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 #include "trimoligos.h"
  13 #include "sequence.hpp"
  14 #include "qualityscores.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> SffInfoCommand::setParameters(){
  18         try {
  19                 CommandParameter psff("sff", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(psff);
  20         CommandParameter poligos("oligos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(poligos);
  21                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  22                 CommandParameter psfftxt("sfftxt", "String", "", "", "", "", "",false,false); parameters.push_back(psfftxt);
  23                 CommandParameter pflow("flow", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pflow);
  24                 CommandParameter ptrim("trim", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(ptrim);
  25                 CommandParameter pfasta("fasta", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pfasta);
  26                 CommandParameter pqfile("name", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pqfile);
  27         CommandParameter ppdiffs("pdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(ppdiffs);
  28                 CommandParameter pbdiffs("bdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(pbdiffs);
  29         CommandParameter pldiffs("ldiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(pldiffs);
  30                 CommandParameter psdiffs("sdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(psdiffs);
  31         CommandParameter ptdiffs("tdiffs", "Number", "", "0", "", "", "",false,false); parameters.push_back(ptdiffs);
  32                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  33                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  34
  35                 vector<string> myArray;
  36                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  37                 return myArray;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "SffInfoCommand", "setParameters");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 string SffInfoCommand::getHelpString(){
  46         try {
  47                 string helpString = "";
  48                 helpString += "The sffinfo command reads a sff file and extracts the sequence data, or you can use it to parse a sfftxt file.\n";
  49                 helpString += "The sffinfo command parameters are sff, fasta, qfile, accnos, flow, sfftxt, oligos, bdiffs, tdiffs, ldiffs, sdiffs, pdiffs and trim. sff is required. \n";
  50                 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";
  51                 helpString += "The fasta parameter allows you to indicate if you would like a fasta formatted file generated.  Default=True. \n";
  52                 helpString += "The qfile parameter allows you to indicate if you would like a quality file generated.  Default=True. \n";
  53         helpString += "The oligos parameter allows you to provide an oligos file to split your sff file into separate sff files by barcode. \n";
  54         helpString += "The tdiffs parameter is used to specify the total number of differences allowed in the sequence. The default is pdiffs + bdiffs + sdiffs + ldiffs.\n";
  55                 helpString += "The bdiffs parameter is used to specify the number of differences allowed in the barcode. The default is 0.\n";
  56                 helpString += "The pdiffs parameter is used to specify the number of differences allowed in the primer. The default is 0.\n";
  57         helpString += "The ldiffs parameter is used to specify the number of differences allowed in the linker. The default is 0.\n";
  58                 helpString += "The sdiffs parameter is used to specify the number of differences allowed in the spacer. The default is 0.\n";
  59                 helpString += "The flow parameter allows you to indicate if you would like a flowgram file generated.  Default=True. \n";
  60                 helpString += "The sfftxt parameter allows you to indicate if you would like a sff.txt file generated.  Default=False. \n";
  61                 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";
  62                 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";
  63                 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";
  64                 helpString += "Example sffinfo(sff=mySffFile.sff, trim=F).\n";
  65                 helpString += "Note: No spaces between parameter labels (i.e. sff), '=' and parameters (i.e.yourSffFileName).\n";
  66                 return helpString;
  67         }
  68         catch(exception& e) {
  69                 m->errorOut(e, "SffInfoCommand", "getHelpString");
  70                 exit(1);
  71         }
  72 }
  73 //**********************************************************************************************************************
  74 string SffInfoCommand::getOutputFileNameTag(string type, string inputName=""){
  75         try {
  76         string outputFileName = "";
  77                 map<string, vector<string> >::iterator it;
  78
  79         //is this a type this command creates
  80         it = outputTypes.find(type);
  81         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  82         else {
  83             if (type == "fasta")            {   outputFileName =  "fasta";   }
  84             else if (type == "flow")    {   outputFileName =  "flow";   }
  85             else if (type == "sfftxt")        {   outputFileName =  "sff.txt";   }
  86             else if (type == "sff")        {   outputFileName =  "sff";   }
  87             else if (type == "qfile")       {   outputFileName =  "qual";   }
  88              else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  89         }
  90         return outputFileName;
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "SffInfoCommand", "getOutputFileNameTag");
  94                 exit(1);
  95         }
  96 }
  97
  98
  99 //**********************************************************************************************************************
 100 SffInfoCommand::SffInfoCommand(){
 101         try {
 102                 abort = true; calledHelp = true;
 103                 setParameters();
 104                 vector<string> tempOutNames;
 105                 outputTypes["fasta"] = tempOutNames;
 106                 outputTypes["flow"] = tempOutNames;
 107                 outputTypes["sfftxt"] = tempOutNames;
 108                 outputTypes["qfile"] = tempOutNames;
 109         outputTypes["sff"] = tempOutNames;
 110         }
 111         catch(exception& e) {
 112                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 113                 exit(1);
 114         }
 115 }
 116 //**********************************************************************************************************************
 117
 118 SffInfoCommand::SffInfoCommand(string option)  {
 119         try {
 120                 abort = false; calledHelp = false;
 121                 hasAccnos = false;
 122         split = 1;
 123
 124                 //allow user to run help
 125                 if(option == "help") { help(); abort = true; calledHelp = true; }
 126                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 127
 128                 else {
 129                         //valid paramters for this command
 130                         vector<string> myArray = setParameters();
 131
 132                         OptionParser parser(option);
 133                         map<string, string> parameters = parser.getParameters();
 134
 135                         ValidParameters validParameter;
 136                         //check to make sure all parameters are valid for command
 137                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 138                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 139                         }
 140
 141                         //initialize outputTypes
 142                         vector<string> tempOutNames;
 143                         outputTypes["fasta"] = tempOutNames;
 144                         outputTypes["flow"] = tempOutNames;
 145                         outputTypes["sfftxt"] = tempOutNames;
 146                         outputTypes["qfile"] = tempOutNames;
 147             outputTypes["sff"] = tempOutNames;
 148
 149                         //if the user changes the output directory command factory will send this info to us in the output parameter
 150                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 151
 152                         //if the user changes the input directory command factory will send this info to us in the output parameter
 153                         string inputDir = validParameter.validFile(parameters, "inputdir", false);        if (inputDir == "not found"){ inputDir = "";          }
 154
 155                         sffFilename = validParameter.validFile(parameters, "sff", false);
 156                         if (sffFilename == "not found") { sffFilename = "";  }
 157                         else {
 158                                 m->splitAtDash(sffFilename, filenames);
 159
 160                                 //go through files and make sure they are good, if not, then disregard them
 161                                 for (int i = 0; i < filenames.size(); i++) {
 162                                         bool ignore = false;
 163                                         if (filenames[i] == "current") {
 164                                                 filenames[i] = m->getSFFFile();
 165                                                 if (filenames[i] != "") {  m->mothurOut("Using " + filenames[i] + " as input file for the sff parameter where you had given current."); m->mothurOutEndLine(); }
 166                                                 else {
 167                                                         m->mothurOut("You have no current sfffile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 168                                                         //erase from file list
 169                                                         filenames.erase(filenames.begin()+i);
 170                                                         i--;
 171                                                 }
 172                                         }
 173
 174                                         if (!ignore) {
 175                                                 if (inputDir != "") {
 176                                                         string path = m->hasPath(filenames[i]);
 177                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                                         if (path == "") {       filenames[i] = inputDir + filenames[i];         }
 179                                                 }
 180
 181                                                 ifstream in;
 182                                                 int ableToOpen = m->openInputFile(filenames[i], in, "noerror");
 183
 184                                                 //if you can't open it, try default location
 185                                                 if (ableToOpen == 1) {
 186                                                         if (m->getDefaultPath() != "") { //default path is set
 187                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(filenames[i]);
 188                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 189                                                                 ifstream in2;
 190                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 191                                                                 in2.close();
 192                                                                 filenames[i] = tryPath;
 193                                                         }
 194                                                 }
 195
 196                                                 //if you can't open it, try default location
 197                                                 if (ableToOpen == 1) {
 198                                                         if (m->getOutputDir() != "") { //default path is set
 199                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(filenames[i]);
 200                                                                 m->mothurOut("Unable to open " + filenames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 201                                                                 ifstream in2;
 202                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 203                                                                 in2.close();
 204                                                                 filenames[i] = tryPath;
 205                                                         }
 206                                                 }
 207
 208                                                 in.close();
 209
 210                                                 if (ableToOpen == 1) {
 211                                                         m->mothurOut("Unable to open " + filenames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 212                                                         //erase from file list
 213                                                         filenames.erase(filenames.begin()+i);
 214                                                         i--;
 215                                                 }else { m->setSFFFile(filenames[i]); }
 216                                         }
 217                                 }
 218
 219                                 //make sure there is at least one valid file left
 220                                 if (filenames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 221                         }
 222
 223                         accnosName = validParameter.validFile(parameters, "accnos", false);
