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(string o) : outputDir(o) {
  15         try {
  16                 globaldata = GlobalData::getInstance();
  17
  18                 //getting output filename
  19                 filename = globaldata->inputFileName;
  20                 if (outputDir == "") { outputDir += hasPath(filename); }
  21
  22                 filename = outputDir + getRootName(getSimpleName(filename));
  23                 filename = filename + "shared";
  24
  25                 openOutputFile(filename, out);
  26                 pickedGroups = false;
  27
  28                 groupMap = globaldata->gGroupmap;
  29
  30                 //if hte user has not specified any groups then use them all
  31                 if (globaldata->Groups.size() == 0) {
  32                         groups = groupMap->namesOfGroups;
  33                 }else{ //they have specified groups
  34                         groups = globaldata->Groups;
  35                         pickedGroups = true;
  36                 }
  37
  38                 //fill filehandles with neccessary ofstreams
  39                 int i;
  40                 ofstream* temp;
  41                 for (i=0; i<groups.size(); i++) {
  42                         temp = new ofstream;
  43                         filehandles[groups[i]] = temp;
  44                 }
  45
  46                 //set fileroot
  47                 fileroot = outputDir + getRootName(getSimpleName(globaldata->getListFile()));
  48
  49                 //clears file before we start to write to it below
  50                 for (int i=0; i<groups.size(); i++) {
  51                         remove((fileroot + groups[i] + ".rabund").c_str());
  52                 }
  53
  54         }
  55         catch(exception& e) {
  56                 errorOut(e, "SharedCommand", "SharedCommand");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61
  62 int SharedCommand::execute(){
  63         try {
  64
  65                 //lookup.clear();
  66                 string errorOff = "no error";
  67                 //errorOff = "";
  68 cout << globaldata->inputFileName << endl;
  69                 //read in listfile
  70                 read = new ReadOTUFile(globaldata->inputFileName);
  71                 read->read(&*globaldata);
  72                 delete read;
  73
  74                 input = globaldata->ginput;
  75                 SharedList = globaldata->gSharedList;
  76                 string lastLabel = SharedList->getLabel();
  77                 vector<SharedRAbundVector*> lookup;
  78
  79                 if ((globaldata->Groups.size() == 0) && (SharedList->getNumSeqs() != groupMap->getNumSeqs())) {  //if the user has not specified any groups and their files don't match exit with error
  80                         mothurOut("Your group file contains " + toString(groupMap->getNumSeqs()) + " sequences and list file contains " + toString(SharedList->getNumSeqs()) + " sequences. Please correct."); mothurOutEndLine();
  81
  82                         out.close();
  83                         remove(filename.c_str()); //remove blank shared file you made
  84
  85                         createMisMatchFile();
  86
  87                         //delete memory
  88                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
  89                                 delete it3->second;
  90                         }
  91                         delete SharedList;
  92                         globaldata->gSharedList = NULL;
  93
  94                         return 1;
  95                 }
  96
  97                 //if user has specified groups make new groupfile for them
  98                 if (globaldata->Groups.size() != 0) { //make new group file
  99                         string groups = "";
 100                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 101                                 groups += globaldata->Groups[i] + ".";
 102                         }
 103
 104                         string newGroupFile = getRootName(globaldata->inputFileName) + groups + "groups";
 105                         ofstream outGroups;
 106                         openOutputFile(newGroupFile, outGroups);
 107
 108                         vector<string> names = groupMap->getNamesSeqs();
 109                         string groupName;
 110                         for (int i = 0; i < names.size(); i++) {
 111                                 groupName = groupMap->getGroup(names[i]);
 112                                 if (isValidGroup(groupName, globaldata->Groups)) {
 113                                         outGroups << names[i] << '\t' << groupName << endl;
 114                                 }
 115                         }
 116                         outGroups.close();
 117                 }
 118
 119                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 120                 set<string> processedLabels;
 121                 set<string> userLabels = globaldata->labels;
 122
 123                 while((SharedList != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0))) {
 124
 125                         if(globaldata->allLines == 1 || globaldata->labels.count(SharedList->getLabel()) == 1){
 126
 127                                         lookup = SharedList->getSharedRAbundVector();
 128                                         if (pickedGroups) { //check for otus with no seqs in them
 129                                                 eliminateZeroOTUS(lookup);
 130                                         }
 131                                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 132
 133                                         printSharedData(lookup); //prints info to the .shared file
 134                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 135
