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                         outputNames.push_back((fileroot + groups[i] + ".rabund"));
  53                 }
  54
  55         }
  56         catch(exception& e) {
  57                 m->errorOut(e, "SharedCommand", "SharedCommand");
  58                 exit(1);
  59         }
  60 }
  61 //**********************************************************************************************************************
  62
  63 int SharedCommand::execute(){
  64         try {
  65
  66                 //lookup.clear();
  67                 string errorOff = "no error";
  68                 //errorOff = "";
  69
  70                 //read in listfile
  71                 read = new ReadOTUFile(globaldata->inputFileName);
  72                 read->read(&*globaldata);
  73                 delete read;
  74
  75                 input = globaldata->ginput;
  76                 SharedList = globaldata->gSharedList;
  77                 string lastLabel = SharedList->getLabel();
  78                 vector<SharedRAbundVector*> lookup;
  79
  80                 if (m->control_pressed) {
  81                         delete input; delete SharedList; globaldata->ginput = NULL; globaldata->gSharedList = NULL;
  82                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {  delete it3->second;  }
  83                         out.close(); remove(filename.c_str());
  84                         for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
  85                         return 1;
  86                 }
  87
  88                 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
  89                         m->mothurOut("Your group file contains " + toString(groupMap->getNumSeqs()) + " sequences and list file contains " + toString(SharedList->getNumSeqs()) + " sequences. Please correct."); m->mothurOutEndLine();
  90
  91                         out.close();
  92                         remove(filename.c_str()); //remove blank shared file you made
  93
  94                         createMisMatchFile();
  95
  96                         //delete memory
  97                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
  98                                 delete it3->second;
  99                         }
 100                         delete input;
 101                         globaldata->ginput = NULL;
 102                         delete SharedList;
 103                         globaldata->gSharedList = NULL;
 104
 105                         return 1;
 106                 }
 107
 108                 //if user has specified groups make new groupfile for them
 109                 if (globaldata->Groups.size() != 0) { //make new group file
 110                         string groups = "";
 111                         for (int i = 0; i < globaldata->Groups.size(); i++) {
 112                                 groups += globaldata->Groups[i] + ".";
 113                         }
 114
 115                         string newGroupFile = outputDir + getRootName(getSimpleName(globaldata->inputFileName)) + groups + "groups";
 116                         ofstream outGroups;
 117                         openOutputFile(newGroupFile, outGroups);
 118
 119                         vector<string> names = groupMap->getNamesSeqs();
 120                         string groupName;
 121                         for (int i = 0; i < names.size(); i++) {
 122                                 groupName = groupMap->getGroup(names[i]);
 123                                 if (isValidGroup(groupName, globaldata->Groups)) {
 124                                         outGroups << names[i] << '\t' << groupName << endl;
 125                                 }
 126                         }
 127                         outGroups.close();
 128                 }
 129
 130                 //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
 131                 set<string> processedLabels;
 132                 set<string> userLabels = globaldata->labels;
 133
 134                 while((SharedList != NULL) && ((globaldata->allLines == 1) || (userLabels.size() != 0))) {
 135                         if (m->control_pressed) {
 136                                 delete input; delete SharedList; globaldata->ginput = NULL; globaldata->gSharedList = NULL;
 137                                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {  delete it3->second;  }
 138                                 out.close(); remove(filename.c_str());
 139                                 for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
 140                                 return 1;
 141                         }
 142
 143                         if(globaldata->allLines == 1 || globaldata->labels.count(SharedList->getLabel()) == 1){
 144
 145                                         lookup = SharedList->getSharedRAbundVector();
 146                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 147                                         if (pickedGroups) { //check for otus with no seqs in them
 148                                                 eliminateZeroOTUS(lookup);
 149                                         }
 150
 151                                         if (m->control_pressed) {
 152                                                 delete input; delete SharedList; globaldata->ginput = NULL; globaldata->gSharedList = NULL;
 153                                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 154                                                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {  delete it3->second;  }