 224                         if (accnosName == "not found") { accnosName = "";  }
 225                         else {
 226                                 hasAccnos = true;
 227                                 m->splitAtDash(accnosName, accnosFileNames);
 228
 229                                 //go through files and make sure they are good, if not, then disregard them
 230                                 for (int i = 0; i < accnosFileNames.size(); i++) {
 231                                         bool ignore = false;
 232                                         if (accnosFileNames[i] == "current") {
 233                                                 accnosFileNames[i] = m->getAccnosFile();
 234                                                 if (accnosFileNames[i] != "") {  m->mothurOut("Using " + accnosFileNames[i] + " as input file for the accnos parameter where you had given current."); m->mothurOutEndLine(); }
 235                                                 else {
 236                                                         m->mothurOut("You have no current accnosfile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 237                                                         //erase from file list
 238                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 239                                                         i--;
 240                                                 }
 241                                         }
 242
 243                                         if (!ignore) {
 244
 245                                                 if (inputDir != "") {
 246                                                         string path = m->hasPath(accnosFileNames[i]);
 247                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 248                                                         if (path == "") {       accnosFileNames[i] = inputDir + accnosFileNames[i];             }
 249                                                 }
 250
 251                                                 ifstream in;
 252                                                 int ableToOpen = m->openInputFile(accnosFileNames[i], in, "noerror");
 253
 254                                                 //if you can't open it, try default location
 255                                                 if (ableToOpen == 1) {
 256                                                         if (m->getDefaultPath() != "") { //default path is set
 257                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(accnosFileNames[i]);
 258                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 259                                                                 ifstream in2;
 260                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 261                                                                 in2.close();
 262                                                                 accnosFileNames[i] = tryPath;
 263                                                         }
 264                                                 }
 265                                                 //if you can't open it, try default location
 266                                                 if (ableToOpen == 1) {
 267                                                         if (m->getOutputDir() != "") { //default path is set
 268                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(accnosFileNames[i]);
 269                                                                 m->mothurOut("Unable to open " + accnosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 270                                                                 ifstream in2;
 271                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 272                                                                 in2.close();
 273                                                                 accnosFileNames[i] = tryPath;
 274                                                         }
 275                                                 }
 276                                                 in.close();
 277
 278                                                 if (ableToOpen == 1) {
 279                                                         m->mothurOut("Unable to open " + accnosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 280                                                         //erase from file list
 281                                                         accnosFileNames.erase(accnosFileNames.begin()+i);
 282                                                         i--;
 283                                                 }
 284                                         }
 285                                 }
 286
 287                                 //make sure there is at least one valid file left
 288                                 if (accnosFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
 289                         }
 290
 291             oligosfile = validParameter.validFile(parameters, "oligos", false);
 292                         if (oligosfile == "not found") { oligosfile = "";  }
 293                         else {
 294                                 hasOligos = true;
 295                                 m->splitAtDash(oligosfile, oligosFileNames);
 296
 297                                 //go through files and make sure they are good, if not, then disregard them
 298                                 for (int i = 0; i < oligosFileNames.size(); i++) {
 299                                         bool ignore = false;
 300                                         if (oligosFileNames[i] == "current") {
 301                                                 oligosFileNames[i] = m->getOligosFile();
 302                                                 if (oligosFileNames[i] != "") {  m->mothurOut("Using " + oligosFileNames[i] + " as input file for the accnos parameter where you had given current."); m->mothurOutEndLine(); }
 303                                                 else {
 304                                                         m->mothurOut("You have no current oligosfile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 305                                                         //erase from file list
 306                                                         oligosFileNames.erase(oligosFileNames.begin()+i);
 307                                                         i--;
 308                                                 }
 309                                         }
 310
 311                                         if (!ignore) {
 312
 313                                                 if (inputDir != "") {
 314                                                         string path = m->hasPath(oligosFileNames[i]);
 315                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 316                                                         if (path == "") {       oligosFileNames[i] = inputDir + oligosFileNames[i];             }
 317                                                 }
 318
 319                                                 ifstream in;
 320                                                 int ableToOpen = m->openInputFile(oligosFileNames[i], in, "noerror");
 321
 322                                                 //if you can't open it, try default location
 323                                                 if (ableToOpen == 1) {
 324                                                         if (m->getDefaultPath() != "") { //default path is set
 325                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(oligosFileNames[i]);
 326                                                                 m->mothurOut("Unable to open " + oligosFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 327                                                                 ifstream in2;
 328                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 329                                                                 in2.close();
 330                                                                 oligosFileNames[i] = tryPath;
 331                                                         }
 332                                                 }
 333                                                 //if you can't open it, try default location
 334                                                 if (ableToOpen == 1) {
 335                                                         if (m->getOutputDir() != "") { //default path is set
 336                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(oligosFileNames[i]);
 337                                                                 m->mothurOut("Unable to open " + oligosFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 338                                                                 ifstream in2;
 339                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 340                                                                 in2.close();
 341                                                                 oligosFileNames[i] = tryPath;
 342                                                         }
 343                                                 }
 344                                                 in.close();
 345
 346                                                 if (ableToOpen == 1) {
 347                                                         m->mothurOut("Unable to open " + oligosFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 348                                                         //erase from file list
 349                                                         oligosFileNames.erase(oligosFileNames.begin()+i);
 350                                                         i--;
 351                                                 }
 352                                         }
 353                                 }
 354
 355                                 //make sure there is at least one valid file left
 356                                 if (oligosFileNames.size() == 0) { m->mothurOut("no valid oligos files."); m->mothurOutEndLine(); abort = true; }
 357                         }
 358
 359                         if (hasOligos) {
 360                 split = 2;
 361                                 if (oligosFileNames.size() != filenames.size()) { abort = true; m->mothurOut("If you provide a oligos file, you must have one for each sff file."); m->mothurOutEndLine(); }
 362                         }
 363
 364                         if (hasAccnos) {
 365                                 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(); }
 366                         }
 367
 368                         string temp = validParameter.validFile(parameters, "qfile", false);                     if (temp == "not found"){       temp = "T";                             }
 369                         qual = m->isTrue(temp);
 370
 371                         temp = validParameter.validFile(parameters, "fasta", false);                            if (temp == "not found"){       temp = "T";                             }
 372                         fasta = m->isTrue(temp);
 373
 374                         temp = validParameter.validFile(parameters, "flow", false);                                     if (temp == "not found"){       temp = "T";                             }
 375                         flow = m->isTrue(temp);
 376
 377                         temp = validParameter.validFile(parameters, "trim", false);                                     if (temp == "not found"){       temp = "T";                             }
 378                         trim = m->isTrue(temp);
 379
 380             temp = validParameter.validFile(parameters, "bdiffs", false);               if (temp == "not found") { temp = "0"; }
 381                         m->mothurConvert(temp, bdiffs);
 382
 383                         temp = validParameter.validFile(parameters, "pdiffs", false);           if (temp == "not found") { temp = "0"; }
 384                         m->mothurConvert(temp, pdiffs);
 385
 386             temp = validParameter.validFile(parameters, "ldiffs", false);               if (temp == "not found") { temp = "0"; }
 387                         m->mothurConvert(temp, ldiffs);
 388
 389             temp = validParameter.validFile(parameters, "sdiffs", false);               if (temp == "not found") { temp = "0"; }
 390                         m->mothurConvert(temp, sdiffs);
 391
 392                         temp = validParameter.validFile(parameters, "tdiffs", false);           if (temp == "not found") { int tempTotal = pdiffs + bdiffs + ldiffs + sdiffs;  temp = toString(tempTotal); }
 393                         m->mothurConvert(temp, tdiffs);
 394
 395                         if(tdiffs == 0){        tdiffs = bdiffs + pdiffs + ldiffs + sdiffs;     }
 396
 397                         temp = validParameter.validFile(parameters, "sfftxt", false);
 398                         if (temp == "not found")        {       temp = "F";      sfftxt = false; sfftxtFilename = "";           }
 399                         else if (m->isTrue(temp))       {       sfftxt = true;          sfftxtFilename = "";                            }
 400                         else {
 401                                 //you are a filename
 402                                 if (inputDir != "") {
 403                                         map<string,string>::iterator it = parameters.find("sfftxt");
 404                                         //user has given a template file
 405                                         if(it != parameters.end()){
 406                                                 string path = m->hasPath(it->second);
 407                                                 //if the user has not given a path then, add inputdir. else leave path alone.
