sffinfocommand.cpp

   1 /*
   2  *  sffinfocommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 7/7/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sffinfocommand.h"
  11 #include "endiannessmacros.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SffInfoCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"sff","qfile","fasta","flow","trim","accnos","sfftxt","outputdir","inputdir", "outputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "SffInfoCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 SffInfoCommand::SffInfoCommand(){
  27         try {
  28                 abort = true; calledHelp = true;
  29                 vector<string> tempOutNames;
  30                 outputTypes["fasta"] = tempOutNames;
  31                 outputTypes["flow"] = tempOutNames;
  32                 outputTypes["sfftxt"] = tempOutNames;
  33                 outputTypes["qfile"] = tempOutNames;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 vector<string> SffInfoCommand::getRequiredParameters(){
  42         try {
  43                 string Array[] =  {"sff", "sfftxt", "or"};
  44                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "SffInfoCommand", "getRequiredParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 vector<string> SffInfoCommand::getRequiredFiles(){
  54         try {
  55                 vector<string> myArray;
  56                 return myArray;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "SffInfoCommand", "getRequiredFiles");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64
  65 SffInfoCommand::SffInfoCommand(string option)  {
  66         try {
  67                 abort = false; calledHelp = false;
  68                 hasAccnos = false;
  69
  70                 //allow user to run help
  71                 if(option == "help") { help(); abort = true; calledHelp = true; }
  72
  73                 else {
  74                         //valid paramters for this command
  75                         string Array[] =  {"sff","qfile","fasta","flow","trim","accnos","sfftxt","outputdir","inputdir", "outputdir"};
  76                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  77
  78                         OptionParser parser(option);
  79                         map<string, string> parameters = parser.getParameters();
  80
  81                         ValidParameters validParameter;
  82                         //check to make sure all parameters are valid for command
  83                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  84                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  85                         }
  86
  87                         //initialize outputTypes
  88                         vector<string> tempOutNames;
  89                         outputTypes["fasta"] = tempOutNames;
  90                         outputTypes["flow"] = tempOutNames;
  91                         outputTypes["sfftxt"] = tempOutNames;
  92                         outputTypes["qfile"] = tempOutNames;
  93
  94                         //if the user changes the output directory command factory will send this info to us in the output parameter
  95                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  96
  97                         //if the user changes the input directory command factory will send this info to us in the output parameter
  98                         string inputDir = validParameter.validFile(parameters, "inputdir", false);        if (inputDir == "not found"){ inputDir = "";          }
  99
 100                         sffFilename = validParameter.validFile(parameters, "sff", false);
 101                         if (sffFilename == "not found") { sffFilename = "";  }
 102                         else {
 103                                 m->splitAtDash(sffFilename, filenames);
 104
 105                                 //go through files and make sure they are good, if not, then disregard them
 106                                 for (int i = 0; i < filenames.size(); i++) {
 107                                         if (inputDir != "") {
 108                                                 string path = m->hasPath(filenames[i]);
 109                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 110                                                 if (path == "") {       filenames[i] = inputDir + filenames[i];         }
 111                                         }
 112
 113                                         ifstream in;
 114                                         int ableToOpen = m->openInputFile(filenames[i], in, "noerror");
 115
 116                                         //if you can't open it, try default location
 117                                         if (ableToOpen == 1) {
 118                                                 if (m->getDefaultPath() != "") { //default path is set
 119                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(filenames[i]);
 120                                                         m->mothurOut("Unable to open " + filenames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 121                                                         ifstream in2;
 122                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 123                                                         in2.close();
 124                                                         filenames[i] = tryPath;
 125                                                 }
 126                                         }
 127
 128                                         //if you can't open it, try default location
 129                                         if (ableToOpen == 1) {
 130                                                 if (m->getOutputDir() != "") { //default path is set
 131                                                         string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 132                                                         m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 133                                                         ifstream in2;
 134                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 135                                                         in2.close();
 136                                                         filenames[i] = tryPath;
 137                                                 }
 138                                         }
 139
 140                                         in.close();
 141
 142                                         if (ableToOpen == 1) {
 143                                                 m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 144                                                 //erase from file list
