sffinfocommand.cpp

   1 /*
   2  *  sffinfocommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 7/7/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "sffinfocommand.h"
  11 #include "endiannessmacros.h"
  12
  13 //**********************************************************************************************************************
  14 vector<string> SffInfoCommand::getValidParameters(){
  15         try {
  16                 string Array[] =  {"sff","qfile","fasta","flow","trim","accnos","sfftxt","outputdir","inputdir", "outputdir"};
  17                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  18                 return myArray;
  19         }
  20         catch(exception& e) {
  21                 m->errorOut(e, "SffInfoCommand", "getValidParameters");
  22                 exit(1);
  23         }
  24 }
  25 //**********************************************************************************************************************
  26 SffInfoCommand::SffInfoCommand(){
  27         try {
  28                 abort = true;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["flow"] = tempOutNames;
  33                 outputTypes["sfftxt"] = tempOutNames;
  34                 outputTypes["qual"] = tempOutNames;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 vector<string> SffInfoCommand::getRequiredParameters(){
  43         try {
  44                 string Array[] =  {"sff"};
  45                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "SffInfoCommand", "getRequiredParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 vector<string> SffInfoCommand::getRequiredFiles(){
  55         try {
  56                 vector<string> myArray;
  57                 return myArray;
  58         }
  59         catch(exception& e) {
  60                 m->errorOut(e, "SffInfoCommand", "getRequiredFiles");
  61                 exit(1);
  62         }
  63 }
  64 //**********************************************************************************************************************
  65
  66 SffInfoCommand::SffInfoCommand(string option)  {
  67         try {
  68                 abort = false;
  69                 hasAccnos = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string Array[] =  {"sff","qfile","fasta","flow","trim","accnos","sfftxt","outputdir","inputdir", "outputdir"};
  77                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  78
  79                         OptionParser parser(option);
  80                         map<string, string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83                         //check to make sure all parameters are valid for command
  84                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  85                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  86                         }
  87
  88                         //initialize outputTypes
  89                         vector<string> tempOutNames;
  90                         outputTypes["fasta"] = tempOutNames;
  91                         outputTypes["flow"] = tempOutNames;
  92                         outputTypes["sfftxt"] = tempOutNames;
  93                         outputTypes["qual"] = tempOutNames;
  94
  95                         //if the user changes the output directory command factory will send this info to us in the output parameter
  96                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  97
  98                         //if the user changes the input directory command factory will send this info to us in the output parameter
  99                         string inputDir = validParameter.validFile(parameters, "inputdir", false);        if (inputDir == "not found"){ inputDir = "";          }
 100
 101                         sffFilename = validParameter.validFile(parameters, "sff", false);
 102                         if (sffFilename == "not found") { m->mothurOut("sff is a required parameter for the sffinfo command."); m->mothurOutEndLine(); abort = true;  }
 103                         else {
 104                                 m->splitAtDash(sffFilename, filenames);
 105
 106                                 //go through files and make sure they are good, if not, then disregard them
 107                                 for (int i = 0; i < filenames.size(); i++) {
 108                                         if (inputDir != "") {
 109                                                 string path = m->hasPath(filenames[i]);
 110                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 111                                                 if (path == "") {       filenames[i] = inputDir + filenames[i];         }
 112                                         }
 113
 114                                         ifstream in;
 115                                         int ableToOpen = m->openInputFile(filenames[i], in, "noerror");
 116
 117                                         //if you can't open it, try default location
 118                                         if (ableToOpen == 1) {
 119                                                 if (m->getDefaultPath() != "") { //default path is set
 120                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(filenames[i]);
 121                                                         m->mothurOut("Unable to open " + filenames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 122                                                         ifstream in2;
 123                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 124                                                         in2.close();
 125                                                         filenames[i] = tryPath;
 126                                                 }
 127                                         }
 128
 129                                         //if you can't open it, try default location
 130                                         if (ableToOpen == 1) {
 131                                                 if (m->getOutputDir() != "") { //default path is set
 132                                                         string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 133                                                         m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 134                                                         ifstream in2;