 408                                                 if (path == "") {       parameters["sfftxt"] = inputDir + it->second;           }
 409                                         }
 410                                 }
 411
 412                                 sfftxtFilename = validParameter.validFile(parameters, "sfftxt", true);
 413                                 if (sfftxtFilename == "not found") { sfftxtFilename = "";  }
 414                                 else if (sfftxtFilename == "not open") { sfftxtFilename = "";  }
 415                         }
 416
 417                         if ((sfftxtFilename == "") && (filenames.size() == 0)) {
 418                                 //if there is a current sff file, use it
 419                                 string filename = m->getSFFFile();
 420                                 if (filename != "") { filenames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the sff parameter."); m->mothurOutEndLine(); }
 421                                 else {  m->mothurOut("[ERROR]: you must provide a valid sff or sfftxt file."); m->mothurOutEndLine(); abort=true;  }
 422                         }
 423
 424
 425                 }
 426         }
 427         catch(exception& e) {
 428                 m->errorOut(e, "SffInfoCommand", "SffInfoCommand");
 429                 exit(1);
 430         }
 431 }
 432 //**********************************************************************************************************************
 433 int SffInfoCommand::execute(){
 434         try {
 435                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 436
 437                 for (int s = 0; s < filenames.size(); s++) {
 438
 439                         if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 440
 441                         int start = time(NULL);
 442
 443             filenames[s] = m->getFullPathName(filenames[s]);
 444                         m->mothurOut("Extracting info from " + filenames[s] + " ..." ); m->mothurOutEndLine();
 445
 446                         string accnos = "";
 447                         if (hasAccnos) { accnos = accnosFileNames[s]; }
 448
 449             string oligos = "";
 450             if (hasOligos) { oligos = oligosFileNames[s]; }
 451
 452                         int numReads = extractSffInfo(filenames[s], accnos, oligos);
 453
 454                         m->mothurOut("It took " + toString(time(NULL) - start) + " secs to extract " + toString(numReads) + ".");
 455                 }
 456
 457                 if (sfftxtFilename != "") {  parseSffTxt(); }
 458
 459                 if (m->control_pressed) {  for (int i = 0; i < outputNames.size(); i++) {       m->mothurRemove(outputNames[i]);        } return 0; }
 460
 461                 //set fasta file as new current fastafile
 462                 string current = "";
 463                 itTypes = outputTypes.find("fasta");
 464                 if (itTypes != outputTypes.end()) {
 465                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 466                 }
 467
 468                 itTypes = outputTypes.find("qfile");
 469                 if (itTypes != outputTypes.end()) {
 470                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 471                 }
 472
 473                 itTypes = outputTypes.find("flow");
 474                 if (itTypes != outputTypes.end()) {
 475                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFlowFile(current); }
 476                 }
 477
 478                 //report output filenames
 479                 m->mothurOutEndLine();
 480                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 481                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 482                 m->mothurOutEndLine();
 483
 484                 return 0;
 485         }
 486         catch(exception& e) {
 487                 m->errorOut(e, "SffInfoCommand", "execute");
 488                 exit(1);
 489         }
 490 }
 491 //**********************************************************************************************************************
 492 int SffInfoCommand::extractSffInfo(string input, string accnos, string oligos){
 493         try {
 494                 currentFileName = input;
 495                 if (outputDir == "") {  outputDir += m->hasPath(input); }
 496
 497                 if (accnos != "")       {  readAccnosFile(accnos);  }
 498                 else                            {       seqNames.clear();               }
 499
 500         if (oligos != "")   {   readOligos(oligos);  split = 2;   }
 501
 502                 ofstream outSfftxt, outFasta, outQual, outFlow;
 503                 string outFastaFileName, outQualFileName;
 504         string rootName = outputDir + m->getRootName(m->getSimpleName(input));
 505         if(rootName.find_last_of(".") == rootName.npos){ rootName += "."; }
 506
 507                 string sfftxtFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("sfftxt");
 508                 string outFlowFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("flow");
 509                 if (trim) {
 510                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("fasta");
 511                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + getOutputFileNameTag("qfile");
 512                 }else{
 513                         outFastaFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw." + getOutputFileNameTag("fasta");
 514                         outQualFileName = outputDir + m->getRootName(m->getSimpleName(input)) + "raw." + getOutputFileNameTag("qfile");
 515                 }
 516
 517                 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); }
 518                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
 519                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
 520                 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);  }
 521
 522                 ifstream in;
 523                 in.open(input.c_str(), ios::binary);
 524
 525                 CommonHeader header;
 526                 readCommonHeader(in, header);
 527
 528                 int count = 0;
 529                 mycount = 0;
 530
 531                 //check magic number and version
 532                 if (header.magicNumber != 779314790) { m->mothurOut("Magic Number is not correct, not a valid .sff file"); m->mothurOutEndLine(); return count; }
 533                 if (header.version != "0001") { m->mothurOut("Version is not supported, only support version 0001."); m->mothurOutEndLine(); return count; }
 534
 535                 //print common header
 536                 if (sfftxt) {   printCommonHeader(outSfftxt, header);           }
 537                 if (flow)       {       outFlow << header.numFlowsPerRead << endl;      }
 538
 539                 //read through the sff file
 540                 while (!in.eof()) {
 541
 542                         bool print = true;
 543
 544                         //read data
 545                         seqRead read;  Header readheader;
 546                         readSeqData(in, read, header.numFlowsPerRead, readheader);
 547             bool okay = sanityCheck(readheader, read);
 548             if (!okay) { break; }
 549
 550                         //if you have provided an accosfile and this seq is not in it, then dont print
 551                         if (seqNames.size() != 0) {   if (seqNames.count(readheader.name) == 0) { print = false; }  }
 552
 553                         //print
 554                         if (print) {
 555                                 if (sfftxt) { printHeader(outSfftxt, readheader); printSffTxtSeqData(outSfftxt, read, readheader); }
 556                                 if (fasta)      {       printFastaSeqData(outFasta, read, readheader);  }
 557                                 if (qual)       {       printQualSeqData(outQual, read, readheader);    }
 558                                 if (flow)       {       printFlowSeqData(outFlow, read, readheader);    }
 559                         }
 560
 561                         count++;
 562                         mycount++;
 563
 564                         //report progress
 565                         if((count+1) % 10000 == 0){     m->mothurOut(toString(count+1)); m->mothurOutEndLine();         }
 566
 567                         if (m->control_pressed) { count = 0; break;   }
 568
 569                         if (count >= header.numReads) { break; }
 570                 }
 571
 572                 //report progress
 573                 if (!m->control_pressed) {   if((count) % 10000 != 0){  m->mothurOut(toString(count)); m->mothurOutEndLine();           }  }
 574
 575                 in.close();
 576
 577                 if (sfftxt) {  outSfftxt.close();       }
 578                 if (fasta)      {  outFasta.close();    }
 579                 if (qual)       {  outQual.close();             }
 580                 if (flow)       {  outFlow.close();             }
 581
 582         if (split > 1) {
 583             //create new common headers for each file with the correct number of reads
 584             adjustCommonHeader(header);
 585
 586             map<string, string> uniqueSffNames;// so we don't add the same sff multiple times
 587                         map<string, string>::iterator it;
 588                         set<string> namesToRemove;
 589                         for(int i=0;i<filehandles.size();i++){
 590                                 for(int j=0;j<filehandles[0].size();j++){
