                 groupMap = globaldata->gGroupmap;
  25
  26                 //fill filehandles with neccessary ofstreams
  27                 int i;
  28                 ofstream* temp;
  29                 for (i=0; i<groupMap->getNumGroups(); i++) {
  30                         temp = new ofstream;
  31                         filehandles[groupMap->namesOfGroups[i]] = temp;
  32                 }
  33
  34                 //set fileroot
  35                 fileroot = getRootName(globaldata->getListFile());
  36
  37                 //clears file before we start to write to it below
  38                 for (int i=0; i<groupMap->getNumGroups(); i++) {
  39                         remove((fileroot + groupMap->namesOfGroups[i] + ".rabund").c_str());
  40                 }
  41
  42         }
  43         catch(exception& e) {
  44                 errorOut(e, "SharedCommand", "SharedCommand");
  45                 exit(1);
  46         }
  47 }
  48 //**********************************************************************************************************************
  49
  50 int SharedCommand::execute(){
  51         try {
  52
  53                 //lookup.clear();
  54                 string errorOff = "no error";
  55                 //errorOff = "";
  56
  57                 //read in listfile
  58                 read = new ReadOTUFile(globaldata->inputFileName);
  59                 read->read(&*globaldata);
  60                 delete read;
  61
  62                 input = globaldata->ginput;
  63                 SharedList = globaldata->gSharedList;
  64                 string lastLabel = SharedList->getLabel();
  65                 vector<SharedRAbundVector*> lookup;
  66
  67                 if (SharedList->getNumSeqs() != groupMap->getNumSeqs()) {
  68                         mothurOut("Your group file contains " + toString(groupMap->getNumSeqs()) + " sequences and list file contains " + toString(SharedList->getNumSeqs()) + " sequences. Please correct."); mothurOutEndLine();
  69
  70                         out.close();
  71                         remove(filename.c_str()); //remove blank shared file you made
  72
  73                         createMisMatchFile();
  74
  75                         //delete memory
  76                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
  77                                 delete it3->second;
  78                         }
  79                         delete SharedList;
  80                         globaldata->gSharedList = NULL;
  81
  82                         return 1;
  83                 }
  84
  85                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
  86                 set<string> processedLabels;
  87                 set<string> userLabels = globaldata->labels;
  88
  89                 while((SharedList != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0))) {
  90
  91
  92                         if(globaldata->allLines == 1 || globaldata->labels.count(SharedList->getLabel()) == 1){
  93
  94                                         lookup = SharedList->getSharedRAbundVector();
  95                                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
  96
  97                                         printSharedData(lookup); //prints info to the .shared file
  98                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
  99
 100                                         processedLabels.insert(SharedList->getLabel());
 101                                         userLabels.erase(SharedList->getLabel());
 102                         }
 103
 104                         if ((anyLabelsToProcess(SharedList->getLabel(), userLabels, errorOff) == true) && (processedLabels.count(lastLabel) != 1)) {
 105                                         string saveLabel = SharedList->getLabel();
 106
 107                                         delete SharedList;
 108                                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 109
 110                                         lookup = SharedList->getSharedRAbundVector();
 111                                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 112
 113                                         printSharedData(lookup); //prints info to the .shared file
 114                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 115
 116                                         processedLabels.insert(SharedList->getLabel());
 117                                         userLabels.erase(SharedList->getLabel());
 118
 119                                         //restore real lastlabel to save below
 120                                         SharedList->setLabel(saveLabel);
 121                         }
 122
 123
 124                         lastLabel = SharedList->getLabel();
 125
 126                         delete SharedList;
 127                         SharedList = input->getSharedListVector(); //get new list vector to process
 128                 }
 129
 130                 //output error messages about any remaining user labels
 131                 set<string>::iterator it;
 132                 bool needToRun = false;
 133                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 134                         if (processedLabels.count(lastLabel) != 1) {
 135                                 needToRun = true;
 136                         }
 137                 }
 138
 139                 //run last label if you need to
 140                 if (needToRun == true)  {
 141                         if (SharedList != NULL) {       delete SharedList;      }
 142                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 143
 144                         lookup = SharedList->getSharedRAbundVector();
 145                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 146
 147                         printSharedData(lookup); //prints info to the .shared file
 148                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 149                         delete SharedList;
 150                 }
 151
 152                 globaldata->gSharedList = NULL;
