sequenceparser.cpp

   1 /*
   2  *  sequenceParser.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/9/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sequenceparser.h"
  11
  12
  13 /************************************************************/
  14 SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFile) {
  15         try {
  16
  17                 m = MothurOut::getInstance();
  18                 int error;
  19
  20                 //read group file
  21                 groupMap = new GroupMap(groupFile);
  22                 error = groupMap->readMap();
  23
  24                 if (error == 1) { m->control_pressed = true; }
  25
  26                 //initialize maps
  27                 vector<string> namesOfGroups = groupMap->getNamesOfGroups();
  28                 for (int i = 0; i < namesOfGroups.size(); i++) {
  29                         vector<Sequence> temp;
  30                         map<string, string> tempMap;
  31                         seqs[namesOfGroups[i]] = temp;
  32                         nameMapPerGroup[namesOfGroups[i]] = tempMap;
  33                 }
  34
  35                 //read fasta file making sure each sequence is in the group file
  36                 ifstream in;
  37                 m->openInputFile(fastaFile, in);
  38
  39                 map<string, string> seqName; //stores name -> sequence string so we can make new "unique" sequences when we parse the name file
  40         int fastaCount = 0;
  41                 while (!in.eof()) {
  42
  43                         if (m->control_pressed) { break; }
  44
  45                         Sequence seq(in); m->gobble(in);
  46             fastaCount++;
  47             if (m->debug) { if((fastaCount) % 1000 == 0){       m->mothurOut("[DEBUG]: reading seq " + toString(fastaCount) + "\n.");   } }
  48
  49         if (seq.getName() != "") {
  50
  51                                  string group = groupMap->getGroup(seq.getName());
  52                                  if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
  53                                  else {
  54                                          seqs[group].push_back(seq);
  55                                          seqName[seq.getName()] = seq.getAligned();
  56                                  }
  57                         }
  58                 }
  59                 in.close();
  60
  61                 if (error == 1) { m->control_pressed = true; }
  62
  63                 //read name file
  64                 ifstream inName;
  65                 m->openInputFile(nameFile, inName);
  66
  67                 //string first, second;
  68                 int countName = 0;
  69                 set<string> thisnames1;
  70
  71         string rest = "";
  72         char buffer[4096];
  73         bool pairDone = false;
  74         bool columnOne = true;
  75         string firstCol, secondCol;
  76
  77                 while (!inName.eof()) {
  78                         if (m->control_pressed) { break; }
  79
  80             inName.read(buffer, 4096);
  81             vector<string> pieces = m->splitWhiteSpace(rest, buffer, inName.gcount());
  82
  83             for (int i = 0; i < pieces.size(); i++) {
  84                 if (columnOne) {  firstCol = pieces[i]; columnOne=false; }
  85                 else  { secondCol = pieces[i]; pairDone = true; columnOne=true; }
  86
  87                 if (pairDone) { //save one line
  88                     if (m->debug) { m->mothurOut("[DEBUG]: reading names: " + firstCol + '\t' + secondCol + ".\n"); }
  89                     vector<string> names;
  90                     m->splitAtChar(secondCol, names, ',');
  91
  92                     //get aligned string for these seqs from the fasta file
  93                     string alignedString = "";
  94                     map<string, string>::iterator itAligned = seqName.find(names[0]);
  95                     if (itAligned == seqName.end()) {
  96                         error = 1; m->mothurOut("[ERROR]: " + names[0] + " is in your name file and not in your fasta file, please correct."); m->mothurOutEndLine();
  97                     }else {
  98                         alignedString = itAligned->second;
  99                     }
 100
 101                     //separate by group - parse one line in name file
 102                     map<string, string> splitMap; //group -> name1,name2,...
 103                     map<string, string>::iterator it;
 104                     for (int i = 0; i < names.size(); i++) {
 105
 106                         string group = groupMap->getGroup(names[i]);
 107                         if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + names[i] + " is in your name file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
 108                         else {
 109
 110                             it = splitMap.find(group);
 111                             if (it != splitMap.end()) { //adding seqs to this group
 112                                 (it->second) += "," + names[i];
 113                                 thisnames1.insert(names[i]);
 114                                 countName++;
 115                             }else { //first sighting of this group
 116                                 splitMap[group] = names[i];
 117                                 countName++;
 118                                 thisnames1.insert(names[i]);
 119
 120                                 //is this seq in the fasta file?