 145                                                 filenames.erase(filenames.begin()+i);
 146                                                 i--;
 147                                         }
 148                                 }
 149
 150                                 //make sure there is at least one valid file left
 151                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 152                         }
 153
 154                         accnosName = validParameter.validFile(parameters, "accnos", false);
 155                         if (accnosName == "not found") { accnosName = "";  }
 156                         else {
 157                                 hasAccnos = true;
 158                                 m->splitAtDash(accnosName, accnosFileNames);
 159
 160                                 //go through files and make sure they are good, if not, then disregard them
 161                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 162                                         if (inputDir != "") {
 163                                                 string path = m->hasPath(accnosFileNames[i]);
 164                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 165                                                 if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 166                                         }
 167
 168                                         ifstream in;
 169                                         int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 170
 171                                         //if you can't open it, try default location
 172                                         if (ableToOpen == 1) {
 173                                                 if (m->getDefaultPath() != "") { //default path is set
 174                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 175                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 176                                                         ifstream in2;
 177                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                         in2.close();
 179                                                         accnosFileNames[i] = tryPath;
 180                                                 }
 181                                         }
 182                                         //if you can't open it, try default location
 183                                         if (ableToOpen == 1) {
 184                                                 if (m->getOutputDir() != "") { //default path is set
 185                                                         string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 186                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 187                                                         ifstream in2;
 188                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 189                                                         in2.close();
 190                                                         accnosFileNames[i] = tryPath;
 191                                                 }
 192                                         }
 193                                         in.close();
 194
 195                                         if (ableToOpen == 1) {
 196                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 197                                                 //erase from file list
 198                                                 accnosFileNames.erase(accnosFileNames.begin()+i);
 199                                                 i--;
 200                                         }
 201                                 }
 202
 203                                 //make sure there is at least one valid file left
 204                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 205                         }
 206
 207                         if (hasAccnos) {
 208                                 if (accnosFileNames.size() != filenames.size()) { abort = true; m->mothurOut("If you provide a accnos file, you must have one for each sff file."); m->mothurOutEndLine(); }
 209                         }
 210
 211                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 212                         qual = m->isTrue(temp);
 213
 214                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 215                         fasta = m->isTrue(temp);
 216
 217                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "F";                             }
 218                         flow = m->isTrue(temp);
 219
 220                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 221                         trim = m->isTrue(temp);
 222
 223                         temp = validParameter.validFile(parameters, "sfftxt", false);
 224                         if (temp == "not found")        {       temp = "F";      sfftxt = false; sfftxtFilename = "";           }
 225                         else if (m->isTrue(temp))       {       sfftxt = true;          sfftxtFilename = "";                            }
 226                         else {
 227                                 //you are a filename
 228                                 if (inputDir != "") {
 229                                         map<string,string>::iterator it = parameters.find("sfftxt");
 230                                         //user has given a template file
 231                                         if(it != parameters.end()){
 232                                                 string path = m->hasPath(it->second);
 233                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 234                                                 if (path == "") {       parameters["sfftxt"] = inputDir + it->second;           }
 235                                         }
 236                                 }
 237
 238                                 sfftxtFilename = validParameter.validFile(parameters, "sfftxt", true);
 239                                 if (sfftxtFilename == "not found") { sfftxtFilename = "";  }
 240                                 else if (sfftxtFilename == "not open") { sfftxtFilename = "";  }
 241                         }
 242
 243                         if ((sfftxtFilename == "") && (filenames.size() == 0)) {  m->mothurOut("[ERROR]: you must provide a valid sff or sfftxt file."); m->mothurOutEndLine(); abort=true; }
 244                 }
 245         }
 246         catch(exception& e) {
 247                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 248                 exit(1);
