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                 //initialize outputTypes
  29                 vector<string> tempOutNames;
  30                 outputTypes["fasta"] = tempOutNames;
  31                 outputTypes["flow"] = tempOutNames;
  32                 outputTypes["sfftxt"] = tempOutNames;
  33                 outputTypes["qual"] = 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"};
  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;
  68                 hasAccnos = false;
  69
  70                 //allow user to run help
  71                 if(option == "help") { help(); abort = 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["qual"] = 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") { m->mothurOut("sff is a required parameter for the sffinfo command."); m->mothurOutEndLine(); abort = true;  }
 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                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 122                                                         filenames[i] = tryPath;
 123                                                 }
 124                                         }
 125
 126                                         //if you can't open it, try default location
 127                                         if (ableToOpen == 1) {
 128                                                 if (m->getOutputDir() != "") { //default path is set
 129                                                         string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 130                                                         m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 131                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 132                                                         filenames[i] = tryPath;
 133                                                 }
 134                                         }
 135
 136                                         in.close();
 137
 138                                         if (ableToOpen == 1) {
 139                                                 m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 140                                                 //erase from file list
 141                                                 filenames.erase(filenames.begin()+i);
 142                                                 i--;
 143                                         }
 144                                 }
 145
 146                                 //make sure there is at least one valid file left
 147                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 148                         }
 149
 150                         accnosName = validParameter.validFile(parameters, "accnos", false);
 151                         if (accnosName == "not found") { accnosName = "";  }
 152                         else {
 153                                 hasAccnos = true;
 154                                 m->splitAtDash(accnosName, accnosFileNames);
 155
 156                                 //go through files and make sure they are good, if not, then disregard them
 157                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 158                                         if (inputDir != "") {
 159                                                 string path = m->hasPath(accnosFileNames[i]);
 160                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 161                                                 if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 162                                         }
 163
 164                                         ifstream in;
 165                                         int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 166
 167                                         //if you can't open it, try default location
 168                                         if (ableToOpen == 1) {
 169                                                 if (m->getDefaultPath() != "") { //default path is set
 170                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 171                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 172                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 173                                                         accnosFileNames[i] = tryPath;
 174                                                 }
 175                                         }
 176                                         //if you can't open it, try default location
 177                                         if (ableToOpen == 1) {
 178                                                 if (m->getOutputDir() != "") { //default path is set
 179                                                         string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 180                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 181                                                         ableToOpen = m->openInputFile(tryPath, in, "noerror");
 182                                                         accnosFileNames[i] = tryPath;
 183                                                 }
 184                                         }
 185                                         in.close();
 186
 187                                         if (ableToOpen == 1) {
 188                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 189                                                 //erase from file list
 190                                                 accnosFileNames.erase(accnosFileNames.begin()+i);
 191                                                 i--;
 192                                         }
 193                                 }
 194
 195                                 //make sure there is at least one valid file left
 196                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 197                         }
 198
 199                         if (hasAccnos) {
 200                                 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(); }
 201                         }
 202
 203                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 204                         qual = m->isTrue(temp);
 205
 206                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 207                         fasta = m->isTrue(temp);
 208
 209                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "F";                             }
 210                         flow = m->isTrue(temp);
 211
 212                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 213                         trim = m->isTrue(temp);
 214
 215                         temp = validParameter.validFile(parameters, "sfftxt", false);                           if (temp == "not found"){       temp = "F";                             }
 216                         sfftxt = m->isTrue(temp);
 217                 }
 218         }
 219         catch(exception& e) {
 220                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 221                 exit(1);
 222         }
 223 }
 224 //**********************************************************************************************************************
 225
 226 void SffInfoCommand::help(){
 227         try {
 228                 m->mothurOut("The sffinfo command reads a sff file and extracts the sequence data.\n");
 229                 m->mothurOut("The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, and trim. sff is required. \n");
 230                 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");
 231                 m->mothurOut("The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n");
 232                 m->mothurOut("The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n");
 233                 m->mothurOut("The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=False. \n");
 234                 m->mothurOut("The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n");
