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