 249         }
 250 }
 251 //**********************************************************************************************************************
 252
 253 void SffInfoCommand::help(){
 254         try {
 255                 m->mothurOut("The sffinfo command reads a sff file and extracts the sequence data, or you can use it to parse a sfftxt file..\n");
 256                 m->mothurOut("The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, and trim. sff is required. \n");
 257                 m->mothurOut("The sff parameter allows you to enter the sff file you would like to extract data from.  You may enter multiple files by separating them by -'s.\n");
 258                 m->mothurOut("The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n");
 259                 m->mothurOut("The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n");
 260                 m->mothurOut("The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=False. \n");
 261                 m->mothurOut("The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n");
 262                 m->mothurOut("If you want to parse an existing sfftxt file into flow, fasta and quality file, enter the file name using the sfftxt parameter. \n");
 263                 m->mothurOut("The trim parameter allows you to indicate if you would like a sequences and quality scores trimmed to the clipQualLeft and clipQualRight values.  Default=True. \n");
 264                 m->mothurOut("The accnos parameter allows you to provide a accnos file containing the names of the sequences you would like extracted. You may enter multiple files by separating them by -'s. \n");
 265                 m->mothurOut("Example sffinfo(sff=mySffFile.sff, trim=F).\n");
 266                 m->mothurOut("Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n\n");
 267         }
 268         catch(exception& e) {
 269                 m->errorOut(e, "SffInfoCommand", "help");
 270                 exit(1);
 271         }
 272 }
 273 //**********************************************************************************************************************
 274
 275 SffInfoCommand::~SffInfoCommand(){}
 276
 277 //**********************************************************************************************************************
 278 int SffInfoCommand::execute(){
 279         try {
 280
 281                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 282
 283                 for (int s = 0; s < filenames.size(); s++) {
 284
 285                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 286
 287                         int start = time(NULL);
 288
 289                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 290
 291                         string accnos = "";
 292                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 293
 294                         int numReads = extractSffInfo(filenames[s], accnos);
 295
 296                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 297                 }
 298
 299                 if (sfftxtFilename != "") {  parseSffTxt(); }
 300
 301                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 302
 303                 //set fasta file as new current fastafile
 304                 string current = "";
 305                 itTypes = outputTypes.find("fasta");
 306                 if (itTypes != outputTypes.end()) {
 307                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 308                 }
 309
 310                 itTypes = outputTypes.find("qfile");
 311                 if (itTypes != outputTypes.end()) {
 312                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 313                 }
 314
 315                 //report output filenames
 316                 m->mothurOutEndLine();
 317                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 318                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 319                 m->mothurOutEndLine();
 320
 321                 return 0;
 322         }
 323         catch(exception& e) {
 324                 m->errorOut(e, "SffInfoCommand", "execute");
 325                 exit(1);
 326         }
 327 }
 328 //**********************************************************************************************************************
 329 int SffInfoCommand::extractSffInfo(string input, string accnos){
 330         try {
 331
 332                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 333
 334                 if (accnos != "")       {  readAccnosFile(accnos);  }
 335                 else                            {       seqNames.clear();               }
 336
 337                 ofstream outSfftxt, outFasta, outQual, outFlow;
 338                 string outFastaFileName, outQualFileName;
 339                 string sfftxtFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "sff.txt";
 340                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "flow";
 341                 if (trim) {
 342                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "fasta";
 343                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "qual";
 344                 }else{
 345                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.fasta";
 346                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.qual";
 347                 }
 348
 349                 if (sfftxt) { m->openOutputFile(sfftxtFileName, outSfftxt); outSfftxt.setf(ios::fixed, ios::floatfield); outSfftxt.setf(ios::showpoint);  outputNames.push_back(sfftxtFileName);  outputTypes["sfftxt"].push_back(sfftxtFileName); }
 350                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 351                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
 352                 if (flow)       { m->openOutputFile(outFlowFileName, outFlow);          outputNames.push_back(outFlowFileName);  outFlow.setf(ios::fixed, ios::floatfield); outFlow.setf(ios::showpoint); outputTypes["flow"].push_back(outFlowFileName);  }
 353
 354                 ifstream in;
 355                 in.open(input.c_str(), ios::binary);
 356
 357                 CommonHeader header;
 358                 readCommonHeader(in, header);
 359
 360                 int count = 0;
 361
 362                 //check magic number and version
 363                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 364                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 365
 366                 //print common header
 367                 if (sfftxt) {   printCommonHeader(outSfftxt, header);           }
 368                 if (flow)       {       outFlow << header.numFlowsPerRead << endl;      }