 591                                         if (filehandles[i][j] != "") {
 592                                                 if (namesToRemove.count(filehandles[i][j]) == 0) {
 593                                                         if(m->isBlank(filehandles[i][j])){
 594                                                                 m->mothurRemove(filehandles[i][j]);
 595                                 m->mothurRemove(filehandlesHeaders[i][j]);
 596                                                                 namesToRemove.insert(filehandles[i][j]);
 597                             }else{
 598                                                                 it = uniqueSffNames.find(filehandles[i][j]);
 599                                                                 if (it == uniqueSffNames.end()) {
 600                                                                         uniqueSffNames[filehandles[i][j]] = barcodeNameVector[i];
 601                                                                 }
 602                                                         }
 603                                                 }
 604                                         }
 605                                 }
 606                         }
 607
 608             //append new header to reads
 609             for (int i = 0; i < filehandles.size(); i++) {
 610                 for (int j = 0; j < filehandles[i].size(); j++) {
 611                     m->appendFiles(filehandles[i][j], filehandlesHeaders[i][j]);
 612                     m->renameFile(filehandlesHeaders[i][j], filehandles[i][j]);
 613                     m->mothurRemove(filehandlesHeaders[i][j]);
 614                     if (numSplitReads[i][j] == 0) { m->mothurRemove(filehandles[i][j]); }
 615                 }
 616             }
 617
 618                         //remove names for outputFileNames, just cleans up the output
 619                         for(int i = 0; i < outputNames.size(); i++) {
 620                 if (namesToRemove.count(outputNames[i]) != 0) {
 621                     outputNames.erase(outputNames.begin()+i);
 622                     i--;
 623                 }
 624             }
 625
 626             if(m->isBlank(noMatchFile)){  m->mothurRemove(noMatchFile); }
 627             else { outputNames.push_back(noMatchFile); outputTypes["sff"].push_back(noMatchFile); }
 628         }
 629
 630                 return count;
 631         }
 632         catch(exception& e) {
 633                 m->errorOut(e, "SffInfoCommand", "extractSffInfo");
 634                 exit(1);
 635         }
 636 }
 637 //**********************************************************************************************************************
 638 int SffInfoCommand::readCommonHeader(ifstream& in, CommonHeader& header){
 639         try {
 640
 641                 if (!in.eof()) {
 642
 643                         //read magic number
 644                         char buffer[4];
 645                         in.read(buffer, 4);
 646                         header.magicNumber = be_int4(*(unsigned int *)(&buffer));
 647
 648                         //read version
 649                         char buffer9[4];
 650                         in.read(buffer9, 4);
 651                         header.version = "";
 652                         for (int i = 0; i < 4; i++) {  header.version += toString((int)(buffer9[i]));  }
 653
 654                         //read offset
 655                         char buffer2 [8];
 656                         in.read(buffer2, 8);
 657                         header.indexOffset =  be_int8(*(unsigned long long *)(&buffer2));
 658
 659                         //read index length
 660                         char buffer3 [4];
 661                         in.read(buffer3, 4);
 662                         header.indexLength =  be_int4(*(unsigned int *)(&buffer3));
 663
 664                         //read num reads
 665                         char buffer4 [4];
 666                         in.read(buffer4, 4);
 667                         header.numReads =  be_int4(*(unsigned int *)(&buffer4));
 668
 669                         //read header length
 670                         char buffer5 [2];
 671                         in.read(buffer5, 2);
 672                         header.headerLength =  be_int2(*(unsigned short *)(&buffer5));
 673
 674                         //read key length
 675                         char buffer6 [2];
 676                         in.read(buffer6, 2);
 677                         header.keyLength = be_int2(*(unsigned short *)(&buffer6));
 678
 679                         //read number of flow reads
 680                         char buffer7 [2];
 681                         in.read(buffer7, 2);
 682                         header.numFlowsPerRead =  be_int2(*(unsigned short *)(&buffer7));
 683
 684                         //read format code
 685                         char buffer8 [1];
 686                         in.read(buffer8, 1);
 687                         header.flogramFormatCode = (int)(buffer8[0]);
 688
 689                         //read flow chars
 690                         char* tempBuffer = new char[header.numFlowsPerRead];
 691                         in.read(&(*tempBuffer), header.numFlowsPerRead);
 692                         header.flowChars = tempBuffer;
 693                         if (header.flowChars.length() > header.numFlowsPerRead) { header.flowChars = header.flowChars.substr(0, header.numFlowsPerRead);  }
 694                         delete[] tempBuffer;
 695
 696                         //read key
 697                         char* tempBuffer2 = new char[header.keyLength];
 698                         in.read(&(*tempBuffer2), header.keyLength);
 699                         header.keySequence = tempBuffer2;
 700                         if (header.keySequence.length() > header.keyLength) { header.keySequence = header.keySequence.substr(0, header.keyLength);  }
 701                         delete[] tempBuffer2;
 702
 703                         /* Pad to 8 chars */
 704                         unsigned long long spotInFile = in.tellg();
 705                         unsigned long long spot = (spotInFile + 7)& ~7;  // ~ inverts
 706                         in.seekg(spot);
 707
 708         }else{
 709                         m->mothurOut("Error reading sff common header."); m->mothurOutEndLine();
 710                 }
 711
 712                 return 0;
 713
 714         }
 715         catch(exception& e) {
 716                 m->errorOut(e, "SffInfoCommand", "readCommonHeader");
 717                 exit(1);
 718         }
 719 }
 720 //**********************************************************************************************************************
 721 int SffInfoCommand::adjustCommonHeader(CommonHeader header){
 722         try {
 723
 724         char* mybuffer = new char[4];
 725         ifstream in;
 726         in.open(currentFileName.c_str(), ios::binary);
 727
 728         //magic number
 729         in.read(mybuffer,4);
 730         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 731             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 732                 ofstream out;
 733                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 734                 out.write(mybuffer, in.gcount());
 735                 out.close();
 736             }
 737         }
 738         delete[] mybuffer;
 739
 740         //version
 741         mybuffer = new char[4];
 742         in.read(mybuffer,4);
 743         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 744             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 745                 ofstream out;
 746                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 747                 out.write(mybuffer, in.gcount());
 748                 out.close();
 749             }
 750         }
 751         delete[] mybuffer;
 752
 753         //offset
 754         mybuffer = new char[8];
 755         in.read(mybuffer,8);
 756         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 757             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 758                 ofstream out;
 759                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 760                 out.write(mybuffer, in.gcount());
 761                 out.close();
 762             }
 763         }
 764         delete[] mybuffer;
 765
 766
 767         //read index length
 768                 mybuffer = new char[4];
 769         in.read(mybuffer,4);
 770         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 771             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 772                 ofstream out;
 773                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 774                 out.write(mybuffer, in.gcount());
 775                 out.close();
 776             }
 777         }
 778         delete[] mybuffer;
 779
 780         //change num reads
 781         mybuffer = new char[4];
 782         in.read(mybuffer,4);
 783         delete[] mybuffer;
 784         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 785             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 786                 ofstream out;
 787                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 788                 //convert number of reads to 4 byte char*
 789                 char* thisbuffer = new char[4];
 790                 thisbuffer[0] = (numSplitReads[i][j] >> 24) & 0xFF;
