makecontigscommand.cpp

   1 //
   2 //  makecontigscommand.cpp
   3 //  Mothur
   4 //
   5 //  Created by Sarah Westcott on 5/15/12.
   6 //  Copyright (c) 2012 Schloss Lab. All rights reserved.
   7 //
   8
   9 #include "makecontigscommand.h"
  10
  11 //**********************************************************************************************************************
  12 vector<string> MakeContigsCommand::setParameters(){
  13         try {
  14                 CommandParameter pfasta("ffastq", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  15         CommandParameter prfasta("rfastq", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(prfasta);
  16                 CommandParameter palign("align", "Multiple", "needleman-gotoh-blast-noalign", "needleman", "", "", "",false,false); parameters.push_back(palign);
  17                 CommandParameter pmatch("match", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pmatch);
  18                 CommandParameter pmismatch("mismatch", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pmismatch);
  19                 CommandParameter pgapopen("gapopen", "Number", "", "-2.0", "", "", "",false,false); parameters.push_back(pgapopen);
  20                 CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend);
  21                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  22                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  23                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  24
  25                 vector<string> myArray;
  26                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  27                 return myArray;
  28         }
  29         catch(exception& e) {
  30                 m->errorOut(e, "MakeContigsCommand", "setParameters");
  31                 exit(1);
  32         }
  33 }
  34 //**********************************************************************************************************************
  35 string MakeContigsCommand::getHelpString(){
  36         try {
  37                 string helpString = "";
  38                 helpString += "The make.contigs command reads a forward fastq file and a reverse fastq file and outputs new fasta and quality files.\n";
  39                 helpString += "The make.contigs command parameters are ffastq, rfastq, align, match, mismatch, gapopen, gapextend and processors.\n";
  40                 helpString += "The ffastq and rfastq parameter is required.\n";
  41                 helpString += "The align parameter allows you to specify the alignment method to use.  Your options are: gotoh, needleman, blast and noalign. The default is needleman.\n";
  42                 helpString += "The match parameter allows you to specify the bonus for having the same base. The default is 1.0.\n";
  43                 helpString += "The mistmatch parameter allows you to specify the penalty for having different bases.  The default is -1.0.\n";
  44                 helpString += "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -2.0.\n";
  45                 helpString += "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment.  The default is -1.0.\n";
  46         helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  47         helpString += "The make.contigs command should be in the following format: \n";
  48                 helpString += "make.contigs(ffastq=yourForwardFastqFile, rfastq=yourReverseFastqFile, align=yourAlignmentMethod) \n";
  49                 helpString += "Note: No spaces between parameter labels (i.e. ffastq), '=' and parameters (i.e.yourForwardFastqFile).\n";
  50                 return helpString;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "MakeContigsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57
  58 //**********************************************************************************************************************
  59 MakeContigsCommand::MakeContigsCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63                 vector<string> tempOutNames;
  64                 outputTypes["fasta"] = tempOutNames;
  65                 outputTypes["qfile"] = tempOutNames;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "MakeContigsCommand", "MakeContigsCommand");
  69                 exit(1);
  70         }
  71 }
  72 //**********************************************************************************************************************
  73 MakeContigsCommand::MakeContigsCommand(string option)  {
  74         try {
  75                 abort = false; calledHelp = false;
  76
  77                 //allow user to run help
  78                 if(option == "help") { help(); abort = true; calledHelp = true; }
  79                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  80
  81                 else {
  82                         vector<string> myArray = setParameters();
  83
  84                         OptionParser parser(option);
  85                         map<string, string> parameters = parser.getParameters();
  86
  87                         ValidParameters validParameter("pairwise.seqs");
  88                         map<string, string>::iterator it;
  89
  90                         //check to make sure all parameters are valid for command
  91                         for (it = parameters.begin(); it != parameters.end(); it++) {
  92                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  93                         }
  94
  95                         //initialize outputTypes
  96                         vector<string> tempOutNames;
  97                         outputTypes["fasta"] = tempOutNames;
  98                         outputTypes["qfile"] = tempOutNames;
  99
 100
 101                         //if the user changes the input directory command factory will send this info to us in the output parameter
 102                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 103                         if (inputDir == "not found"){   inputDir = "";          }
 104                         else {
 105                                 string path;
 106                 it = parameters.find("ffastq");
 107                                 //user has given a template file
 108                                 if(it != parameters.end()){
 109                                         path = m->hasPath(it->second);
 110                                         //if the user has not given a path then, add inputdir. else leave path alone.