 369
 370                 //read through the sff file
 371                 while (!in.eof()) {
 372
 373                         bool print = true;
 374
 375                         //read header
 376                         Header readheader;
 377                         readHeader(in, readheader);
 378
 379                         //read data
 380                         seqRead read;
 381                         readSeqData(in, read, header.numFlowsPerRead, readheader.numBases);
 382
 383                         //if you have provided an accosfile and this seq is not in it, then dont print
 384                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 385
 386                         //print
 387                         if (print) {
 388                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 389                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 390                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 391                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 392                         }
 393
 394                         count++;
 395
 396                         //report progress
 397                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 398
 399                         if (m->control_pressed) { count = 0; break;   }
 400
 401                         if (count >= header.numReads) { break; }
 402                 }
 403
 404                 //report progress
 405                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 406
 407                 in.close();
 408
 409                 if (sfftxt) {  outSfftxt.close();       }
 410                 if (fasta)      {  outFasta.close();    }
 411                 if (qual)       {  outQual.close();             }
 412                 if (flow)       {  outFlow.close();             }
 413
 414                 return count;
 415         }
 416         catch(exception& e) {
 417                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 418                 exit(1);
 419         }
 420 }
 421 //**********************************************************************************************************************
 422 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 423         try {
 424
 425                 if (!in.eof()) {
 426
 427                         //read magic number
 428                         char buffer[4];
 429                         in.read(buffer, 4);
 430                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 431
 432                         //read version
 433                         char buffer9[4];
 434                         in.read(buffer9, 4);
 435                         header.version = "";
 436                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i])); }
 437
 438                         //read offset
 439                         char buffer2 [8];
 440                         in.read(buffer2, 8);
 441                         header.indexOffset =  be_int8(*(unsigned long int *)(&buffer2));
 442
 443                         //read index length
 444                         char buffer3 [4];
 445                         in.read(buffer3, 4);
 446                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 447
 448                         //read num reads
 449                         char buffer4 [4];
 450                         in.read(buffer4, 4);
 451                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 452
 453                         //read header length
 454                         char buffer5 [2];
 455                         in.read(buffer5, 2);
 456                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 457
 458                         //read key length
 459                         char buffer6 [2];
 460                         in.read(buffer6, 2);
 461                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 462
 463                         //read number of flow reads
 464                         char buffer7 [2];
 465                         in.read(buffer7, 2);
 466                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 467
 468                         //read format code
 469                         char buffer8 [1];
 470                         in.read(buffer8, 1);
 471                         header.flogramFormatCode = (int)(buffer8[0]);
 472
 473                         //read flow chars
 474                         char* tempBuffer = new char[header.numFlowsPerRead];
 475                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 476                         header.flowChars = tempBuffer;
 477                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 478                         delete[] tempBuffer;
 479
 480                         //read key
 481                         char* tempBuffer2 = new char[header.keyLength];
 482                         in.read(&(*tempBuffer2), header.keyLength);
 483                         header.keySequence = tempBuffer2;
 484                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 485                         delete[] tempBuffer2;
 486
 487                         /* Pad to 8 chars */
 488                         unsigned long int spotInFile = in.tellg();
 489                         unsigned long int spot = (spotInFile + 7)& ~7;  // ~ inverts
 490                         in.seekg(spot);
 491
 492                 }else{
 493                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 494                 }
 495
 496                 return 0;
 497         }
 498         catch(exception& e) {
 499                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 500                 exit(1);
 501         }
 502 }
 503 //**********************************************************************************************************************
 504 int SffInfoCommand::readHeader(ifstream& in, Header& header){
 505         try {
 506
 507                 if (!in.eof()) {
 508
 509                         //read header length
 510                         char buffer [2];
 511                         in.read(buffer, 2);
 512                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 513
 514                         //read name length
 515                         char buffer2 [2];
 516                         in.read(buffer2, 2);
 517                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 518
 519                         //read num bases
 520                         char buffer3 [4];
 521                         in.read(buffer3, 4);
 522                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 523
 524                         //read clip qual left