 791                 thisbuffer[1] = (numSplitReads[i][j] >> 16) & 0xFF;
 792                 thisbuffer[2] = (numSplitReads[i][j] >> 8) & 0xFF;
 793                 thisbuffer[3] = numSplitReads[i][j] & 0xFF;
 794                 out.write(thisbuffer, 4);
 795                 out.close();
 796                 delete[] thisbuffer;
 797             }
 798         }
 799
 800         //read header length
 801         mybuffer = new char[2];
 802         in.read(mybuffer,2);
 803         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 804             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 805                 ofstream out;
 806                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 807                 out.write(mybuffer, in.gcount());
 808                 out.close();
 809             }
 810         }
 811         delete[] mybuffer;
 812
 813         //read key length
 814         mybuffer = new char[2];
 815         in.read(mybuffer,2);
 816         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 817             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 818                 ofstream out;
 819                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 820                 out.write(mybuffer, in.gcount());
 821                 out.close();
 822             }
 823         }
 824         delete[] mybuffer;
 825
 826         //read number of flow reads
 827         mybuffer = new char[2];
 828         in.read(mybuffer,2);
 829         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 830             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 831                 ofstream out;
 832                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 833                 out.write(mybuffer, in.gcount());
 834                 out.close();
 835             }
 836         }
 837         delete[] mybuffer;
 838
 839         //read format code
 840         mybuffer = new char[1];
 841         in.read(mybuffer,1);
 842         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 843             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 844                 ofstream out;
 845                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 846                 out.write(mybuffer, in.gcount());
 847                 out.close();
 848             }
 849         }
 850         delete[] mybuffer;
 851
 852         //read flow chars
 853         mybuffer = new char[header.numFlowsPerRead];
 854         in.read(mybuffer,header.numFlowsPerRead);
 855         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 856             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 857                 ofstream out;
 858                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 859                 out.write(mybuffer, in.gcount());
 860                 out.close();
 861             }
 862         }
 863         delete[] mybuffer;
 864
 865         //read key
 866         mybuffer = new char[header.keyLength];
 867         in.read(mybuffer,header.keyLength);
 868         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 869             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 870                 ofstream out;
 871                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 872                 out.write(mybuffer, in.gcount());
 873                 out.close();
 874             }
 875         }
 876         delete[] mybuffer;
 877
 878
 879         /* Pad to 8 chars */
 880         unsigned long long spotInFile = in.tellg();
 881         unsigned long long spot = (spotInFile + 7)& ~7;  // ~ inverts
 882         in.seekg(spot);
 883
 884         mybuffer = new char[spot-spotInFile];
 885         for (int i = 0; i < filehandlesHeaders.size(); i++) {
 886             for (int j = 0; j < filehandlesHeaders[i].size(); j++) {
 887                 ofstream out;
 888                 m->openOutputFileAppend(filehandlesHeaders[i][j], out);
 889                 out.write(mybuffer, spot-spotInFile);
 890                 out.close();
 891             }
 892         }
 893         delete[] mybuffer;
 894         in.close();
 895                 return 0;
 896
 897         }
 898         catch(exception& e) {
 899                 m->errorOut(e, "SffInfoCommand", "adjustCommonHeader");
 900                 exit(1);
 901         }
 902 }
 903 //**********************************************************************************************************************
 904 int SffInfoCommand::readSeqData(ifstream& in, seqRead& read, int numFlowReads, Header& header){
 905         try {
 906         unsigned long long startSpotInFile = in.tellg();
 907                 if (!in.eof()) {
 908
 909             /*****************************************/
 910             //read header
 911
 912             //read header length
 913                         char buffer [2];
 914                         in.read(buffer, 2);
 915                         header.headerLength = be_int2(*(unsigned short *)(&buffer));
 916
 917                         //read name length
 918                         char buffer2 [2];
 919                         in.read(buffer2, 2);
 920                         header.nameLength = be_int2(*(unsigned short *)(&buffer2));
 921
 922                         //read num bases
 923                         char buffer3 [4];
 924                         in.read(buffer3, 4);
 925                         header.numBases =  be_int4(*(unsigned int *)(&buffer3));
 926
 927                         //read clip qual left
 928                         char buffer4 [2];
 929                         in.read(buffer4, 2);
 930                         header.clipQualLeft =  be_int2(*(unsigned short *)(&buffer4));
 931                         header.clipQualLeft = 5;
 932
 933                         //read clip qual right
 934                         char buffer5 [2];
 935                         in.read(buffer5, 2);
 936                         header.clipQualRight =  be_int2(*(unsigned short *)(&buffer5));
 937
 938                         //read clipAdapterLeft
 939                         char buffer6 [2];
 940                         in.read(buffer6, 2);
 941                         header.clipAdapterLeft = be_int2(*(unsigned short *)(&buffer6));
 942
 943                         //read clipAdapterRight
 944                         char buffer7 [2];
 945                         in.read(buffer7, 2);
 946                         header.clipAdapterRight = be_int2(*(unsigned short *)(&buffer7));
 947
 948                         //read name
 949                         char* tempBuffer = new char[header.nameLength];
 950                         in.read(&(*tempBuffer), header.nameLength);
 951                         header.name = tempBuffer;
 952                         if (header.name.length() > header.nameLength) { header.name = header.name.substr(0, header.nameLength);  }
 953                         delete[] tempBuffer;
 954
 955                         //extract info from name
 956                         decodeName(header.timestamp, header.region, header.xy, header.name);
 957
 958                         /* Pad to 8 chars */
 959                         unsigned long long spotInFile = in.tellg();
 960                         unsigned long long spot = (spotInFile + 7)& ~7;
 961                         in.seekg(spot);
 962
 963             /*****************************************/
 964             //sequence read
 965
 966                         //read flowgram
 967                         read.flowgram.resize(numFlowReads);
 968                         for (int i = 0; i < numFlowReads; i++) {
 969                                 char buffer [2];
 970                                 in.read(buffer, 2);
 971                                 read.flowgram[i] = be_int2(*(unsigned short *)(&buffer));
 972                         }
 973
 974                         //read flowIndex
 975                         read.flowIndex.resize(header.numBases);
 976                         for (int i = 0; i < header.numBases; i++) {
 977                                 char temp[1];
 978                                 in.read(temp, 1);
 979                                 read.flowIndex[i] = be_int1(*(unsigned char *)(&temp));
 980                         }
 981
 982                         //read bases
 983                         char* tempBuffer6 = new char[header.numBases];
 984                         in.read(&(*tempBuffer6), header.numBases);
 985                         read.bases = tempBuffer6;
 986                         if (read.bases.length() > header.numBases) { read.bases = read.bases.substr(0, header.numBases);  }
 987                         delete[] tempBuffer6;
 988
 989                         //read qual scores
 990                         read.qualScores.resize(header.numBases);
 991                         for (int i = 0; i < header.numBases; i++) {
 992                                 char temp[1];
 993                                 in.read(temp, 1);
 994                                 read.qualScores[i] = be_int1(*(unsigned char *)(&temp));
 995                         }
 996
 997                         /* Pad to 8 chars */
 998                         spotInFile = in.tellg();
 999                         spot = (spotInFile + 7)& ~7;
1000                         in.seekg(spot);
1001
1002             if (split > 1) {
1003                 char * mybuffer;
1004                 mybuffer = new char [spot-startSpotInFile];
1005                 ifstream in2;