 121                                 if (i != 0) { //if not then we need to add a duplicate sequence to the seqs for this group so the new "fasta" and "name" files will match
 122                                     Sequence tempSeq(names[i], alignedString); //get the first guys sequence string since he's in the fasta file.
 123                                     seqs[group].push_back(tempSeq);
 124                                 }
 125                             }
 126                         }
 127
 128                         allSeqsMap[names[i]] = names[0];
 129                     }
 130
 131
 132                     //fill nameMapPerGroup - holds all lines in namefile separated by group
 133                     for (it = splitMap.begin(); it != splitMap.end(); it++) {
 134                         //grab first name
 135                         string firstName = "";
 136                         for(int i = 0; i < (it->second).length(); i++) {
 137                             if (((it->second)[i]) != ',') {
 138                                 firstName += ((it->second)[i]);
 139                             }else { break; }
 140                         }
 141
 142                         //group1 -> seq1 -> seq1,seq2,seq3
 143                         nameMapPerGroup[it->first][firstName] = it->second;
 144                     }
 145
 146                     pairDone = false;
 147                 }
 148             }
 149                 }
 150                 inName.close();
 151
 152                 if (error == 1) { m->control_pressed = true; }
 153
 154                 if (countName != (groupMap->getNumSeqs())) {
 155                         vector<string> groupseqsnames = groupMap->getNamesSeqs();
 156
 157                         for (int i = 0; i < groupseqsnames.size(); i++) {
 158                                 set<string>::iterator itnamesfile = thisnames1.find(groupseqsnames[i]);
 159                                 if (itnamesfile == thisnames1.end()){
 160                                         cout << "missing name " + groupseqsnames[i] << '\t' << allSeqsMap[groupseqsnames[i]] << endl;
 161                                 }
 162                         }
 163
 164                         m->mothurOutEndLine();
 165                         m->mothurOut("[ERROR]: Your name file contains " + toString(countName) + " valid sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 166                         m->mothurOutEndLine();
 167                         m->control_pressed = true;
 168                 }
 169
 170         }
 171         catch(exception& e) {
 172                 m->errorOut(e, "SequenceParser", "SequenceParser");
 173                 exit(1);
 174         }
 175 }
 176 /************************************************************/
 177 SequenceParser::SequenceParser(string groupFile, string fastaFile) {
 178         try {
 179
 180                 m = MothurOut::getInstance();
 181                 int error;
 182
 183                 //read group file
 184                 groupMap = new GroupMap(groupFile);
 185                 error = groupMap->readMap();
 186
 187                 if (error == 1) { m->control_pressed = true; }
 188
 189                 //initialize maps
 190                 vector<string> namesOfGroups = groupMap->getNamesOfGroups();
 191                 for (int i = 0; i < namesOfGroups.size(); i++) {
 192                         vector<Sequence> temp;
 193                         seqs[namesOfGroups[i]] = temp;
 194                 }
 195
 196                 //read fasta file making sure each sequence is in the group file
 197                 ifstream in;
 198                 m->openInputFile(fastaFile, in);
 199                 int count = 0;
 200
 201                 while (!in.eof()) {
 202
 203                         if (m->control_pressed) { break; }
 204
 205                         Sequence seq(in); m->gobble(in);
 206
 207                         if (seq.getName() != "") {
 208
 209                                 string group = groupMap->getGroup(seq.getName());
 210                                 if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + seq.getName() + " is in your fasta file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
 211                                 else {  seqs[group].push_back(seq);     count++; }
 212                         }
 213                 }
 214                 in.close();
 215
 216                 if (error == 1) { m->control_pressed = true; }
 217
 218                 if (count != (groupMap->getNumSeqs())) {
 219                         m->mothurOutEndLine();