 525                         char buffer4 [2];
 526                         in.read(buffer4, 2);
 527                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 528                         header.clipQualLeft = 5;
 529
 530                         //read clip qual right
 531                         char buffer5 [2];
 532                         in.read(buffer5, 2);
 533                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 534
 535                         //read clipAdapterLeft
 536                         char buffer6 [2];
 537                         in.read(buffer6, 2);
 538                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 539
 540                         //read clipAdapterRight
 541                         char buffer7 [2];
 542                         in.read(buffer7, 2);
 543                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 544
 545                         //read name
 546                         char* tempBuffer = new char[header.nameLength];
 547                         in.read(&(*tempBuffer), header.nameLength);
 548                         header.name = tempBuffer;
 549                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 550                         delete[] tempBuffer;
 551
 552                         //extract info from name
 553                         decodeName(header.timestamp, header.region, header.xy, header.name);
 554
 555                         /* Pad to 8 chars */
 556                         unsigned long int spotInFile = in.tellg();
 557                         unsigned long int spot = (spotInFile + 7)& ~7;
 558                         in.seekg(spot);
 559
 560                 }else{
 561                         m->mothurOut("Error reading sff header info."); m->mothurOutEndLine();
 562                 }
 563
 564                 return 0;
 565         }
 566         catch(exception& e) {
 567                 m->errorOut(e, "SffInfoCommand", "readHeader");
 568                 exit(1);
 569         }
 570 }
 571 //**********************************************************************************************************************
 572 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, int numBases){
 573         try {
 574
 575                 if (!in.eof()) {
 576
 577                         //read flowgram
 578                         read.flowgram.resize(numFlowReads);
 579                         for (int i = 0; i < numFlowReads; i++) {
 580                                 char buffer [2];
 581                                 in.read(buffer, 2);
 582                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 583                         }
 584
 585                         //read flowIndex
 586                         read.flowIndex.resize(numBases);
 587                         for (int i = 0; i < numBases; i++) {
 588                                 char temp[1];
 589                                 in.read(temp, 1);
 590                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 591                         }
 592
 593                         //read bases
 594                         char* tempBuffer = new char[numBases];
 595                         in.read(&(*tempBuffer), numBases);
 596                         read.bases = tempBuffer;
 597                         if (read.bases.length() > numBases) { read.bases = read.bases.substr(0, numBases);  }
 598                         delete[] tempBuffer;
 599
 600                         //read qual scores
 601                         read.qualScores.resize(numBases);
 602                         for (int i = 0; i < numBases; i++) {
 603                                 char temp[1];
 604                                 in.read(temp, 1);
 605                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 606                         }
 607
 608                         /* Pad to 8 chars */
 609                         unsigned long int spotInFile = in.tellg();
 610                         unsigned long int spot = (spotInFile + 7)& ~7;
 611                         in.seekg(spot);
 612
 613                 }else{
 614                         m->mothurOut("Error reading."); m->mothurOutEndLine();
 615                 }
 616
 617                 return 0;
 618         }
 619         catch(exception& e) {
 620                 m->errorOut(e, "SffInfoCommand", "readSeqData");
 621                 exit(1);
 622         }
 623 }
 624 //**********************************************************************************************************************
 625 int SffInfoCommand::decodeName(string& timestamp, string& region, string& xy, string name) {
 626         try {
 627
 628                 if (name.length() >= 6) {
 629                         string time = name.substr(0, 6);
 630                         unsigned int timeNum = m->fromBase36(time);
 631
 632                         int q1 = timeNum / 60;
 633                         int sec = timeNum - 60 * q1;
 634                         int q2 = q1 / 60;
 635                         int minute = q1 - 60 * q2;
 636                         int q3 = q2 / 24;
 637                         int hr = q2 - 24 * q3;
 638                         int q4 = q3 / 32;
 639                         int day = q3 - 32 * q4;
 640                         int q5 = q4 / 13;
 641                         int mon = q4 - 13 * q5;
 642                         int year = 2000 + q5;
 643
 644                         timestamp = toString(year) + "_" + toString(mon) + "_" + toString(day) + "_" + toString(hr) + "_" + toString(minute) + "_" + toString(sec);
 645                 }
 646
 647                 if (name.length() >= 9) {
 648                         region = name.substr(7, 2);
 649
 650                         string xyNum = name.substr(9);
 651                         unsigned int myXy = m->fromBase36(xyNum);
 652                         int x = myXy >> 12;
 653                         int y = myXy & 4095;
 654
 655                         xy = toString(x) + "_" + toString(y);
 656                 }
 657
 658                 return 0;
 659         }
 660         catch(exception& e) {
 661                 m->errorOut(e, "SffInfoCommand", "decodeName");
 662                 exit(1);
 663         }
 664 }
 665 //**********************************************************************************************************************
 666 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
 667         try {
 668
 669                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
 670                 out << "Version: " << header.version << endl;