1006                 m->openInputFile(currentFileName, in2);
1007                 in2.seekg(startSpotInFile);
1008                 in2.read(mybuffer,spot-startSpotInFile);
1009                 in2.close();
1010
1011                 int barcodeIndex, primerIndex;
1012                 int trashCodeLength = findGroup(header, read, barcodeIndex, primerIndex);
1013
1014                 if(trashCodeLength == 0){
1015                     ofstream out;
1016                     m->openOutputFileAppend(filehandles[barcodeIndex][primerIndex], out);
1017                     out.write(mybuffer, in2.gcount());
1018                     out.close();
1019                     delete[] mybuffer;
1020                     numSplitReads[barcodeIndex][primerIndex]++;
1021                                 }
1022                                 else{
1023                                         ofstream out;
1024                     m->openOutputFileAppend(noMatchFile, out);
1025                     out.write(mybuffer, in2.gcount());
1026                     out.close();
1027                     delete[] mybuffer;
1028                                 }
1029
1030                         }
1031                 }else{
1032                         m->mothurOut("Error reading."); m->mothurOutEndLine();
1033                 }
1034
1035                 return 0;
1036         }
1037         catch(exception& e) {
1038                 m->errorOut(e, "SffInfoCommand", "readSeqData");
1039                 exit(1);
1040         }
1041 }
1042 //**********************************************************************************************************************
1043 int SffInfoCommand::findGroup(Header header, seqRead read, int& barcode, int& primer) {
1044         try {
1045         //find group read belongs to
1046         TrimOligos trimOligos(pdiffs, bdiffs, ldiffs, sdiffs, primers, barcodes, rbarcodes, revPrimer, linker, spacer);
1047
1048         int success = 1;
1049         string trashCode = "";
1050         int currentSeqsDiffs = 0;
1051
1052         string seq = read.bases;
1053
1054         if (trim) {
1055             if(header.clipQualRight < header.clipQualLeft){
1056                 seq = "NNNN";
1057             }
1058             else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1059                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
1060             }
1061             else {
1062                 seq = seq.substr(header.clipQualLeft-1);
1063             }
1064         }else{
1065             //if you wanted the sfftxt then you already converted the bases to the right case
1066             if (!sfftxt) {
1067                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1068                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
1069                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
1070                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
1071                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
1072             }
1073         }
1074
1075         Sequence currSeq(header.name, seq);
1076         QualityScores currQual;
1077
1078         if(numLinkers != 0){
1079             success = trimOligos.stripLinker(currSeq, currQual);
1080             if(success > ldiffs)                {       trashCode += 'k';       }
1081             else{ currentSeqsDiffs += success;  }
1082
1083         }
1084
1085         if(barcodes.size() != 0){
1086             success = trimOligos.stripBarcode(currSeq, currQual, barcode);
1087             if(success > bdiffs)                {       trashCode += 'b';       }
1088             else{ currentSeqsDiffs += success;  }
1089         }
1090
1091         if(rbarcodes.size() != 0){
1092             success = trimOligos.stripRBarcode(currSeq, currQual, barcode);
1093             if(success > bdiffs)                {       trashCode += 'b';       }
1094             else{ currentSeqsDiffs += success;  }
1095         }
1096
1097         if(numSpacers != 0){
1098             success = trimOligos.stripSpacer(currSeq, currQual);
1099             if(success > sdiffs)                {       trashCode += 's';       }
1100             else{ currentSeqsDiffs += success;  }
1101
1102         }
1103
1104         if(numFPrimers != 0){
1105             success = trimOligos.stripForward(currSeq, currQual, primer, true);
1106             if(success > pdiffs)                {       trashCode += 'f';       }
1107             else{ currentSeqsDiffs += success;  }
1108         }
1109
1110         if (currentSeqsDiffs > tdiffs)  {       trashCode += 't';   }
1111
1112         if(revPrimer.size() != 0){
1113             success = trimOligos.stripReverse(currSeq, currQual);
1114             if(!success)                                {       trashCode += 'r';       }
1115         }
1116
1117
1118         return trashCode.length();
1119     }
1120         catch(exception& e) {
1121                 m->errorOut(e, "SffInfoCommand", "findGroup");
1122                 exit(1);
1123         }
1124 }
1125 //**********************************************************************************************************************
1126 int SffInfoCommand::decodeName(string& timestamp, string& region, string& xy, string name) {
1127         try {
1128
1129                 if (name.length() >= 6) {
1130                         string time = name.substr(0, 6);
1131                         unsigned int timeNum = m->fromBase36(time);
1132
1133                         int q1 = timeNum / 60;
1134                         int sec = timeNum - 60 * q1;
1135                         int q2 = q1 / 60;
1136                         int minute = q1 - 60 * q2;
1137                         int q3 = q2 / 24;
1138                         int hr = q2 - 24 * q3;
1139                         int q4 = q3 / 32;
1140                         int day = q3 - 32 * q4;
1141                         int q5 = q4 / 13;
1142                         int mon = q4 - 13 * q5;
1143                         int year = 2000 + q5;
1144
1145                         timestamp = toString(year) + "_" + toString(mon) + "_" + toString(day) + "_" + toString(hr) + "_" + toString(minute) + "_" + toString(sec);
1146                 }
1147
1148                 if (name.length() >= 9) {
1149                         region = name.substr(7, 2);
1150
1151                         string xyNum = name.substr(9);
1152                         unsigned int myXy = m->fromBase36(xyNum);
1153                         int x = myXy >> 12;
1154                         int y = myXy & 4095;
1155
1156                         xy = toString(x) + "_" + toString(y);
1157                 }
1158
1159                 return 0;
1160         }
1161         catch(exception& e) {
1162                 m->errorOut(e, "SffInfoCommand", "decodeName");
1163                 exit(1);
1164         }
1165 }
1166 //**********************************************************************************************************************
1167 int SffInfoCommand::printCommonHeader(ofstream& out, CommonHeader& header) {
1168         try {
1169
1170                 out << "Common Header:\nMagic Number: " << header.magicNumber << endl;
1171                 out << "Version: " << header.version << endl;
1172                 out << "Index Offset: " << header.indexOffset << endl;
1173                 out << "Index Length: " << header.indexLength << endl;
1174                 out << "Number of Reads: " << header.numReads << endl;
1175                 out << "Header Length: " << header.headerLength << endl;
1176                 out << "Key Length: " << header.keyLength << endl;
1177                 out << "Number of Flows: " << header.numFlowsPerRead << endl;
1178                 out << "Format Code: " << header.flogramFormatCode << endl;
1179                 out << "Flow Chars: " << header.flowChars << endl;
1180                 out << "Key Sequence: " << header.keySequence << endl << endl;
1181
1182                 return 0;
1183         }
1184         catch(exception& e) {
1185                 m->errorOut(e, "SffInfoCommand", "printCommonHeader");
1186                 exit(1);
1187         }
1188 }
1189 //**********************************************************************************************************************
1190 int SffInfoCommand::printHeader(ofstream& out, Header& header) {
1191         try {
1192
1193                 out << ">" << header.name << endl;
1194                 out << "Run Prefix: " << header.timestamp << endl;
1195                 out << "Region #:  " << header.region << endl;
1196                 out << "XY Location: " << header.xy << endl << endl;
1197
1198                 out << "Run Name:  " << endl;
1199                 out << "Analysis Name:  " << endl;
1200                 out << "Full Path: " << endl << endl;
1201
1202                 out << "Read Header Len: " << header.headerLength << endl;
1203                 out << "Name Length: " << header.nameLength << endl;
1204                 out << "# of Bases: " << header.numBases << endl;
1205                 out << "Clip Qual Left: " << header.clipQualLeft << endl;
1206                 out << "Clip Qual Right: " << header.clipQualRight << endl;
1207                 out << "Clip Adap Left: " << header.clipAdapterLeft << endl;
1208                 out << "Clip Adap Right: " << header.clipAdapterRight << endl << endl;
1209
1210                 return 0;
1211         }
1212         catch(exception& e) {
1213                 m->errorOut(e, "SffInfoCommand", "printHeader");
1214                 exit(1);