 235                 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");
 236                 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");
 237                 m->mothurOut("Example sffinfo(sff=mySffFile.sff, trim=F).\n");
 238                 m->mothurOut("Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n\n");
 239         }
 240         catch(exception& e) {
 241                 m->errorOut(e, "SffInfoCommand", "help");
 242                 exit(1);
 243         }
 244 }
 245 //**********************************************************************************************************************
 246
 247 SffInfoCommand::~SffInfoCommand(){}
 248
 249 //**********************************************************************************************************************
 250 int SffInfoCommand::execute(){
 251         try {
 252
 253                 if (abort == true) { return 0; }
 254
 255                 for (int s = 0; s < filenames.size(); s++) {
 256
 257                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 258
 259                         int start = time(NULL);
 260
 261                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 262
 263                         string accnos = "";
 264                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 265
 266                         int numReads = extractSffInfo(filenames[s], accnos);
 267
 268                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 269                 }
 270
 271                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 272
 273                 //report output filenames
 274                 m->mothurOutEndLine();
 275                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 276                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 277                 m->mothurOutEndLine();
 278
 279                 return 0;
 280         }
 281         catch(exception& e) {
 282                 m->errorOut(e, "SffInfoCommand", "execute");
 283                 exit(1);
 284         }
 285 }
 286 //**********************************************************************************************************************
 287 int SffInfoCommand::extractSffInfo(string input, string accnos){
 288         try {
 289
 290                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 291
 292                 if (accnos != "")       {  readAccnosFile(accnos);  }
 293                 else                            {       seqNames.clear();               }
 294
 295                 ofstream outSfftxt, outFasta, outQual, outFlow;
 296                 string outFastaFileName, outQualFileName;
 297                 string sfftxtFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "sff.txt";
 298                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "flow";
 299                 if (trim) {
 300                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "fasta";
 301                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "qual";
 302                 }else{
 303                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.fasta";
 304                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.qual";
 305                 }
 306
 307                 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); }
 308                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 309                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qual"].push_back(outQualFileName);  }
 310                 if (flow)       { m->openOutputFile(outFlowFileName, outFlow);          outputNames.push_back(outFlowFileName);  outputTypes["flow"].push_back(outFlowFileName);  }
 311
 312                 ifstream in;
 313                 in.open(input.c_str(), ios::binary);
 314
 315                 CommonHeader header;
 316                 readCommonHeader(in, header);
 317
 318                 int count = 0;
 319
 320                 //check magic number and version
 321                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 322                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 323
 324                 //print common header
 325                 if (sfftxt) { printCommonHeader(outSfftxt, header); }
 326
 327                 //read through the sff file
 328                 while (!in.eof()) {
 329
 330                         bool print = true;
 331
 332                         //read header
 333                         Header readheader;
 334                         readHeader(in, readheader);
 335
 336                         //read data
 337                         seqRead read;
 338                         readSeqData(in, read, header.numFlowsPerRead, readheader.numBases);
 339
 340                         //if you have provided an accosfile and this seq is not in it, then dont print
 341                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 342
 343                         //print
 344                         if (print) {
 345                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 346                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 347                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 348                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 349                         }
 350
 351                         count++;
 352
 353                         //report progress
 354                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 355
 356                         if (m->control_pressed) { count = 0; break;   }
 357
 358                         if (count >= header.numReads) { break; }
 359                 }
 360
 361                 //report progress
 362                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 363
 364                 in.close();
 365
 366                 if (sfftxt) {  outSfftxt.close();       }
 367                 if (fasta)      {  outFasta.close();    }
 368                 if (qual)       {  outQual.close();             }
 369                 if (flow)       {  outFlow.close();             }
 370
 371                 return count;
 372         }
 373         catch(exception& e) {
 374                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 375                 exit(1);
 376         }
 377 }
 378 //**********************************************************************************************************************
 379 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 380         try {
 381
 382                 if (!in.eof()) {
 383
 384                         //read magic number
 385                         char buffer[4];
 386                         in.read(buffer, 4);
 387                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 388
 389                         //read version
 390                         char buffer9[4];
 391                         in.read(buffer9, 4);
 392                         header.version = "";
 393                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i])); }
 394
 395                         //read offset
 396                         char buffer2 [8];
 397                         in.read(buffer2, 8);
 398                         header.indexOffset =  be_int8(*(unsigned long int *)(&buffer2));
 399
 400                         //read index length
 401                         char buffer3 [4];
 402                         in.read(buffer3, 4);
 403                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 404
 405                         //read num reads
 406                         char buffer4 [4];
 407                         in.read(buffer4, 4);
 408                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 409
 410                         //read header length
 411                         char buffer5 [2];
 412                         in.read(buffer5, 2);
 413                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 414
 415                         //read key length