 671                 out << "Index Offset: " << header.indexOffset << endl;
 672                 out << "Index Length: " << header.indexLength << endl;
 673                 out << "Number of Reads: " << header.numReads << endl;
 674                 out << "Header Length: " << header.headerLength << endl;
 675                 out << "Key Length: " << header.keyLength << endl;
 676                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
 677                 out << "Format Code: " << header.flogramFormatCode << endl;
 678                 out << "Flow Chars: " << header.flowChars << endl;
 679                 out << "Key Sequence: " << header.keySequence << endl << endl;
 680
 681                 return 0;
 682         }
 683         catch(exception& e) {
 684                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
 685                 exit(1);
 686         }
 687 }
 688 //**********************************************************************************************************************
 689 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
 690         try {
 691
 692                 out << ">" << header.name << endl;
 693                 out << "Run Prefix: " << header.timestamp << endl;
 694                 out << "Region #:  " << header.region << endl;
 695                 out << "XY Location: " << header.xy << endl << endl;
 696
 697                 out << "Run Name:  " << endl;
 698                 out << "Analysis Name:  " << endl;
 699                 out << "Full Path: " << endl << endl;
 700
 701                 out << "Read Header Len: " << header.headerLength << endl;
 702                 out << "Name Length: " << header.nameLength << endl;
 703                 out << "# of Bases: " << header.numBases << endl;
 704                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
 705                 out << "Clip Qual Right: " << header.clipQualRight << endl;
 706                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
 707                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
 708
 709                 return 0;
 710         }
 711         catch(exception& e) {
 712                 m->errorOut(e, "SffInfoCommand", "printHeader");
 713                 exit(1);
 714         }
 715 }
 716
 717 //**********************************************************************************************************************
 718 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
 719         try {
 720
 721                 out << "Flowgram: ";
 722                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
 723
 724                 out << endl <<  "Flow Indexes: ";
 725                 int sum = 0;
 726                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
 727
 728                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 729                 if(header.clipQualRight == 0){  header.clipQualRight = read.bases.length();     }
 730                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
 731                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
 732                 for (int i = (header.clipQualRight-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
 733
 734                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
 735                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 736
 737
 738                 out << endl << endl;
 739
 740                 return 0;
 741         }
 742         catch(exception& e) {
 743                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
 744                 exit(1);
 745         }
 746 }
 747 //**********************************************************************************************************************
 748 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
 749         try {
 750
 751                 string seq = read.bases;
 752
 753                 if (trim) {
 754                         if(header.clipQualRight < header.clipQualLeft){
 755                                 seq = "NNNN";
 756                         }
 757                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 758                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
 759                         }
 760                         else {
 761                                 seq = seq.substr(header.clipQualLeft-1);
 762                         }
 763                 }else{
 764                         //if you wanted the sfftxt then you already converted the bases to the right case
 765                         if (!sfftxt) {
 766                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 767                                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
 768                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
 769                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
 770                                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
 771                         }
 772                 }
 773
 774                 out << ">" << header.name  << " xy=" << header.xy << endl;
 775                 out << seq << endl;
 776
 777                 return 0;
 778         }
 779         catch(exception& e) {
 780                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
 781                 exit(1);
 782         }
 783 }
 784
 785 //**********************************************************************************************************************
 786 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
 787         try {
 788
 789                 if (trim) {
 790                         if(header.clipQualRight < header.clipQualLeft){
 791                                 out << "0\t0\t0\t0";
 792                         }
 793                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 794                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 795                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
 796                         }
 797                         else{
 798                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 799                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
 800                         }
 801                 }else{
 802                         out << ">" << header.name << " xy=" << header.xy << " length=" << read.qualScores.size() << endl;
 803                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 804                 }
 805
 806                 out << endl;
 807
 808                 return 0;
 809         }
 810         catch(exception& e) {
 811                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
 812                 exit(1);
 813         }