 220                         m->mothurOut("[ERROR]: Your fasta file contains " + toString(count) + " valid sequences, and your groupfile contains " + toString(groupMap->getNumSeqs()) + ", please correct.");
 221                         if (count < (groupMap->getNumSeqs())) { m->mothurOut(" Did you forget to include the name file?"); }
 222                         m->mothurOutEndLine();
 223                         m->control_pressed = true;
 224                 }
 225
 226         }
 227         catch(exception& e) {
 228                 m->errorOut(e, "SequenceParser", "SequenceParser");
 229                 exit(1);
 230         }
 231 }
 232 /************************************************************/
 233 SequenceParser::~SequenceParser(){ delete groupMap; }
 234 /************************************************************/
 235 int SequenceParser::getNumGroups(){ return groupMap->getNumGroups(); }
 236 /************************************************************/
 237 vector<string> SequenceParser::getNamesOfGroups(){ return groupMap->getNamesOfGroups(); }
 238 /************************************************************/
 239 bool SequenceParser::isValidGroup(string g){ return groupMap->isValidGroup(g); }
 240 /************************************************************/
 241 string SequenceParser::getGroup(string g){ return groupMap->getGroup(g); }
 242 /************************************************************/
 243 int SequenceParser::getNumSeqs(string g){
 244         try {
 245                 map<string, vector<Sequence> >::iterator it;
 246                 int num = 0;
 247
 248                 it = seqs.find(g);
 249                 if(it == seqs.end()) {
 250                         m->mothurOut("[ERROR]: " + g + " is not a valid group, please correct."); m->mothurOutEndLine();
 251                 }else {
 252                         num = (it->second).size();
 253                 }
 254
 255                 return num;
 256         }
 257         catch(exception& e) {
 258                 m->errorOut(e, "SequenceParser", "getNumSeqs");
 259                 exit(1);
 260         }
 261 }
 262 /************************************************************/
 263 vector<Sequence> SequenceParser::getSeqs(string g){
 264         try {
 265                 map<string, vector<Sequence> >::iterator it;
 266                 vector<Sequence> seqForThisGroup;
 267
 268                 it = seqs.find(g);
 269                 if(it == seqs.end()) {
 270                         m->mothurOut("[ERROR]: No sequences available for group " + g + ", please correct."); m->mothurOutEndLine();
 271                 }else {
 272                         seqForThisGroup = it->second;
 273             if (m->debug) {  m->mothurOut("[DEBUG]: group " + g + " fasta file has " + toString(seqForThisGroup.size()) + " sequences.");  }
 274                 }
 275
 276                 return seqForThisGroup;
 277         }
 278         catch(exception& e) {
 279                 m->errorOut(e, "SequenceParser", "getSeqs");
 280                 exit(1);
 281         }
 282 }
 283 /************************************************************/
 284 int SequenceParser::getSeqs(string g, string filename, bool uchimeFormat=false){
 285         try {
 286                 map<string, vector<Sequence> >::iterator it;
 287                 vector<Sequence> seqForThisGroup;
 288                 vector<seqPriorityNode> nameVector;
 289
 290                 it = seqs.find(g);
 291                 if(it == seqs.end()) {
 292                         m->mothurOut("[ERROR]: No sequences available for group " + g + ", please correct."); m->mothurOutEndLine();
 293                 }else {
 294
 295                         ofstream out;
 296                         m->openOutputFile(filename, out);
 297
 298                         seqForThisGroup = it->second;
 299
 300                         if (uchimeFormat) {
 301                                 // format should look like
 302                                 //>seqName /ab=numRedundantSeqs/
 303                                 //sequence
 304
 305                                 map<string, string> nameMapForThisGroup = getNameMap(g);
 306                                 map<string, string>::iterator itNameMap;
 307                                 int error = 0;
 308
 309                                 for (int i = 0; i < seqForThisGroup.size(); i++) {
 310                                         itNameMap = nameMapForThisGroup.find(seqForThisGroup[i].getName());
 311
 312                                         if (itNameMap == nameMapForThisGroup.end()){
 313                                                 error = 1;