 135                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 136                                                         in2.close();
 137                                                         filenames[i] = tryPath;
 138                                                 }
 139                                         }
 140
 141                                         in.close();
 142
 143                                         if (ableToOpen == 1) {
 144                                                 m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 145                                                 //erase from file list
 146                                                 filenames.erase(filenames.begin()+i);
 147                                                 i--;
 148                                         }
 149                                 }
 150
 151                                 //make sure there is at least one valid file left
 152                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 153                         }
 154
 155                         accnosName = validParameter.validFile(parameters, "accnos", false);
 156                         if (accnosName == "not found") { accnosName = "";  }
 157                         else {
 158                                 hasAccnos = true;
 159                                 m->splitAtDash(accnosName, accnosFileNames);
 160
 161                                 //go through files and make sure they are good, if not, then disregard them
 162                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 163                                         if (inputDir != "") {
 164                                                 string path = m->hasPath(accnosFileNames[i]);
 165                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 166                                                 if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 167                                         }
 168
 169                                         ifstream in;
 170                                         int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 171
 172                                         //if you can't open it, try default location
 173                                         if (ableToOpen == 1) {
 174                                                 if (m->getDefaultPath() != "") { //default path is set
 175                                                         string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 176                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 177                                                         ifstream in2;
 178                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 179                                                         in2.close();
 180                                                         accnosFileNames[i] = tryPath;
 181                                                 }
 182                                         }
 183                                         //if you can't open it, try default location
 184                                         if (ableToOpen == 1) {
 185                                                 if (m->getOutputDir() != "") { //default path is set
 186                                                         string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 187                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 188                                                         ifstream in2;
 189                                                         ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 190                                                         in2.close();
 191                                                         accnosFileNames[i] = tryPath;
 192                                                 }
 193                                         }
 194                                         in.close();
 195
 196                                         if (ableToOpen == 1) {
 197                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 198                                                 //erase from file list
 199                                                 accnosFileNames.erase(accnosFileNames.begin()+i);
 200                                                 i--;
 201                                         }
 202                                 }
 203
 204                                 //make sure there is at least one valid file left
 205                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 206                         }
 207
 208                         if (hasAccnos) {
 209                                 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(); }
 210                         }
 211
 212                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 213                         qual = m->isTrue(temp);
 214
 215                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 216                         fasta = m->isTrue(temp);
 217
 218                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "F";                             }
 219                         flow = m->isTrue(temp);
 220
 221                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 222                         trim = m->isTrue(temp);
 223
 224                         temp = validParameter.validFile(parameters, "sfftxt", false);                           if (temp == "not found"){       temp = "F";                             }
 225                         sfftxt = m->isTrue(temp);
 226                 }
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234
 235 void SffInfoCommand::help(){
 236         try {
 237                 m->mothurOut("The sffinfo command reads a sff file and extracts the sequence data.\n");
 238                 m->mothurOut("The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, and trim. sff is required. \n");
 239                 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");
 240                 m->mothurOut("The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n");
 241                 m->mothurOut("The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n");
 242                 m->mothurOut("The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=False. \n");
 243                 m->mothurOut("The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n");