 136                                         processedLabels.insert(SharedList->getLabel());
 137                                         userLabels.erase(SharedList->getLabel());
 138                         }
 139
 140                         if ((anyLabelsToProcess(SharedList->getLabel(), userLabels, errorOff) == true) && (processedLabels.count(lastLabel) != 1)) {
 141                                         string saveLabel = SharedList->getLabel();
 142
 143                                         delete SharedList;
 144                                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 145
 146                                         lookup = SharedList->getSharedRAbundVector();
 147                                         if (pickedGroups) { //check for otus with no seqs in them
 148                                                 eliminateZeroOTUS(lookup);
 149                                         }
 150                                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 151
 152                                         printSharedData(lookup); //prints info to the .shared file
 153                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 154
 155                                         processedLabels.insert(SharedList->getLabel());
 156                                         userLabels.erase(SharedList->getLabel());
 157
 158                                         //restore real lastlabel to save below
 159                                         SharedList->setLabel(saveLabel);
 160                         }
 161
 162
 163                         lastLabel = SharedList->getLabel();
 164
 165                         delete SharedList;
 166                         SharedList = input->getSharedListVector(); //get new list vector to process
 167                 }
 168
 169                 //output error messages about any remaining user labels
 170                 set<string>::iterator it;
 171                 bool needToRun = false;
 172                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 173                         if (processedLabels.count(lastLabel) != 1) {
 174                                 needToRun = true;
 175                         }
 176                 }
 177
 178                 //run last label if you need to
 179                 if (needToRun == true)  {
 180                         if (SharedList != NULL) {       delete SharedList;      }
 181                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 182
 183                         lookup = SharedList->getSharedRAbundVector();
 184                         if (pickedGroups) { //check for otus with no seqs in them
 185                                 eliminateZeroOTUS(lookup);
 186                         }
 187                         mothurOut(lookup[0]->getLabel()); mothurOutEndLine();
 188
 189                         printSharedData(lookup); //prints info to the .shared file
 190                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 191                         delete SharedList;
 192                 }
 193
 194                 globaldata->gSharedList = NULL;
 195
 196                 out.close();
 197
 198                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 199                         delete it3->second;
 200                 }
 201
 202                 globaldata->setSharedFile(filename);
 203
 204                 return 0;
 205         }
 206         catch(exception& e) {
 207                 errorOut(e, "SharedCommand", "execute");
 208                 exit(1);
 209         }
 210 }
 211 //**********************************************************************************************************************
 212 void SharedCommand::printSharedData(vector<SharedRAbundVector*> thislookup) {
 213         try {
 214
 215                 //initialize bin values
 216                 for (int i = 0; i < thislookup.size(); i++) {
 217 //cout << "in printData " << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() <<  endl;
 218                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 219                         thislookup[i]->print(out);
 220
 221                         RAbundVector rav = thislookup[i]->getRAbundVector();
 222                         openOutputFileAppend(fileroot + thislookup[i]->getGroup() + ".rabund", *(filehandles[thislookup[i]->getGroup()]));
 223                         rav.print(*(filehandles[thislookup[i]->getGroup()]));
 224                         (*(filehandles[thislookup[i]->getGroup()])).close();
 225                 }
 226
 227         }
 228         catch(exception& e) {
 229                 errorOut(e, "SharedCommand", "printSharedData");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234 void SharedCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
 235         try {
 236
 237                 vector<SharedRAbundVector*> newLookup;
 238                 for (int i = 0; i < thislookup.size(); i++) {
 239                         SharedRAbundVector* temp = new SharedRAbundVector();
 240                         temp->setLabel(thislookup[i]->getLabel());
 241                         temp->setGroup(thislookup[i]->getGroup());
 242                         newLookup.push_back(temp);
 243                 }
 244
 245                 //for each bin
 246                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
 247
 248                         //look at each sharedRabund and make sure they are not all zero
 249                         bool allZero = true;
 250                         for (int j = 0; j < thislookup.size(); j++) {
 251                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
 252                         }
 253
 254                         //if they are not all zero add this bin