 155                                                 out.close(); remove(filename.c_str());
 156                                                 for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
 157                                                 return 1;
 158                                         }
 159
 160                                         printSharedData(lookup); //prints info to the .shared file
 161                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 162
 163                                         processedLabels.insert(SharedList->getLabel());
 164                                         userLabels.erase(SharedList->getLabel());
 165                         }
 166
 167                         if ((anyLabelsToProcess(SharedList->getLabel(), userLabels, errorOff) == true) && (processedLabels.count(lastLabel) != 1)) {
 168                                         string saveLabel = SharedList->getLabel();
 169
 170                                         delete SharedList;
 171                                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 172
 173                                         lookup = SharedList->getSharedRAbundVector();
 174                                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 175                                         if (pickedGroups) { //check for otus with no seqs in them
 176                                                 eliminateZeroOTUS(lookup);
 177                                         }
 178
 179
 180                                         if (m->control_pressed) {
 181                                                 delete input; delete SharedList; globaldata->ginput = NULL; globaldata->gSharedList = NULL;
 182                                                 for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 183                                                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {  delete it3->second;  }
 184                                                 out.close(); remove(filename.c_str());
 185                                                 for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
 186                                                 return 1;
 187                                         }
 188
 189                                         printSharedData(lookup); //prints info to the .shared file
 190                                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 191
 192                                         processedLabels.insert(SharedList->getLabel());
 193                                         userLabels.erase(SharedList->getLabel());
 194
 195                                         //restore real lastlabel to save below
 196                                         SharedList->setLabel(saveLabel);
 197                         }
 198
 199
 200                         lastLabel = SharedList->getLabel();
 201
 202                         delete SharedList;
 203                         SharedList = input->getSharedListVector(); //get new list vector to process
 204                 }
 205
 206                 //output error messages about any remaining user labels
 207                 set<string>::iterator it;
 208                 bool needToRun = false;
 209                 for (it = userLabels.begin(); it != userLabels.end(); it++) {
 210                         if (processedLabels.count(lastLabel) != 1) {
 211                                 needToRun = true;
 212                         }
 213                 }
 214
 215                 //run last label if you need to
 216                 if (needToRun == true)  {
 217                         if (SharedList != NULL) {       delete SharedList;      }
 218                         SharedList = input->getSharedListVector(lastLabel); //get new list vector to process
 219
 220                         lookup = SharedList->getSharedRAbundVector();
 221                         m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
 222                         if (pickedGroups) { //check for otus with no seqs in them
 223                                 eliminateZeroOTUS(lookup);
 224                         }
 225
 226                         if (m->control_pressed) {
 227                                         delete input;  globaldata->ginput = NULL;
 228                                         for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {  delete it3->second;   }
 229                                         out.close(); remove(filename.c_str());
 230                                         for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
 231                                         return 1;
 232                         }
 233
 234                         printSharedData(lookup); //prints info to the .shared file
 235                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 236                         delete SharedList;
 237                 }
 238
 239                 globaldata->gSharedList = NULL;
 240
 241                 out.close();
 242
 243                 for (it3 = filehandles.begin(); it3 != filehandles.end(); it3++) {
 244                         delete it3->second;
 245                 }
 246
 247
 248                 //change format to shared  to speed up commands
 249                 globaldata->setFormat("sharedfile");
 250                 globaldata->setListFile("");
 251                 globaldata->setGroupFile("");
 252                 globaldata->setSharedFile(filename);
 253