 244                 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");
 245                 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");
 246                 m->mothurOut("Example sffinfo(sff=mySffFile.sff, trim=F).\n");
 247                 m->mothurOut("Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n\n");
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "SffInfoCommand", "help");
 251                 exit(1);
 252         }
 253 }
 254 //**********************************************************************************************************************
 255
 256 SffInfoCommand::~SffInfoCommand(){}
 257
 258 //**********************************************************************************************************************
 259 int SffInfoCommand::execute(){
 260         try {
 261
 262                 if (abort == true) { return 0; }
 263
 264                 for (int s = 0; s < filenames.size(); s++) {
 265
 266                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 267
 268                         int start = time(NULL);
 269
 270                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 271
 272                         string accnos = "";
 273                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 274
 275                         int numReads = extractSffInfo(filenames[s], accnos);
 276
 277                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 278                 }
 279
 280                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       remove(outputNames[i].c_str());         } return 0; }
 281
 282                 //report output filenames
 283                 m->mothurOutEndLine();
 284                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 285                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 286                 m->mothurOutEndLine();
 287
 288                 return 0;
 289         }
 290         catch(exception& e) {
 291                 m->errorOut(e, "SffInfoCommand", "execute");
 292                 exit(1);
 293         }
 294 }
 295 //**********************************************************************************************************************
 296 int SffInfoCommand::extractSffInfo(string input, string accnos){
 297         try {
 298
 299                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 300
 301                 if (accnos != "")       {  readAccnosFile(accnos);  }
 302                 else                            {       seqNames.clear();               }
 303
 304                 ofstream outSfftxt, outFasta, outQual, outFlow;
 305                 string outFastaFileName, outQualFileName;
 306                 string sfftxtFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "sff.txt";
 307                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "flow";
 308                 if (trim) {
 309                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "fasta";
 310                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "qual";
 311                 }else{
 312                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.fasta";
 313                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw.qual";
 314                 }
 315
 316                 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); }
 317                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 318                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qual"].push_back(outQualFileName);  }
 319                 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);  }
 320
 321                 ifstream in;
 322                 in.open(input.c_str(), ios::binary);
 323
 324                 CommonHeader header;
 325                 readCommonHeader(in, header);
 326
 327                 int count = 0;
 328
 329                 //check magic number and version
 330                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 331                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 332
 333                 //print common header
 334                 if (sfftxt) { printCommonHeader(outSfftxt, header); }
 335
 336                 //read through the sff file
 337                 while (!in.eof()) {
 338
 339                         bool print = true;
 340
 341                         //read header
 342                         Header readheader;
 343                         readHeader(in, readheader);
 344
 345                         //read data
 346                         seqRead read;
 347                         readSeqData(in, read, header.numFlowsPerRead, readheader.numBases);
 348
 349                         //if you have provided an accosfile and this seq is not in it, then dont print
 350                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 351
 352                         //print
 353                         if (print) {
 354                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 355                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 356                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 357                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 358                         }
 359
 360                         count++;
 361
 362                         //report progress
 363                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 364
 365                         if (m->control_pressed) { count = 0; break;   }
 366
 367                         if (count >= header.numReads) { break; }
 368                 }
 369
 370                 //report progress
 371                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 372
 373                 in.close();
 374
 375                 if (sfftxt) {  outSfftxt.close();       }
 376                 if (fasta)      {  outFasta.close();    }
 377                 if (qual)       {  outQual.close();             }
 378                 if (flow)       {  outFlow.close();             }
 379
 380                 return count;
 381         }
 382         catch(exception& e) {
 383                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 384                 exit(1);
 385         }
 386 }
 387 //**********************************************************************************************************************
 388 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 389         try {
 390