1215         }
1216 }
1217 //**********************************************************************************************************************
1218 bool SffInfoCommand::sanityCheck(Header& header, seqRead& read) {
1219         try {
1220         bool okay = true;
1221         string message = "[WARNING]: Your sff file may be corrupted! Sequence: " + header.name + "\n";
1222
1223         if (header.clipQualLeft > read.bases.length()) {
1224             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
1225         }
1226         if (header.clipQualRight > read.bases.length()) {
1227             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.bases.length()) + " bases.\n";
1228         }
1229         if (header.clipQualLeft > read.qualScores.size()) {
1230             okay = false; message += "Clip Qual Left = " + toString(header.clipQualLeft) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
1231         }
1232         if (header.clipQualRight > read.qualScores.size()) {
1233             okay = false; message += "Clip Qual Right = " + toString(header.clipQualRight) + ", but we only read " + toString(read.qualScores.size()) + " quality scores.\n";
1234         }
1235
1236         if (okay == false) {
1237             m->mothurOut(message); m->mothurOutEndLine();
1238         }
1239
1240                 return okay;
1241         }
1242         catch(exception& e) {
1243                 m->errorOut(e, "SffInfoCommand", "sanityCheck");
1244                 exit(1);
1245         }
1246 }
1247 //**********************************************************************************************************************
1248 int SffInfoCommand::printSffTxtSeqData(ofstream& out, seqRead& read, Header& header) {
1249         try {
1250                 out << "Flowgram: ";
1251                 for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << (read.flowgram[i]/(float)100) << '\t';  }
1252
1253                 out << endl <<  "Flow Indexes: ";
1254                 int sum = 0;
1255                 for (int i = 0; i < read.flowIndex.size(); i++) {  sum +=  read.flowIndex[i];  out << sum << '\t'; }
1256
1257                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1258                 if(header.clipQualRight == 0){  header.clipQualRight = read.bases.length();     }
1259                 for (int i = 0; i < (header.clipQualLeft-1); i++) { read.bases[i] = tolower(read.bases[i]); }
1260                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   read.bases[i] = toupper(read.bases[i]);  }
1261                 for (int i = (header.clipQualRight-1); i < read.bases.length(); i++) {   read.bases[i] = tolower(read.bases[i]);  }
1262
1263                 out << endl <<  "Bases: " << read.bases << endl << "Quality Scores: ";
1264                 for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
1265
1266
1267                 out << endl << endl;
1268
1269                 return 0;
1270         }
1271         catch(exception& e) {
1272                 m->errorOut(e, "SffInfoCommand", "printSffTxtSeqData");
1273                 exit(1);
1274         }
1275 }
1276 //**********************************************************************************************************************
1277 int SffInfoCommand::printFastaSeqData(ofstream& out, seqRead& read, Header& header) {
1278         try {
1279                 string seq = read.bases;
1280
1281         if (trim) {
1282                         if(header.clipQualRight < header.clipQualLeft){
1283                                 seq = "NNNN";
1284                         }
1285                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1286                                 seq = seq.substr((header.clipQualLeft-1), (header.clipQualRight-header.clipQualLeft));
1287                         }
1288                         else {
1289                                 seq = seq.substr(header.clipQualLeft-1);
1290                         }
1291                 }else{
1292                         //if you wanted the sfftxt then you already converted the bases to the right case
1293                         if (!sfftxt) {
1294                                 //make the bases you want to clip lowercase and the bases you want to keep upper case
1295                                 if(header.clipQualRight == 0){  header.clipQualRight = seq.length();    }
1296                                 for (int i = 0; i < (header.clipQualLeft-1); i++) { seq[i] = tolower(seq[i]);  }
1297                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++)  {   seq[i] = toupper(seq[i]);  }
1298                                 for (int i = (header.clipQualRight-1); i < seq.length(); i++) {   seq[i] = tolower(seq[i]);  }
1299                         }
1300                 }
1301
1302                 out << ">" << header.name  << " xy=" << header.xy << endl;
1303                 out << seq << endl;
1304
1305                 return 0;
1306         }
1307         catch(exception& e) {
1308                 m->errorOut(e, "SffInfoCommand", "printFastaSeqData");
1309                 exit(1);
1310         }
1311 }
1312
1313 //**********************************************************************************************************************
1314 int SffInfoCommand::printQualSeqData(ofstream& out, seqRead& read, Header& header) {
1315         try {
1316
1317                 if (trim) {
1318                         if(header.clipQualRight < header.clipQualLeft){
1319                                 out << ">" << header.name << " xy=" << header.xy << endl;
1320                                 out << "0\t0\t0\t0";
1321                         }
1322                         else if((header.clipQualRight != 0) && ((header.clipQualRight-header.clipQualLeft) >= 0)){
1323                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
1324                                 for (int i = (header.clipQualLeft-1); i < (header.clipQualRight-1); i++) {   out << read.qualScores[i] << '\t'; }
1325                         }
1326                         else{
1327                                 out << ">" << header.name << " xy=" << header.xy << " length=" << (header.clipQualRight-header.clipQualLeft) << endl;
1328                                 for (int i = (header.clipQualLeft-1); i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';   }
1329                         }
1330                 }else{
1331                         out << ">" << header.name << " xy=" << header.xy << " length=" << read.qualScores.size() << endl;
1332                         for (int i = 0; i < read.qualScores.size(); i++) {   out << read.qualScores[i] << '\t';  }
1333                 }
1334
1335                 out << endl;
1336
1337                 return 0;
1338         }
1339         catch(exception& e) {
1340                 m->errorOut(e, "SffInfoCommand", "printQualSeqData");
1341                 exit(1);
1342         }
1343 }
1344
1345 //**********************************************************************************************************************
1346 int SffInfoCommand::printFlowSeqData(ofstream& out, seqRead& read, Header& header) {
1347         try {
1348                 if(header.clipQualRight > header.clipQualLeft){
1349
1350                         int rightIndex = 0;
1351                         for (int i = 0; i < header.clipQualRight; i++) {  rightIndex +=  read.flowIndex[i];     }
1352
1353                         out << header.name << ' ' << rightIndex;
1354                         for (int i = 0; i < read.flowgram.size(); i++) { out << setprecision(2) << ' ' << (read.flowgram[i]/(float)100);  }
1355                         out << endl;
1356                 }
1357
1358
1359                 return 0;
1360         }
1361         catch(exception& e) {
1362                 m->errorOut(e, "SffInfoCommand", "printFlowSeqData");
1363                 exit(1);
1364         }
1365 }
1366 //**********************************************************************************************************************
1367 int SffInfoCommand::readAccnosFile(string filename) {
1368         try {
1369                 //remove old names
1370                 seqNames.clear();
1371
1372                 ifstream in;
1373                 m->openInputFile(filename, in);
1374                 string name;
1375
1376                 while(!in.eof()){
1377                         in >> name; m->gobble(in);
1378
1379                         seqNames.insert(name);
1380
1381                         if (m->control_pressed) { seqNames.clear(); break; }
1382                 }
1383                 in.close();
1384
1385                 return 0;
1386         }
1387         catch(exception& e) {
1388                 m->errorOut(e, "SffInfoCommand", "readAccnosFile");
1389                 exit(1);
1390         }
1391 }
1392 //**********************************************************************************************************************
1393 int SffInfoCommand::parseSffTxt() {
1394         try {
1395
1396                 ifstream inSFF;
1397                 m->openInputFile(sfftxtFilename, inSFF);
1398
1399                 if (outputDir == "") {  outputDir += m->hasPath(sfftxtFilename); }
1400
1401                 //output file names
1402                 ofstream outFasta, outQual, outFlow;
1403                 string outFastaFileName, outQualFileName;
1404                 string fileRoot = m->getRootName(m->getSimpleName(sfftxtFilename));
1405                 if (fileRoot.length() > 0) {
1406                         //rip off last .