 814 }
 815
 816 //**********************************************************************************************************************
 817 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
 818         try {
 819                 if(header.clipQualRight > header.clipQualLeft){
 820
 821                         int rightIndex = 0;
 822                         for (int i = 0; i < header.clipQualRight; i++) {  rightIndex +=  read.flowIndex[i];     }
 823
 824                         out << header.name << ' ' << rightIndex;
 825                         for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << ' ' << (read.flowgram[i]/(float)100);  }
 826                         out << endl;
 827                 }
 828
 829
 830                 return 0;
 831         }
 832         catch(exception& e) {
 833                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
 834                 exit(1);
 835         }
 836 }
 837 //**********************************************************************************************************************
 838 int SffInfoCommand::readAccnosFile(string filename) {
 839         try {
 840                 //remove old names
 841                 seqNames.clear();
 842
 843                 ifstream in;
 844                 m->openInputFile(filename, in);
 845                 string name;
 846
 847                 while(!in.eof()){
 848                         in >> name; m->gobble(in);
 849
 850                         seqNames.insert(name);
 851
 852                         if (m->control_pressed) { seqNames.clear(); break; }
 853                 }
 854                 in.close();
 855
 856                 return 0;
 857         }
 858         catch(exception& e) {
 859                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
 860                 exit(1);
 861         }
 862 }
 863 //**********************************************************************************************************************
 864 int SffInfoCommand::parseSffTxt() {
 865         try {
 866
 867                 ifstream inSFF;
 868                 m->openInputFile(sfftxtFilename, inSFF);
 869
 870                 if (outputDir == "") {  outputDir += m->hasPath(sfftxtFilename); }
 871
 872                 //output file names
 873                 ofstream outFasta, outQual, outFlow;
 874                 string outFastaFileName, outQualFileName;
 875                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(sfftxtFilename)) + "flow";
 876                 if (trim) {
 877                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(sfftxtFilename)) + "fasta";
 878                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(sfftxtFilename)) + "qual";
 879                 }else{
 880                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(sfftxtFilename)) + "raw.fasta";
 881                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(sfftxtFilename)) + "raw.qual";
 882                 }
 883
 884                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 885                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qual"].push_back(outQualFileName);  }
 886                 if (flow)       { m->openOutputFile(outFlowFileName, outFlow);          outputNames.push_back(outFlowFileName);  outFlow.setf(ios::fixed, ios::floatfield); outFlow.setf(ios::showpoint); outputTypes["flow"].push_back(outFlowFileName);  }
 887
 888                 //read common header
 889                 string commonHeader = m->getline(inSFF);
 890                 string magicNumber = m->getline(inSFF);
 891                 string version = m->getline(inSFF);
 892                 string indexOffset = m->getline(inSFF);
 893                 string indexLength = m->getline(inSFF);
 894                 int numReads = parseHeaderLineToInt(inSFF);
 895                 string headerLength = m->getline(inSFF);
 896                 string keyLength = m->getline(inSFF);
 897                 int numFlows = parseHeaderLineToInt(inSFF);
 898                 string flowgramCode = m->getline(inSFF);
 899                 string flowChars = m->getline(inSFF);
 900                 string keySequence = m->getline(inSFF);
 901                 m->gobble(inSFF);
 902
 903                 string seqName;
 904
 905                 if (flow)       {       outFlow << numFlows << endl;    }
 906
 907                 for(int i=0;i<numReads;i++){
 908
 909                         //sanity check
 910                         if (inSFF.eof()) { m->mothurOut("[ERROR]: Expected " + toString(numReads) + " but reached end of file at " + toString(i+1) + "."); m->mothurOutEndLine(); break; }
 911
 912                         Header header;
 913
 914                         //parse read header
 915                         inSFF >> seqName;
 916                         seqName = seqName.substr(1);
 917                         m->gobble(inSFF);
 918                         header.name = seqName;
 919
 920                         string runPrefix = parseHeaderLineToString(inSFF);              header.timestamp = runPrefix;
 921                         string regionNumber = parseHeaderLineToString(inSFF);   header.region = regionNumber;
 922                         string xyLocation = parseHeaderLineToString(inSFF);             header.xy = xyLocation;
 923                         m->gobble(inSFF);
 924
 925                         string runName = parseHeaderLineToString(inSFF);
 926                         string analysisName = parseHeaderLineToString(inSFF);
 927                         string fullPath = parseHeaderLineToString(inSFF);
 928                         m->gobble(inSFF);
 929
 930                         string readHeaderLen = parseHeaderLineToString(inSFF);  convert(readHeaderLen, header.headerLength);
 931                         string nameLength = parseHeaderLineToString(inSFF);             convert(nameLength, header.nameLength);
 932                         int numBases = parseHeaderLineToInt(inSFF);                             header.numBases = numBases;
 933                         string clipQualLeft = parseHeaderLineToString(inSFF);   convert(clipQualLeft, header.clipQualLeft);
 934                         int clipQualRight = parseHeaderLineToInt(inSFF);                header.clipQualRight = clipQualRight;
 935                         string clipAdapLeft = parseHeaderLineToString(inSFF);   convert(clipAdapLeft, header.clipAdapterLeft);
 936                         string clipAdapRight = parseHeaderLineToString(inSFF);  convert(clipAdapRight, header.clipAdapterRight);
 937                         m->gobble(inSFF);
 938
 939                         seqRead read;
 940