 255                         if (!allZero) {
 256                                 for (int j = 0; j < thislookup.size(); j++) {
 257                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
 258                                 }
 259                         }
 260                         //else{  cout << "bin # " << i << " is all zeros" << endl;  }
 261                 }
 262
 263                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
 264                 thislookup = newLookup;
 265
 266
 267         }
 268         catch(exception& e) {
 269                 errorOut(e, "SharedCommand", "eliminateZeroOTUS");
 270                 exit(1);
 271         }
 272 }
 273 //**********************************************************************************************************************
 274 void SharedCommand::createMisMatchFile() {
 275         try {
 276                 ofstream outMisMatch;
 277                 string outputMisMatchName = getRootName(globaldata->inputFileName);
 278
 279                 //you have sequences in your list file that are not in your group file
 280                 if (SharedList->getNumSeqs() > groupMap->getNumSeqs()) {
 281                         outputMisMatchName += "missing.group";
 282                         mothurOut("For a list of names that are in your list file and not in your group file, please refer to " + outputMisMatchName + "."); mothurOutEndLine();
 283
 284                         openOutputFile(outputMisMatchName, outMisMatch);
 285
 286                         //go through list and if group returns "not found" output it
 287                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 288
 289                                 string names = SharedList->get(i);
 290
 291                                 while (names.find_first_of(',') != -1) {
 292                                         string name = names.substr(0,names.find_first_of(','));
 293                                         names = names.substr(names.find_first_of(',')+1, names.length());
 294                                         string group = groupMap->getGroup(name);
 295
 296                                         if(group == "not found") {      outMisMatch << name << endl;  }
 297                                 }
 298
 299                                 //get last name
 300                                 string group = groupMap->getGroup(names);
 301                                 if(group == "not found") {      outMisMatch << names << endl;  }
 302                         }
 303
 304                         outMisMatch.close();
 305
 306
 307                 }else {//you have sequences in your group file that are not in you list file
 308
 309                         outputMisMatchName += "missing.name";
 310                         mothurOut("For a list of names that are in your group file and not in your list file, please refer to " + outputMisMatchName + "."); mothurOutEndLine();
 311
 312                         map<string, string> namesInList;
 313
 314                         //go through listfile and get names
 315                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 316
 317                                 string names = SharedList->get(i);
 318
 319                                 while (names.find_first_of(',') != -1) {
 320                                         string name = names.substr(0,names.find_first_of(','));
 321                                         names = names.substr(names.find_first_of(',')+1, names.length());
 322
 323                                         namesInList[name] = name;
 324                                 }
 325
 326                                 //get last name
 327                                 namesInList[names] = names;
 328                         }
 329
 330                         //get names of sequences in groupfile
 331                         vector<string> seqNames = groupMap->getNamesSeqs();
 332
 333                         map<string, string>::iterator itMatch;
 334
 335                         openOutputFile(outputMisMatchName, outMisMatch);
 336
 337                         //loop through names in seqNames and if they aren't in namesIn list output them
 338                         for (int i = 0; i < seqNames.size(); i++) {
 339
 340                                 itMatch = namesInList.find(seqNames[i]);
 341
 342                                 if (itMatch == namesInList.end()) {
 343
 344                                         outMisMatch << seqNames[i] << endl;
 345                                 }
 346                         }
 347                         outMisMatch.close();
 348                 }
 349
 350         }
 351         catch(exception& e) {
 352                 errorOut(e, "SharedCommand", "createMisMatchFile");
 353                 exit(1);
 354         }
 355 }
 356
 357 //**********************************************************************************************************************
 358
 359 SharedCommand::~SharedCommand(){
 360         //delete list;
 361
 362
 363 }
 364
 365 //**********************************************************************************************************************
 366
 367 bool SharedCommand::isValidGroup(string groupname, vector<string> groups) {
 368         try {
 369                 for (int i = 0; i < groups.size(); i++) {
 370                         if (groupname == groups[i]) { return true; }
 371                 }
 372
 373                 return false;
 374         }
 375         catch(exception& e) {
 376                 errorOut(e, "SharedCommand", "isValidGroup");
 377                 exit(1);
 378         }
 379 }
 380 /************************************************************/
 381
 382