 391                 if (!in.eof()) {
 392
 393                         //read magic number
 394                         char buffer[4];
 395                         in.read(buffer, 4);
 396                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 397
 398                         //read version
 399                         char buffer9[4];
 400                         in.read(buffer9, 4);
 401                         header.version = "";
 402                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i])); }
 403
 404                         //read offset
 405                         char buffer2 [8];
 406                         in.read(buffer2, 8);
 407                         header.indexOffset =  be_int8(*(unsigned long int *)(&buffer2));
 408
 409                         //read index length
 410                         char buffer3 [4];
 411                         in.read(buffer3, 4);
 412                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 413
 414                         //read num reads
 415                         char buffer4 [4];
 416                         in.read(buffer4, 4);
 417                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 418
 419                         //read header length
 420                         char buffer5 [2];
 421                         in.read(buffer5, 2);
 422                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 423
 424                         //read key length
 425                         char buffer6 [2];
 426                         in.read(buffer6, 2);
 427                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 428
 429                         //read number of flow reads
 430                         char buffer7 [2];
 431                         in.read(buffer7, 2);
 432                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 433
 434                         //read format code
 435                         char buffer8 [1];
 436                         in.read(buffer8, 1);
 437                         header.flogramFormatCode = (int)(buffer8[0]);
 438
 439                         //read flow chars
 440                         char* tempBuffer = new char[header.numFlowsPerRead];
 441                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 442                         header.flowChars = tempBuffer;
 443                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 444                         delete[] tempBuffer;
 445
 446                         //read key
 447                         char* tempBuffer2 = new char[header.keyLength];
 448                         in.read(&(*tempBuffer2), header.keyLength);
 449                         header.keySequence = tempBuffer2;
 450                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 451                         delete[] tempBuffer2;
 452
 453                         /* Pad to 8 chars */
 454                         unsigned long int spotInFile = in.tellg();
 455                         unsigned long int spot = (spotInFile + 7)& ~7;  // ~ inverts
 456                         in.seekg(spot);
 457
 458                 }else{
 459                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 460                 }
 461
 462                 return 0;
 463         }
 464         catch(exception& e) {
 465                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 466                 exit(1);
 467         }
 468 }
 469 //**********************************************************************************************************************
 470 int SffInfoCommand::readHeader(ifstream& in, Header& header){
 471         try {
 472
 473                 if (!in.eof()) {
 474
 475                         //read header length
 476                         char buffer [2];
 477                         in.read(buffer, 2);
 478                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 479
 480                         //read name length
 481                         char buffer2 [2];
 482                         in.read(buffer2, 2);
 483                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 484
 485                         //read num bases
 486                         char buffer3 [4];
 487                         in.read(buffer3, 4);
 488                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 489
 490                         //read clip qual left
 491                         char buffer4 [2];
 492                         in.read(buffer4, 2);
 493                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 494                         header.clipQualLeft = 5;
 495
 496                         //read clip qual right
 497                         char buffer5 [2];
 498                         in.read(buffer5, 2);
 499                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 500
 501                         //read clipAdapterLeft
 502                         char buffer6 [2];
 503                         in.read(buffer6, 2);
 504                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 505
 506                         //read clipAdapterRight
 507                         char buffer7 [2];
 508                         in.read(buffer7, 2);
 509                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 510
 511                         //read name
 512                         char* tempBuffer = new char[header.nameLength];
 513                         in.read(&(*tempBuffer), header.nameLength);
 514                         header.name = tempBuffer;
 515                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 516                         delete[] tempBuffer;
 517
 518                         //extract info from name
 519                         decodeName(header.timestamp, header.region, header.xy, header.name);
 520
 521                         /* Pad to 8 chars */
 522                         unsigned long int spotInFile = in.tellg();
 523                         unsigned long int spot = (spotInFile + 7)& ~7;
 524                         in.seekg(spot);
 525
 526                 }else{
 527                         m->mothurOut("Error reading sff header info."); m->mothurOutEndLine();
 528                 }
 529
 530                 return 0;
 531         }
 532         catch(exception& e) {
 533                 m->errorOut(e, "SffInfoCommand", "readHeader");
 534                 exit(1);