 153
 154                 out.close();
 155
 156                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 157                         delete it3->second;
 158                 }
 159
 160
 161                 return 0;
 162         }
 163         catch(exception& e) {
 164                 errorOut(e, "SharedCommand", "execute");
 165                 exit(1);
 166         }
 167 }
 168 //**********************************************************************************************************************
 169 void SharedCommand::printSharedData(vector<SharedRAbundVector*> thislookup) {
 170         try {
 171
 172                 //initialize bin values
 173                 for (int i = 0; i < thislookup.size(); i++) {
 174 //cout << "in printData " << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() <<  endl;
 175                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 176                         thislookup[i]->print(out);
 177
 178                         RAbundVector rav = thislookup[i]->getRAbundVector();
 179                         openOutputFileAppend(fileroot + thislookup[i]->getGroup() + ".rabund", *(filehandles[thislookup[i]->getGroup()]));
 180                         rav.print(*(filehandles[thislookup[i]->getGroup()]));
 181                         (*(filehandles[thislookup[i]->getGroup()])).close();
 182                 }
 183
 184         }
 185         catch(exception& e) {
 186                 errorOut(e, "SharedCommand", "printSharedData");
 187                 exit(1);
 188         }
 189 }
 190 //**********************************************************************************************************************
 191 void SharedCommand::createMisMatchFile() {
 192         try {
 193                 ofstream outMisMatch;
 194                 string outputMisMatchName = getRootName(globaldata->inputFileName);
 195
 196                 //you have sequences in your list file that are not in your group file
 197                 if (SharedList->getNumSeqs() > groupMap->getNumSeqs()) {
 198                         outputMisMatchName += "missing.group";
 199                         mothurOut("For a list of names that are in your list file and not in your group file, please refer to " + outputMisMatchName + "."); mothurOutEndLine();
 200
 201                         openOutputFile(outputMisMatchName, outMisMatch);
 202
 203                         //go through list and if group returns "not found" output it
 204                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 205
 206                                 string names = SharedList->get(i);
 207
 208                                 while (names.find_first_of(',') != -1) {
 209                                         string name = names.substr(0,names.find_first_of(','));
 210                                         names = names.substr(names.find_first_of(',')+1, names.length());
 211                                         string group = groupMap->getGroup(name);
 212
 213                                         if(group == "not found") {      outMisMatch << name << endl;  }
 214                                 }
 215
 216                                 //get last name
 217                                 string group = groupMap->getGroup(names);
 218                                 if(group == "not found") {      outMisMatch << names << endl;  }
 219                         }
 220
 221                         outMisMatch.close();
 222
 223
 224                 }else {//you have sequences in your group file that are not in you list file
 225
 226                         outputMisMatchName += "missing.name";
 227                         mothurOut("For a list of names that are in your group file and not in your list file, please refer to " + outputMisMatchName + "."); mothurOutEndLine();
 228
 229                         map<string, string> namesInList;
 230
 231                         //go through listfile and get names
 232                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 233
 234                                 string names = SharedList->get(i);
 235
 236                                 while (names.find_first_of(',') != -1) {
 237                                         string name = names.substr(0,names.find_first_of(','));
 238                                         names = names.substr(names.find_first_of(',')+1, names.length());
 239
 240                                         namesInList[name] = name;
 241                                 }
 242
 243                                 //get last name
 244                                 namesInList[names] = names;
 245                         }
 246
 247                         //get names of sequences in groupfile
 248                         vector<string> seqNames = groupMap->getNamesSeqs();
 249
 250                         map<string, string>::iterator itMatch;
 251
 252                         openOutputFile(outputMisMatchName, outMisMatch);
 253
 254                         //loop through names in seqNames and if they aren't in namesIn list output them
 255                         for (int i = 0; i < seqNames.size(); i++) {
 256
 257                                 itMatch = namesInList.find(seqNames[i]);
 258
 259                                 if (itMatch == namesInList.end()) {
 260
 261                                         outMisMatch << seqNames[i] << endl;
 262                                 }
 263                         }
 264                         outMisMatch.close();
 265                 }
 266
 267         }
 268         catch(exception& e) {
 269                 errorOut(e, "SharedCommand", "createMisMatchFile");
 270                 exit(1);
 271         }
 272 }
 273
 274 //**********************************************************************************************************************
 275
 276 SharedCommand::~SharedCommand(){
 277         //delete list;
 278
 279
 280 }
 281
 282 //**********************************************************************************************************************