 416                         char buffer6 [2];
 417                         in.read(buffer6, 2);
 418                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 419
 420                         //read number of flow reads
 421                         char buffer7 [2];
 422                         in.read(buffer7, 2);
 423                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 424
 425                         //read format code
 426                         char buffer8 [1];
 427                         in.read(buffer8, 1);
 428                         header.flogramFormatCode = (int)(buffer8[0]);
 429
 430                         //read flow chars
 431                         char* tempBuffer = new char[header.numFlowsPerRead];
 432                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 433                         header.flowChars = tempBuffer;
 434                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 435                         delete[] tempBuffer;
 436
 437                         //read key
 438                         char* tempBuffer2 = new char[header.keyLength];
 439                         in.read(&(*tempBuffer2), header.keyLength);
 440                         header.keySequence = tempBuffer2;
 441                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 442                         delete[] tempBuffer2;
 443
 444                         /* Pad to 8 chars */
 445                         unsigned long int spotInFile = in.tellg();
 446                         unsigned long int spot = (spotInFile + 7)& ~7;  // ~ inverts
 447                         in.seekg(spot);
 448
 449                 }else{
 450                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 451                 }
 452
 453                 return 0;
 454         }
 455         catch(exception& e) {
 456                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 457                 exit(1);
 458         }
 459 }
 460 //**********************************************************************************************************************
 461 int SffInfoCommand::readHeader(ifstream& in, Header& header){
 462         try {
 463
 464                 if (!in.eof()) {
 465
 466                         //read header length
 467                         char buffer [2];
 468                         in.read(buffer, 2);
 469                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 470
 471                         //read name length
 472                         char buffer2 [2];
 473                         in.read(buffer2, 2);
 474                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 475
 476                         //read num bases
 477                         char buffer3 [4];
 478                         in.read(buffer3, 4);
 479                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 480
 481                         //read clip qual left
 482                         char buffer4 [2];
 483                         in.read(buffer4, 2);
 484                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 485                         header.clipQualLeft = 5;
 486
 487                         //read clip qual right
 488                         char buffer5 [2];
 489                         in.read(buffer5, 2);
 490                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 491
 492                         //read clipAdapterLeft
 493                         char buffer6 [2];
 494                         in.read(buffer6, 2);
 495                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 496
 497                         //read clipAdapterRight
 498                         char buffer7 [2];
 499                         in.read(buffer7, 2);
 500                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 501
 502                         //read name
 503                         char* tempBuffer = new char[header.nameLength];
 504                         in.read(&(*tempBuffer), header.nameLength);
 505                         header.name = tempBuffer;
 506                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 507                         delete[] tempBuffer;
 508
 509                         /* Pad to 8 chars */
 510                         unsigned long int spotInFile = in.tellg();
 511                         unsigned long int spot = (spotInFile + 7)& ~7;
 512                         in.seekg(spot);
 513
 514                 }else{
 515                         m->mothurOut("Error reading sff header info."); m->mothurOutEndLine();
 516                 }
 517
 518                 return 0;
 519         }
 520         catch(exception& e) {
 521                 m->errorOut(e, "SffInfoCommand", "readHeader");
 522                 exit(1);
 523         }
 524 }
 525 //**********************************************************************************************************************
 526 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, int numBases){
 527         try {
 528
 529                 if (!in.eof()) {
 530
 531                         //read flowgram
 532                         read.flowgram.resize(numFlowReads);
 533                         for (int i = 0; i < numFlowReads; i++) {
 534                                 char buffer [2];
 535                                 in.read(buffer, 2);
 536                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 537                         }
 538
 539                         //read flowIndex
 540                         read.flowIndex.resize(numBases);
 541                         for (int i = 0; i < numBases; i++) {
 542                                 char temp[1];
 543                                 in.read(temp, 1);
 544                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 545                         }
 546
 547                         //read bases
 548                         char* tempBuffer = new char[numBases];
 549                         in.read(&(*tempBuffer), numBases);
 550                         read.bases = tempBuffer;
 551                         if (read.bases.length() > numBases) { read.bases = read.bases.substr(0, numBases);  }
 552                         delete[] tempBuffer;
 553
 554                         //read qual scores
 555                         read.qualScores.resize(numBases);
 556                         for (int i = 0; i < numBases; i++) {
 557                                 char temp[1];
 558                                 in.read(temp, 1);
 559                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 560                         }
 561
 562                         /* Pad to 8 chars */
 563                         unsigned long int spotInFile = in.tellg();
 564                         unsigned long int spot = (spotInFile + 7)& ~7;
 565                         in.seekg(spot);
 566
 567                 }else{
 568                         m->mothurOut("Error reading."); m->mothurOutEndLine();
 569                 }
 570
 571                 return 0;
 572         }
 573         catch(exception& e) {
 574                 m->errorOut(e, "SffInfoCommand", "readSeqData");
 575                 exit(1);
 576         }
 577 }
 578 //**********************************************************************************************************************
 579 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
 580         try {
 581
 582                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
 583                 out << "Version: " << header.version << endl;
 584                 out << "Index Offset: " << header.indexOffset << endl;
 585                 out << "Index Length: " << header.indexLength << endl;
 586                 out << "Number of Reads: " << header.numReads << endl;
 587                 out << "Header Length: " << header.headerLength << endl;
 588                 out << "Key Length: " << header.keyLength << endl;