 254                 if (m->control_pressed) {
 255                                 delete input;  globaldata->ginput = NULL;
 256                                 remove(filename.c_str());
 257                                 for (int i=0; i<groups.size(); i++) {  remove((fileroot + groups[i] + ".rabund").c_str());              }
 258                                 return 1;
 259                 }
 260
 261                 m->mothurOutEndLine();
 262                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 263                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 264                 m->mothurOut(filename); m->mothurOutEndLine();
 265                 m->mothurOutEndLine();
 266
 267                 return 0;
 268         }
 269         catch(exception& e) {
 270                 m->errorOut(e, "SharedCommand", "execute");
 271                 exit(1);
 272         }
 273 }
 274 //**********************************************************************************************************************
 275 void SharedCommand::printSharedData(vector<SharedRAbundVector*> thislookup) {
 276         try {
 277
 278                 //initialize bin values
 279                 for (int i = 0; i < thislookup.size(); i++) {
 280 //cout << "in printData " << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() <<  endl;
 281                         out << thislookup[i]->getLabel() << '\t' << thislookup[i]->getGroup() << '\t';
 282                         thislookup[i]->print(out);
 283
 284                         RAbundVector rav = thislookup[i]->getRAbundVector();
 285                         openOutputFileAppend(fileroot + thislookup[i]->getGroup() + ".rabund", *(filehandles[thislookup[i]->getGroup()]));
 286                         rav.print(*(filehandles[thislookup[i]->getGroup()]));
 287                         (*(filehandles[thislookup[i]->getGroup()])).close();
 288                 }
 289
 290         }
 291         catch(exception& e) {
 292                 m->errorOut(e, "SharedCommand", "printSharedData");
 293                 exit(1);
 294         }
 295 }
 296 //**********************************************************************************************************************
 297 int SharedCommand::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
 298         try {
 299
 300                 vector<SharedRAbundVector*> newLookup;
 301                 for (int i = 0; i < thislookup.size(); i++) {
 302                         SharedRAbundVector* temp = new SharedRAbundVector();
 303                         temp->setLabel(thislookup[i]->getLabel());
 304                         temp->setGroup(thislookup[i]->getGroup());
 305                         newLookup.push_back(temp);
 306                 }
 307
 308                 //for each bin
 309                 for (int i = 0; i < thislookup[0]->getNumBins(); i++) {
 310                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
 311
 312                         //look at each sharedRabund and make sure they are not all zero
 313                         bool allZero = true;
 314                         for (int j = 0; j < thislookup.size(); j++) {
 315                                 if (thislookup[j]->getAbundance(i) != 0) { allZero = false;  break;  }
 316                         }
 317
 318                         //if they are not all zero add this bin
 319                         if (!allZero) {
 320                                 for (int j = 0; j < thislookup.size(); j++) {
 321                                         newLookup[j]->push_back(thislookup[j]->getAbundance(i), thislookup[j]->getGroup());
 322                                 }
 323                         }
 324                         //else{  cout << "bin # " << i << " is all zeros" << endl;  }
 325                 }
 326
 327                 for (int j = 0; j < thislookup.size(); j++) {  delete thislookup[j];  }
 328                 thislookup = newLookup;
 329
 330                 return 0;
 331
 332         }
 333         catch(exception& e) {
 334                 m->errorOut(e, "SharedCommand", "eliminateZeroOTUS");
 335                 exit(1);
 336         }
 337 }
 338 //**********************************************************************************************************************
 339 int SharedCommand::createMisMatchFile() {
 340         try {
 341                 ofstream outMisMatch;
 342                 string outputMisMatchName = outputDir + getRootName(getSimpleName(globaldata->inputFileName));
 343
 344                 //you have sequences in your list file that are not in your group file
 345                 if (SharedList->getNumSeqs() > groupMap->getNumSeqs()) {
 346                         outputMisMatchName += "missing.group";
 347                         m->mothurOut("For a list of names that are in your list file and not in your group file, please refer to " + outputMisMatchName + "."); m->mothurOutEndLine();
 348
 349                         openOutputFile(outputMisMatchName, outMisMatch);
 350
 351                         map<string, string> listNames;
 352                         map<string, string>::iterator itList;
 353
 354                         //go through list and if group returns "not found" output it
 355                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 356                                 if (m->control_pressed) { outMisMatch.close(); remove(outputMisMatchName.c_str()); return 0; }
 357
 358                                 string names = SharedList->get(i);
 359
 360                                 while (names.find_first_of(',') != -1) {