 941                         //parse read
 942                         vector<unsigned short> flowVector = parseHeaderLineToFloatVector(inSFF, numFlows);      read.flowgram = flowVector;
 943                         vector<unsigned int> flowIndices = parseHeaderLineToIntVector(inSFF, numBases);
 944
 945                         //adjust for print
 946                         vector<unsigned int> flowIndicesAdjusted; flowIndicesAdjusted.push_back(flowIndices[0]);
 947                         for (int j = 1; j < flowIndices.size(); j++) {   flowIndicesAdjusted.push_back(flowIndices[j] - flowIndices[j-1]);   }
 948                         read.flowIndex = flowIndicesAdjusted;
 949
 950                         string bases = parseHeaderLineToString(inSFF);                                                                          read.bases = bases;
 951                         vector<unsigned int> qualityScores = parseHeaderLineToIntVector(inSFF, numBases);       read.qualScores = qualityScores;
 952                         m->gobble(inSFF);
 953
 954                         //if you have provided an accosfile and this seq is not in it, then dont print
 955                         bool print = true;
 956                         if (seqNames.size() != 0) {   if (seqNames.count(header.name) == 0) { print = false; }  }
 957
 958                         //print
 959                         if (print) {
 960                                 if (fasta)      {       printFastaSeqData(outFasta, read, header);      }
 961                                 if (qual)       {       printQualSeqData(outQual, read, header);        }
 962                                 if (flow)       {       printFlowSeqData(outFlow, read, header);        }
 963                         }
 964
 965                         //report progress
 966                         if((i+1) % 10000 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine();             }
 967
 968                         if (m->control_pressed) {  break;  }
 969                 }
 970
 971                 //report progress
 972                 if (!m->control_pressed) {   if((numReads) % 10000 != 0){       m->mothurOut(toString(numReads)); m->mothurOutEndLine();                }  }
 973
 974                 inSFF.close();
 975
 976                 if (fasta)      {  outFasta.close();    }
 977                 if (qual)       {  outQual.close();             }
 978                 if (flow)       {  outFlow.close();             }
 979
 980                 return 0;
 981         }
 982         catch(exception& e) {
 983                 m->errorOut(e, "SffInfoCommand", "parseSffTxt");
 984                 exit(1);
 985         }
 986 }
 987 //**********************************************************************************************************************
 988
 989 int SffInfoCommand::parseHeaderLineToInt(ifstream& file){
 990         try {
 991                 int number;
 992
 993                 while (!file.eof())     {
 994
 995                         char c = file.get();
 996                         if (c == ':'){
 997                                 file >> number;
 998                                 break;
 999                         }
1000
1001                 }
1002                 m->gobble(file);
1003                 return number;
1004         }
1005         catch(exception& e) {
1006                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToInt");
1007                 exit(1);
1008         }
1009
1010 }
1011
1012 //**********************************************************************************************************************
1013
1014 string SffInfoCommand::parseHeaderLineToString(ifstream& file){
1015         try {
1016                 string text;
1017
1018                 while (!file.eof())     {
1019                         char c = file.get();
1020
1021                         if (c == ':'){
1022                                 //m->gobble(file);
1023                                 //text = m->getline(file);
1024                                 file >> text;
1025                                 break;
1026                         }
1027                 }
1028                 m->gobble(file);
1029
1030                 return text;
1031         }
1032         catch(exception& e) {
1033                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToString");
1034                 exit(1);
1035         }
1036 }
1037
1038 //**********************************************************************************************************************
1039
1040 vector<unsigned short> SffInfoCommand::parseHeaderLineToFloatVector(ifstream& file, int length){
1041         try {
1042                 vector<unsigned short> floatVector(length);
1043
1044                 while (!file.eof())     {
1045                         char c = file.get();
1046                         if (c == ':'){
1047                                 float temp;
1048                                 for(int i=0;i<length;i++){
1049                                         file >> temp;
1050                                         floatVector[i] = temp * 100;
1051                                 }
1052                                 break;
1053                         }
1054                 }
1055                 m->gobble(file);
1056                 return floatVector;
1057         }
1058         catch(exception& e) {
1059                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToFloatVector");
1060                 exit(1);
1061         }
1062 }
1063
1064 //**********************************************************************************************************************
1065
1066 vector<unsigned int> SffInfoCommand::parseHeaderLineToIntVector(ifstream& file, int length){
1067         try {
1068                 vector<unsigned int> intVector(length);
1069
1070                 while (!file.eof())     {
1071                         char c = file.get();
1072                         if (c == ':'){
1073                                 for(int i=0;i<length;i++){
1074                                         file >> intVector[i];
1075                                 }
1076                                 break;
1077                         }
1078                 }
1079                 m->gobble(file);
1080                 return intVector;
1081         }
1082         catch(exception& e) {
1083                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToIntVector");
1084                 exit(1);
1085         }
1086 }
1087
1088 //**********************************************************************************************************************
1089
1090
1091
1092