 314                                                 m->mothurOut("[ERROR]: " + seqForThisGroup[i].getName() + " is in your fastafile, but is not in your namesfile, please correct."); m->mothurOutEndLine();
 315                                         }else {
 316                                                 int num = m->getNumNames(itNameMap->second);
 317
 318                                                 seqPriorityNode temp(num, seqForThisGroup[i].getAligned(), seqForThisGroup[i].getName());
 319                                                 nameVector.push_back(temp);
 320                                         }
 321                                 }
 322
 323                                 if (error == 1) { out.close(); m->mothurRemove(filename); return 1; }
 324
 325                                 //sort by num represented
 326                                 sort(nameVector.begin(), nameVector.end(), compareSeqPriorityNodes);
 327
 328                                 //print new file in order of
 329                                 for (int i = 0; i < nameVector.size(); i++) {
 330
 331                                         if(m->control_pressed) { out.close(); m->mothurRemove(filename); return 1; }
 332
 333                                         out << ">" << nameVector[i].name  << "/ab=" << nameVector[i].numIdentical << "/" << endl << nameVector[i].seq << endl;
 334                                 }
 335
 336                         }else {
 337                 //m->mothurOut("Group " + g +  " contains " + toString(seqForThisGroup.size()) + " unique seqs.\n");
 338                                 for (int i = 0; i < seqForThisGroup.size(); i++) {
 339
 340                                         if(m->control_pressed) { out.close(); m->mothurRemove(filename); return 1; }
 341
 342                                         seqForThisGroup[i].printSequence(out);
 343                                 }
 344                         }
 345                         out.close();
 346                 }
 347
 348                 return 0;
 349         }
 350         catch(exception& e) {
 351                 m->errorOut(e, "SequenceParser", "getSeqs");
 352                 exit(1);
 353         }
 354 }
 355
 356 /************************************************************/
 357 map<string, string> SequenceParser::getNameMap(string g){
 358         try {
 359                 map<string, map<string, string> >::iterator it;
 360                 map<string, string> nameMapForThisGroup;
 361
 362                 it = nameMapPerGroup.find(g);
 363                 if(it == nameMapPerGroup.end()) {
 364                         m->mothurOut("[ERROR]: No nameMap available for group " + g + ", please correct."); m->mothurOutEndLine();
 365                 }else {
 366                         nameMapForThisGroup = it->second;
 367             if (m->debug) {  m->mothurOut("[DEBUG]: group " + g + " name file has " + toString(nameMapForThisGroup.size()) + " unique sequences.");  }
 368                 }
 369
 370                 return nameMapForThisGroup;
 371         }
 372         catch(exception& e) {
 373                 m->errorOut(e, "SequenceParser", "getNameMap");
 374                 exit(1);
 375         }
 376 }
 377 /************************************************************/
 378 int SequenceParser::getNameMap(string g, string filename){
 379         try {
 380                 map<string, map<string, string> >::iterator it;
 381                 map<string, string> nameMapForThisGroup;
 382
 383                 it = nameMapPerGroup.find(g);
 384                 if(it == nameMapPerGroup.end()) {
 385                         m->mothurOut("[ERROR]: No nameMap available for group " + g + ", please correct."); m->mothurOutEndLine();
 386                 }else {
 387                         nameMapForThisGroup = it->second;
 388
 389                         ofstream out;
 390                         m->openOutputFile(filename, out);
 391
 392                         for (map<string, string>::iterator itFile = nameMapForThisGroup.begin(); itFile != nameMapForThisGroup.end(); itFile++) {
 393
 394                                 if(m->control_pressed) { out.close(); m->mothurRemove(filename); return 1; }
 395
 396                                 out << itFile->first << '\t' << itFile->second << endl;
 397                         }
 398
 399                         out.close();
 400                 }
 401
 402                 return 0;
 403         }
 404         catch(exception& e) {
 405                 m->errorOut(e, "SequenceParser", "getNameMap");
 406                 exit(1);
 407         }
 408 }
 409 /************************************************************/
 410
 411
 412