 361                                         string name = names.substr(0,names.find_first_of(','));
 362                                         names = names.substr(names.find_first_of(',')+1, names.length());
 363                                         string group = groupMap->getGroup(name);
 364
 365                                         if(group == "not found") {      outMisMatch << name << endl;  }
 366
 367                                         itList = listNames.find(name);
 368                                         if (itList != listNames.end()) {  m->mothurOut(name + " is in your list file more than once.  Sequence names must be unique. please correct."); m->mothurOutEndLine(); }
 369                                         else { listNames[name] = name; }
 370                                 }
 371
 372                                 //get last name
 373                                 string group = groupMap->getGroup(names);
 374                                 if(group == "not found") {      outMisMatch << names << endl;  }
 375
 376                                 itList = listNames.find(names);
 377                                 if (itList != listNames.end()) {  m->mothurOut(names + " is in your list file more than once.  Sequence names must be unique. please correct."); m->mothurOutEndLine(); }
 378                                 else { listNames[names] = names; }
 379
 380                         }
 381
 382                         outMisMatch.close();
 383
 384
 385                 }else {//you have sequences in your group file that are not in you list file
 386
 387                         outputMisMatchName += "missing.name";
 388                         m->mothurOut("For a list of names that are in your group file and not in your list file, please refer to " + outputMisMatchName + "."); m->mothurOutEndLine();
 389
 390                         map<string, string> namesInList;
 391                         map<string, string>::iterator itList;
 392
 393                         //go through listfile and get names
 394                         for (int i = 0; i < SharedList->getNumBins(); i++) {
 395                                 if (m->control_pressed) {  return 0; }
 396
 397
 398                                 string names = SharedList->get(i);
 399
 400                                 while (names.find_first_of(',') != -1) {
 401                                         string name = names.substr(0,names.find_first_of(','));
 402                                         names = names.substr(names.find_first_of(',')+1, names.length());
 403
 404                                         itList = namesInList.find(name);
 405                                         if (itList != namesInList.end()) {  m->mothurOut(name + " is in your list file more than once.  Sequence names must be unique. please correct."); m->mothurOutEndLine(); }
 406
 407                                         namesInList[name] = name;
 408
 409                                 }
 410
 411                                 itList = namesInList.find(names);
 412                                 if (itList != namesInList.end()) {  m->mothurOut(names + " is in your list file more than once.  Sequence names must be unique. please correct."); m->mothurOutEndLine(); }
 413
 414                                 //get last name
 415                                 namesInList[names] = names;
 416                         }
 417
 418                         //get names of sequences in groupfile
 419                         vector<string> seqNames = groupMap->getNamesSeqs();
 420
 421                         map<string, string>::iterator itMatch;
 422
 423                         openOutputFile(outputMisMatchName, outMisMatch);
 424
 425                         //loop through names in seqNames and if they aren't in namesIn list output them
 426                         for (int i = 0; i < seqNames.size(); i++) {
 427                                 if (m->control_pressed) { outMisMatch.close(); remove(outputMisMatchName.c_str()); return 0; }
 428
 429                                 itMatch = namesInList.find(seqNames[i]);
 430
 431                                 if (itMatch == namesInList.end()) {
 432
 433                                         outMisMatch << seqNames[i] << endl;
 434                                 }
 435                         }
 436                         outMisMatch.close();
 437                 }
 438
 439                 return 0;
 440         }
 441         catch(exception& e) {
 442                 m->errorOut(e, "SharedCommand", "createMisMatchFile");
 443                 exit(1);
 444         }
 445 }
 446
 447 //**********************************************************************************************************************
 448
 449 SharedCommand::~SharedCommand(){
 450         //delete list;
 451
 452
 453 }
 454
 455 //**********************************************************************************************************************
 456
 457 bool SharedCommand::isValidGroup(string groupname, vector<string> groups) {
 458         try {
 459                 for (int i = 0; i < groups.size(); i++) {
 460                         if (groupname == groups[i]) { return true; }
 461                 }
 462
 463                 return false;
 464         }
 465         catch(exception& e) {
 466                 m->errorOut(e, "SharedCommand", "isValidGroup");
 467                 exit(1);
 468         }
 469 }
 470 /************************************************************/
 471
 472