 535         }
 536 }
 537 //**********************************************************************************************************************
 538 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, int numBases){
 539         try {
 540
 541                 if (!in.eof()) {
 542
 543                         //read flowgram
 544                         read.flowgram.resize(numFlowReads);
 545                         for (int i = 0; i < numFlowReads; i++) {
 546                                 char buffer [2];
 547                                 in.read(buffer, 2);
 548                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 549                         }
 550
 551                         //read flowIndex
 552                         read.flowIndex.resize(numBases);
 553                         for (int i = 0; i < numBases; i++) {
 554                                 char temp[1];
 555                                 in.read(temp, 1);
 556                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 557                         }
 558
 559                         //read bases
 560                         char* tempBuffer = new char[numBases];
 561                         in.read(&(*tempBuffer), numBases);
 562                         read.bases = tempBuffer;
 563                         if (read.bases.length() > numBases) { read.bases = read.bases.substr(0, numBases);  }
 564                         delete[] tempBuffer;
 565
 566                         //read qual scores
 567                         read.qualScores.resize(numBases);
 568                         for (int i = 0; i < numBases; i++) {
 569                                 char temp[1];
 570                                 in.read(temp, 1);
 571                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 572                         }
 573
 574                         /* Pad to 8 chars */
 575                         unsigned long int spotInFile = in.tellg();
 576                         unsigned long int spot = (spotInFile + 7)& ~7;
 577                         in.seekg(spot);
 578
 579                 }else{
 580                         m->mothurOut("Error reading."); m->mothurOutEndLine();
 581                 }
 582
 583                 return 0;
 584         }
 585         catch(exception& e) {
 586                 m->errorOut(e, "SffInfoCommand", "readSeqData");
 587                 exit(1);
 588         }
 589 }
 590 //**********************************************************************************************************************
 591 int SffInfoCommand::decodeName(string& timestamp, string& region, string& xy, string name) {
 592         try {
 593
 594                 string time = name.substr(0, 6);
 595                 unsigned int timeNum = m->fromBase36(time);
 596
 597                 int q1 = timeNum / 60;
 598                 int sec = timeNum - 60 * q1;
 599                 int q2 = q1 / 60;
 600                 int minute = q1 - 60 * q2;
 601                 int q3 = q2 / 24;
 602                 int hr = q2 - 24 * q3;
 603                 int q4 = q3 / 32;
 604                 int day = q3 - 32 * q4;
 605                 int q5 = q4 / 13;
 606                 int mon = q4 - 13 * q5;
 607                 int year = 2000 + q5;
 608
 609                 timestamp = toString(year) + "_" + toString(mon) + "_" + toString(day) + "_" + toString(hr) + "_" + toString(minute) + "_" + toString(sec);
 610
 611                 region = name.substr(7, 2);
 612
 613                 string xyNum = name.substr(9);
 614                 unsigned int myXy = m->fromBase36(xyNum);
 615                 int x = myXy >> 12;
 616                 int y = myXy & 4095;
 617
 618                 xy = toString(x) + "_" + toString(y);
 619
 620                 return 0;
 621         }
 622         catch(exception& e) {
 623                 m->errorOut(e, "SffInfoCommand", "decodeName");
 624                 exit(1);
 625         }
 626 }
 627 //**********************************************************************************************************************
 628 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
 629         try {
 630
 631                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
 632                 out << "Version: " << header.version << endl;
 633                 out << "Index Offset: " << header.indexOffset << endl;
 634                 out << "Index Length: " << header.indexLength << endl;
 635                 out << "Number of Reads: " << header.numReads << endl;
 636                 out << "Header Length: " << header.headerLength << endl;
 637                 out << "Key Length: " << header.keyLength << endl;
 638                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
 639                 out << "Format Code: " << header.flogramFormatCode << endl;
 640                 out << "Flow Chars: " << header.flowChars << endl;
 641                 out << "Key Sequence: " << header.keySequence << endl << endl;
 642
 643                 return 0;
 644         }
 645         catch(exception& e) {
 646                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
 647                 exit(1);
 648         }
 649 }
 650 //**********************************************************************************************************************
 651 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
 652         try {
 653
 654                 out << ">" << header.name << endl;
 655                 out << "Run Prefix: " << header.timestamp << endl;
 656                 out << "Region #:  " << header.region << endl;
 657                 out << "XY Location: " << header.xy << endl << endl;
 658
 659                 out << "Run Name:  " << endl;
 660                 out << "Analysis Name:  " << endl;
 661                 out << "Full Path: " << endl << endl;
 662
 663                 out << "Read Header Len: " << header.headerLength << endl;
 664                 out << "Name Length: " << header.nameLength << endl;
 665                 out << "# of Bases: " << header.numBases << endl;
 666                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
 667                 out << "Clip Qual Right: " << header.clipQualRight << endl;
 668                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
 669                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
 670
 671                 return 0;
 672         }