 111                                         if (path == "") {       parameters["ffastq"] = inputDir + it->second;           }
 112                                 }
 113
 114                 it = parameters.find("rfastq");
 115                                 //user has given a template file
 116                                 if(it != parameters.end()){
 117                                         path = m->hasPath(it->second);
 118                                         //if the user has not given a path then, add inputdir. else leave path alone.
 119                                         if (path == "") {       parameters["rfastq"] = inputDir + it->second;           }
 120                                 }
 121             }
 122
 123             ffastqfile = validParameter.validFile(parameters, "ffastq", true);
 124                         if (ffastqfile == "not open") { ffastqfile = ""; abort = true; }
 125                         else if (ffastqfile == "not found") { ffastqfile = ""; abort=true;  m->mothurOut("The ffastq parameter is required.\n"); }
 126
 127                         rfastqfile = validParameter.validFile(parameters, "rfastq", true);
 128                         if (rfastqfile == "not open") { rfastqfile = ""; abort = true; }
 129                         else if (rfastqfile == "not found") { rfastqfile = ""; abort=true;  m->mothurOut("The rfastq parameter is required.\n"); }
 130
 131             //if the user changes the output directory command factory will send this info to us in the output parameter
 132                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(ffastqfile);             }
 133
 134
 135                         //check for optional parameter and set defaults
 136                         // ...at some point should added some additional type checking...
 137                         string temp;
 138                         temp = validParameter.validFile(parameters, "match", false);            if (temp == "not found"){       temp = "1.0";                   }
 139                         m->mothurConvert(temp, match);
 140
 141                         temp = validParameter.validFile(parameters, "mismatch", false);         if (temp == "not found"){       temp = "-1.0";                  }
 142                         m->mothurConvert(temp, misMatch);
 143             if (misMatch > 0) { m->mothurOut("[ERROR]: mismatch must be negative.\n"); abort=true; }
 144
 145                         temp = validParameter.validFile(parameters, "gapopen", false);          if (temp == "not found"){       temp = "-2.0";                  }
 146                         m->mothurConvert(temp, gapOpen);
 147             if (gapOpen > 0) { m->mothurOut("[ERROR]: gapopen must be negative.\n"); abort=true; }
 148
 149                         temp = validParameter.validFile(parameters, "gapextend", false);        if (temp == "not found"){       temp = "-1.0";                  }
 150                         m->mothurConvert(temp, gapExtend);
 151             if (gapExtend > 0) { m->mothurOut("[ERROR]: gapextend must be negative.\n"); abort=true; }
 152
 153                         temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
 154                         m->setProcessors(temp);
 155                         m->mothurConvert(temp, processors);
 156
 157                         align = validParameter.validFile(parameters, "align", false);           if (align == "not found"){      align = "needleman";    }
 158                         if ((align != "needleman") && (align != "blast") && (align != "gotoh") && (align != "noalign")) { m->mothurOut(align + " is not a valid alignment method. Options are needleman, blast, gotoh and noalign. I will use needleman."); m->mothurOutEndLine(); align = "needleman"; }
 159         }
 160
 161         }
 162         catch(exception& e) {
 163                 m->errorOut(e, "MakeContigsCommand", "MakeContigsCommand");
 164                 exit(1);
 165         }
 166 }
 167 //**********************************************************************************************************************
 168 int MakeContigsCommand::execute(){
 169         try {
 170                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 171
 172         //read ffastq and rfastq files creating fasta and qual files.
 173         //this function will create a forward and reverse, fasta and qual files for each processor.