 589                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
 590                 out << "Format Code: " << header.flogramFormatCode << endl;
 591                 out << "Flow Chars: " << header.flowChars << endl;
 592                 out << "Key Sequence: " << header.keySequence << endl << endl;
 593
 594                 return 0;
 595         }
 596         catch(exception& e) {
 597                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
 598                 exit(1);
 599         }
 600 }
 601 //**********************************************************************************************************************
 602 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
 603         try {
 604
 605                 out << ">" << header.name << endl;
 606                 out << "Run Prefix: " << endl;
 607                 out << "Region #:  " << endl;
 608                 out << "XY Location: " << endl << endl;
 609
 610                 out << "Run Name:  " << endl;
 611                 out << "Analysis Name:  " << endl;
 612                 out << "Full Path: " << endl << endl;
 613
 614                 out << "Read Header Len: " << header.headerLength << endl;
 615                 out << "Name Length: " << header.nameLength << endl;
 616                 out << "# of Bases: " << header.numBases << endl;
 617                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
 618                 out << "Clip Qual Right: " << header.clipQualRight << endl;
 619                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
 620                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
 621
 622                 return 0;
 623         }
 624         catch(exception& e) {
 625                 m->errorOut(e, "SffInfoCommand", "printHeader");
 626                 exit(1);
 627         }
 628 }
 629
 630 //**********************************************************************************************************************
 631 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
 632         try {
 633
 634                 out << "Flowgram: ";
 635                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
 636
 637                 out << endl <<  "Flow Indexes: ";
 638                 int sum = 0;
 639                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
 640
 641                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 642                 if(header.clipQualRight == 0){  header.clipQualRight = read.bases.length();     }
 643                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
 644                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
 645                 for (int i = (header.clipQualRight-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
 646
 647                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
 648                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 649
 650
 651                 out << endl << endl;
 652
 653                 return 0;
 654         }
 655         catch(exception& e) {
 656                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
 657                 exit(1);
 658         }
 659 }
 660 //**********************************************************************************************************************
 661 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
 662         try {
 663
 664                 string seq = read.bases;
 665
 666                 if (trim) {
 667                         if(header.clipQualRight < header.clipQualLeft){
 668                                 seq = "NNNN";
 669                         }
 670                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 671                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
 672                         }
 673                         else {
 674                                 seq = seq.substr(header.clipQualLeft-1);
 675                         }
 676                 }else{
 677                         //if you wanted the sfftxt then you already converted the bases to the right case
 678                         if (!sfftxt) {
 679                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 680                                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
 681                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
 682                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
 683                                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
 684                         }
 685                 }
 686
 687                 out << ">" << header.name << endl;
 688                 out << seq << endl;
 689
 690                 return 0;
 691         }
 692         catch(exception& e) {
 693                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
 694                 exit(1);
 695         }
 696 }
 697
 698 //**********************************************************************************************************************
 699 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
 700         try {
 701
 702                 if (trim) {
 703                         if(header.clipQualRight < header.clipQualLeft){
 704                                 out << "0\t0\t0\t0";
 705                         }
 706                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 707                                 out << ">" << header.name << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 708                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
 709                         }
 710                         else{
 711                                 out << ">" << header.name << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 712                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
 713                         }
 714                 }else{
 715                         out << ">" << header.name << " length=" << read.qualScores.size() << endl;
 716                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 717                 }
 718
 719                 out << endl;
 720
 721                 return 0;
 722         }
 723         catch(exception& e) {
 724                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
 725                 exit(1);
 726         }
 727 }
 728
 729 //**********************************************************************************************************************
 730 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
 731         try {
 732
 733                 out << ">" << header.name << endl;
 734                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
 735                 out << endl;
 736
 737                 return 0;
 738         }
 739         catch(exception& e) {
 740                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
 741                 exit(1);
 742         }
 743 }
 744 //**********************************************************************************************************************
 745 int SffInfoCommand::readAccnosFile(string filename) {
 746         try {
 747                 //remove old names
 748                 seqNames.clear();
 749
 750                 ifstream in;
 751                 m->openInputFile(filename, in);
 752                 string name;
 753
 754                 while(!in.eof()){
 755                         in >> name; m->gobble(in);
 756
 757                         seqNames.insert(name);
 758
 759                         if (m->control_pressed) { seqNames.clear(); break; }
 760                 }
 761                 in.close();
 762
 763                 return 0;
 764         }
 765         catch(exception& e) {
 766                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
 767                 exit(1);
 768         }
 769 }
 770 //**********************************************************************************************************************/