 673         catch(exception& e) {
 674                 m->errorOut(e, "SffInfoCommand", "printHeader");
 675                 exit(1);
 676         }
 677 }
 678
 679 //**********************************************************************************************************************
 680 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
 681         try {
 682
 683                 out << "Flowgram: ";
 684                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
 685
 686                 out << endl <<  "Flow Indexes: ";
 687                 int sum = 0;
 688                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
 689
 690                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 691                 if(header.clipQualRight == 0){  header.clipQualRight = read.bases.length();     }
 692                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
 693                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
 694                 for (int i = (header.clipQualRight-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
 695
 696                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
 697                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 698
 699
 700                 out << endl << endl;
 701
 702                 return 0;
 703         }
 704         catch(exception& e) {
 705                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
 706                 exit(1);
 707         }
 708 }
 709 //**********************************************************************************************************************
 710 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
 711         try {
 712
 713                 string seq = read.bases;
 714
 715                 if (trim) {
 716                         if(header.clipQualRight < header.clipQualLeft){
 717                                 seq = "NNNN";
 718                         }
 719                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 720                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
 721                         }
 722                         else {
 723                                 seq = seq.substr(header.clipQualLeft-1);
 724                         }
 725                 }else{
 726                         //if you wanted the sfftxt then you already converted the bases to the right case
 727                         if (!sfftxt) {
 728                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
 729                                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
 730                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
 731                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
 732                                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
 733                         }
 734                 }
 735
 736                 out << ">" << header.name  << " xy=" << header.xy << endl;
 737                 out << seq << endl;
 738
 739                 return 0;
 740         }
 741         catch(exception& e) {
 742                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
 743                 exit(1);
 744         }
 745 }
 746
 747 //**********************************************************************************************************************
 748 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
 749         try {
 750
 751                 if (trim) {
 752                         if(header.clipQualRight < header.clipQualLeft){
 753                                 out << "0\t0\t0\t0";
 754                         }
 755                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
 756                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 757                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
 758                         }
 759                         else{
 760                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
 761                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
 762                         }
 763                 }else{
 764                         out << ">" << header.name << " xy=" << header.xy << " length=" << read.qualScores.size() << endl;
 765                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
 766                 }
 767
 768                 out << endl;
 769
 770                 return 0;
 771         }
 772         catch(exception& e) {
 773                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
 774                 exit(1);
 775         }
 776 }
 777
 778 //**********************************************************************************************************************
 779 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
 780         try {
 781                 if(header.clipQualRight > header.clipQualLeft){
 782                         out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << " numflows=" << read.flowgram.size() << endl;
 783                         for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << ' ';  }
 784                         out << endl;
 785                 }
 786
 787
 788                 return 0;
 789         }
 790         catch(exception& e) {
 791                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
 792                 exit(1);
 793         }
 794 }
 795 //**********************************************************************************************************************
 796 int SffInfoCommand::readAccnosFile(string filename) {
 797         try {
 798                 //remove old names
 799                 seqNames.clear();
 800
 801                 ifstream in;
 802                 m->openInputFile(filename, in);
 803                 string name;
 804
 805                 while(!in.eof()){
 806                         in >> name; m->gobble(in);
 807
 808                         seqNames.insert(name);
 809
 810                         if (m->control_pressed) { seqNames.clear(); break; }
 811                 }
 812                 in.close();
 813
 814                 return 0;
 815         }
 816         catch(exception& e) {
 817                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
 818                 exit(1);
 819         }
 820 }
 821 //**********************************************************************************************************************/