 174         //files has an entry for each processor. files[i][0] = forwardFasta, files[i][1] = forwardQual, files[i][2] = reverseFasta, files[i][3] = reverseQual
 175         int numReads = 0;
 176         m->mothurOut("Reading fastq data..."); cout.flush();
 177         vector< vector<string> > files = readFastqFiles(numReads);
 178         m->mothurOut("Done.\n");
 179
 180
 181
 182
 183         string currentFasta = "";
 184                 itTypes = outputTypes.find("fasta");
 185                 if (itTypes != outputTypes.end()) {
 186                         if ((itTypes->second).size() != 0) { currentFasta = (itTypes->second)[0]; m->setFastaFile(currentFasta); }
 187                 }
 188
 189         string currentQual = "";
 190                 itTypes = outputTypes.find("qfile");
 191                 if (itTypes != outputTypes.end()) {
 192                         if ((itTypes->second).size() != 0) { currentQual = (itTypes->second)[0]; m->setQualFile(currentQual); }
 193                 }
 194
 195         //output files created by command
 196                 m->mothurOutEndLine();
 197                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 198                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 199                 m->mothurOutEndLine();
 200
 201
 202         return 0;
 203     }
 204         catch(exception& e) {
 205                 m->errorOut(e, "MakeContigsCommand", "execute");
 206                 exit(1);
 207         }
 208 }
 209 //**********************************************************************************************************************
 210 vector< vector<string> > MakeContigsCommand::readFastqFiles(int& count){
 211     try {
 212         vector< vector<string> > files;
 213
 214         //maps processors number to file pointer
 215         map<int, vector<ofstream*> > tempfiles;  //tempfiles[0] = forwardFasta, [1] = forwardQual, [2] = reverseFasta, [3] = reverseQual
 216         map<int, vector<ofstream*> >::iterator it;
 217
 218         //create files to write to
 219         for (int i = 0; i < processors; i++) {
 220             vector<ofstream*> temp;
 221             ofstream* outFF = new ofstream;     temp.push_back(outFF);
 222             ofstream* outFQ = new ofstream;     temp.push_back(outFQ);
 223             ofstream* outRF = new ofstream;     temp.push_back(outRF);
 224             ofstream* outRQ = new ofstream;     temp.push_back(outRQ);
 225             tempfiles[i] = temp;
 226
 227             vector<string> names;
 228             string ffastafilename = outputDir + m->getRootName(m->getSimpleName(ffastqfile)) + toString(i) + "ffasta.temp";
 229             string rfastafilename = outputDir + m->getRootName(m->getSimpleName(rfastqfile)) + toString(i) + "rfasta.temp";
 230             string fqualfilename = outputDir + m->getRootName(m->getSimpleName(ffastqfile)) + toString(i) + "fqual.temp";
 231             string rqualfilename = outputDir + m->getRootName(m->getSimpleName(rfastqfile)) + toString(i) + "rqual.temp";
 232             names.push_back(ffastafilename); names.push_back(fqualfilename);
 233             names.push_back(rfastafilename); names.push_back(rqualfilename);
 234             files.push_back(names);
 235
 236             m->openOutputFile(ffastafilename, *outFF);
 237             m->openOutputFile(rfastafilename, *outRF);
 238             m->openOutputFile(fqualfilename, *outFQ);
 239             m->openOutputFile(rqualfilename, *outRQ);
 240         }
 241
 242         if (m->control_pressed) {
 243             //close files, delete ofstreams
 244             for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
 245             //remove files
 246             for (int i = 0; i < files.size(); i++) {
 247                 for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); }
 248             }
 249         }
 250
 251         ifstream inForward;
 252         m->openInputFile(ffastqfile, inForward);
 253
 254         ifstream inReverse;
 255         m->openInputFile(rfastqfile, inReverse);
 256
 257         count = 0;
 258         while ((!inForward.eof()) && (!inReverse.eof())) {
 259
 260             if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inForward.close(); inReverse.close(); return files; }
 261
 262             //get a read from forward and reverse fastq files
 263             fastqRead fread = readFastq(inForward);
 264             fastqRead rread = readFastq(inReverse);
 265             checkReads(fread, rread);
 266
 267             if (m->control_pressed) { for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } } for (int i = 0; i < files.size(); i++) {  for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } } inForward.close(); inReverse.close(); return files; }
 268
 269             //if the reads are okay write to output files
 270             int process = count % processors;
 271
 272             *(tempfiles[process][0]) << ">" << fread.name << endl << fread.sequence << endl;
 273             *(tempfiles[process][1]) << ">" << fread.name << endl;
 274             for (int i = 0; i < fread.scores.size(); i++) { *(tempfiles[process][1]) << fread.scores[i] << " "; }
 275             *(tempfiles[process][1]) << endl;
 276             *(tempfiles[process][2]) << ">" << rread.name << endl << rread.sequence << endl;
 277             *(tempfiles[process][3]) << ">" << rread.name << endl;
 278             for (int i = 0; i < rread.scores.size(); i++) { *(tempfiles[process][3]) << rread.scores[i] << " "; }
 279             *(tempfiles[process][3]) << endl;
 280
 281             count++;
 282
 283             //report progress
 284                         if((count) % 10000 == 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 285
 286                 }
 287                 //report progress
 288                 if((count) % 10000 != 0){       m->mothurOut(toString(count)); m->mothurOutEndLine();           }
 289
 290
 291
 292         //close files, delete ofstreams
 293         for (it = tempfiles.begin(); it!=tempfiles.end(); it++) { for (int i = 0; i < (it->second).size(); i++) { (*(it->second)[i]).close();  delete (it->second)[i]; } }
 294         inForward.close();
 295         inReverse.close();
 296
 297         //adjust for really large processors or really small files
 298         if (count < processors) {
 299             for (int i = count; i < processors; i++) { for(int j = 0; j < files[i].size(); j++) { m->mothurRemove(files[i][j]); } files[i].clear(); }
 300             files.resize(count);
 301             processors = count;
 302         }
 303
 304         return files;
 305     }
 306     catch(exception& e) {
 307         m->errorOut(e, "MakeContigsCommand", "readFastqFiles");
 308         exit(1);
 309     }
 310 }
 311
 312 //**********************************************************************************************************************
 313 fastqRead MakeContigsCommand::readFastq(ifstream& in){
 314     try {
 315         fastqRead read;
 316
 317         //read sequence name
 318         string name = m->getline(in); m->gobble(in);
 319         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 320         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 321         else { name = name.substr(1); }
 322
 323         //read sequence
 324         string sequence = m->getline(in); m->gobble(in);
 325         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 326
 327         //read sequence name
 328         string name2 = m->getline(in); m->gobble(in);
 329         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 330         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 331         else { name2 = name2.substr(1);  }
 332
 333         //read quality scores
 334         string quality = m->getline(in); m->gobble(in);
 335         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 336
 337         //sanity check sequence length and number of quality scores match
 338         if (name2 != "") { if (name != name2) { m->mothurOut("[ERROR]: names do not match. read " + name + " for fasta and " + name2 + " for quality."); m->mothurOutEndLine(); m->control_pressed = true; return read; } }
 339         if (quality.length() != sequence.length()) { m->mothurOut("[ERROR]: Lengths do not match for sequence " + name + ". Read " + toString(sequence.length()) + " characters for fasta and " + toString(quality.length()) + " characters for quality scores."); m->mothurOutEndLine(); m->control_pressed = true; return read; }
 340
 341         vector<int> qualScores;
 342                 int controlChar = int('@');
 343                 for (int i = 0; i < quality.length(); i++) {
 344                         int temp = int(quality[i]);
 345                         temp -= controlChar;
 346
 347                         qualScores.push_back(temp);
 348                 }
 349
 350         read.name = name;
 351         read.sequence = sequence;
 352         read.scores = qualScores;
 353
 354         return read;
 355     }
 356     catch(exception& e) {
 357         m->errorOut(e, "MakeContigsCommand", "readFastq");
 358         exit(1);
 359     }
 360 }
 361 //**********************************************************************************************************************
 362 bool MakeContigsCommand::checkReads(fastqRead& forward, fastqRead& reverse){
 363     try {
 364         bool good = true;
 365
 366         //fix names
 367         if ((forward.name.length() > 2) && (reverse.name.length() > 2)) {
 368             forward.name = forward.name.substr(0, forward.name.length()-2);
 369             reverse.name = reverse.name.substr(0, reverse.name.length()-2);
 370         }else { good = false; m->control_pressed = true; }
 371
 372         //do names match
 373         if (forward.name != reverse.name) {
 374             m->mothurOut("[ERROR]: read " + forward.name + " from " + ffastqfile + ", but read " + reverse.name + " from " + rfastqfile + ".\n");
 375             good = false; m->control_pressed = true;
 376         }
 377
 378         //do sequence lengths match
 379         if (forward.sequence.length() != reverse.sequence.length()) {
 380             m->mothurOut("[ERROR]: For sequence " + forward.name + " I read a sequence of length " + toString(forward.sequence.length()) + " from " + ffastqfile + ", but read a sequence of length " + toString(reverse.sequence.length()) + " from " + rfastqfile + ".\n");
 381             good = false; m->control_pressed = true;
 382         }
 383
 384         //do number of qual scores match
 385         if (forward.scores.size() != reverse.scores.size()) {
 386             m->mothurOut("[ERROR]: For sequence " + forward.name + " I read " + toString(forward.scores.size()) + " quality scores from " + ffastqfile + ", but read  " + toString(reverse.scores.size()) + " quality scores from " + rfastqfile + ".\n");
 387             good = false; m->control_pressed = true;
 388         }
 389
 390         return good;
 391     }
 392     catch(exception& e) {
 393         m->errorOut(e, "MakeContigsCommand", "readFastq");
 394         exit(1);
 395     }
 396 }
 397 //**********************************************************************************************************************
 398
 399
 400
 401