1407                         fileRoot = fileRoot.substr(0, fileRoot.length()-1);
1408                         fileRoot = m->getRootName(fileRoot);
1409                 }
1410
1411                 string outFlowFileName = outputDir + fileRoot + getOutputFileNameTag("flow");
1412                 if (trim) {
1413                         outFastaFileName = outputDir + fileRoot + getOutputFileNameTag("fasta");
1414                         outQualFileName = outputDir + fileRoot + getOutputFileNameTag("qfile");
1415                 }else{
1416                         outFastaFileName = outputDir + fileRoot + "raw." + getOutputFileNameTag("fasta");
1417                         outQualFileName = outputDir + fileRoot + "raw." + getOutputFileNameTag("qfile");
1418                 }
1419
1420                 if (fasta)      { m->openOutputFile(outFastaFileName, outFasta);        outputNames.push_back(outFastaFileName); outputTypes["fasta"].push_back(outFastaFileName); }
1421                 if (qual)       { m->openOutputFile(outQualFileName, outQual);          outputNames.push_back(outQualFileName); outputTypes["qfile"].push_back(outQualFileName);  }
1422                 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);  }
1423
1424                 //read common header
1425                 string commonHeader = m->getline(inSFF);
1426                 string magicNumber = m->getline(inSFF);
1427                 string version = m->getline(inSFF);
1428                 string indexOffset = m->getline(inSFF);
1429                 string indexLength = m->getline(inSFF);
1430                 int numReads = parseHeaderLineToInt(inSFF);
1431                 string headerLength = m->getline(inSFF);
1432                 string keyLength = m->getline(inSFF);
1433                 int numFlows = parseHeaderLineToInt(inSFF);
1434                 string flowgramCode = m->getline(inSFF);
1435                 string flowChars = m->getline(inSFF);
1436                 string keySequence = m->getline(inSFF);
1437                 m->gobble(inSFF);
1438
1439                 string seqName;
1440
1441                 if (flow)       {       outFlow << numFlows << endl;    }
1442
1443                 for(int i=0;i<numReads;i++){
1444
1445                         //sanity check
1446                         if (inSFF.eof()) { m->mothurOut("[ERROR]: Expected " + toString(numReads) + " but reached end of file at " + toString(i+1) + "."); m->mothurOutEndLine(); break; }
1447
1448                         Header header;
1449
1450                         //parse read header
1451                         inSFF >> seqName;
1452                         seqName = seqName.substr(1);
1453                         m->gobble(inSFF);
1454                         header.name = seqName;
1455
1456                         string runPrefix = parseHeaderLineToString(inSFF);              header.timestamp = runPrefix;
1457                         string regionNumber = parseHeaderLineToString(inSFF);   header.region = regionNumber;
1458                         string xyLocation = parseHeaderLineToString(inSFF);             header.xy = xyLocation;
1459                         m->gobble(inSFF);
1460
1461                         string runName = parseHeaderLineToString(inSFF);
1462                         string analysisName = parseHeaderLineToString(inSFF);
1463                         string fullPath = parseHeaderLineToString(inSFF);
1464                         m->gobble(inSFF);
1465
1466                         string readHeaderLen = parseHeaderLineToString(inSFF);  convert(readHeaderLen, header.headerLength);
1467                         string nameLength = parseHeaderLineToString(inSFF);             convert(nameLength, header.nameLength);
1468                         int numBases = parseHeaderLineToInt(inSFF);                             header.numBases = numBases;
1469                         string clipQualLeft = parseHeaderLineToString(inSFF);   convert(clipQualLeft, header.clipQualLeft);
1470                         int clipQualRight = parseHeaderLineToInt(inSFF);                header.clipQualRight = clipQualRight;
1471                         string clipAdapLeft = parseHeaderLineToString(inSFF);   convert(clipAdapLeft, header.clipAdapterLeft);
1472                         string clipAdapRight = parseHeaderLineToString(inSFF);  convert(clipAdapRight, header.clipAdapterRight);
1473                         m->gobble(inSFF);
1474
1475                         seqRead read;
1476
1477                         //parse read
1478                         vector<unsigned short> flowVector = parseHeaderLineToFloatVector(inSFF, numFlows);      read.flowgram = flowVector;
1479                         vector<unsigned int> flowIndices = parseHeaderLineToIntVector(inSFF, numBases);
1480
1481                         //adjust for print
1482                         vector<unsigned int> flowIndicesAdjusted; flowIndicesAdjusted.push_back(flowIndices[0]);
1483                         for (int j = 1; j < flowIndices.size(); j++) {   flowIndicesAdjusted.push_back(flowIndices[j] - flowIndices[j-1]);   }
1484                         read.flowIndex = flowIndicesAdjusted;
1485
1486                         string bases = parseHeaderLineToString(inSFF);                                                                          read.bases = bases;
1487                         vector<unsigned int> qualityScores = parseHeaderLineToIntVector(inSFF, numBases);       read.qualScores = qualityScores;
1488                         m->gobble(inSFF);
1489
1490                         //if you have provided an accosfile and this seq is not in it, then dont print
1491                         bool print = true;
1492                         if (seqNames.size() != 0) {   if (seqNames.count(header.name) == 0) { print = false; }  }
1493
1494                         //print
1495                         if (print) {
1496                                 if (fasta)      {       printFastaSeqData(outFasta, read, header);      }
1497                                 if (qual)       {       printQualSeqData(outQual, read, header);        }
1498                                 if (flow)       {       printFlowSeqData(outFlow, read, header);        }
1499                         }
1500
1501                         //report progress
1502                         if((i+1) % 10000 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine();             }
1503
1504                         if (m->control_pressed) {  break;  }
1505                 }
1506
1507                 //report progress
1508                 if (!m->control_pressed) {   if((numReads) % 10000 != 0){       m->mothurOut(toString(numReads)); m->mothurOutEndLine();                }  }
1509
1510                 inSFF.close();
1511
1512                 if (fasta)      {  outFasta.close();    }
1513                 if (qual)       {  outQual.close();             }
1514                 if (flow)       {  outFlow.close();             }
1515
1516                 return 0;
1517         }
1518         catch(exception& e) {
1519                 m->errorOut(e, "SffInfoCommand", "parseSffTxt");
1520                 exit(1);
1521         }
1522 }
1523 //**********************************************************************************************************************
1524
1525 int SffInfoCommand::parseHeaderLineToInt(ifstream& file){
1526         try {
1527                 int number;
1528
1529                 while (!file.eof())     {
1530
1531                         char c = file.get();
1532                         if (c == ':'){
1533                                 file >> number;
1534                                 break;
1535                         }
1536
1537                 }
1538                 m->gobble(file);
1539                 return number;
1540         }
1541         catch(exception& e) {
1542                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToInt");
1543                 exit(1);
1544         }
1545
1546 }
1547
1548 //**********************************************************************************************************************
1549
1550 string SffInfoCommand::parseHeaderLineToString(ifstream& file){
1551         try {
1552                 string text;
1553
1554                 while (!file.eof())     {
1555                         char c = file.get();
1556
1557                         if (c == ':'){
1558                                 //m->gobble(file);
1559                                 //text = m->getline(file);
1560                                 file >> text;
1561                                 break;
1562                         }
1563                 }
1564                 m->gobble(file);
1565
1566                 return text;
1567         }
1568         catch(exception& e) {
1569                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToString");
1570                 exit(1);
1571         }
1572 }
1573
1574 //**********************************************************************************************************************
1575
1576 vector<unsigned short> SffInfoCommand::parseHeaderLineToFloatVector(ifstream& file, int length){
1577         try {
1578                 vector<unsigned short> floatVector(length);
1579
1580                 while (!file.eof())     {
1581                         char c = file.get();
1582                         if (c == ':'){
1583                                 float temp;
1584                                 for(int i=0;i<length;i++){
1585                                         file >> temp;
1586                                         floatVector[i] = temp * 100;
1587                                 }
1588                                 break;
1589                         }
1590                 }
1591                 m->gobble(file);
1592                 return floatVector;
1593         }
1594         catch(exception& e) {
1595                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToFloatVector");
1596                 exit(1);
1597         }
1598 }
1599
1600 //**********************************************************************************************************************
1601
1602 vector<unsigned int> SffInfoCommand::parseHeaderLineToIntVector(ifstream& file, int length){
1603         try {
1604                 vector<unsigned int> intVector(length);
1605
1606                 while (!file.eof())     {
1607                         char c = file.get();
1608                         if (c == ':'){
1609                                 for(int i=0;i<length;i++){
1610                                         file >> intVector[i];
1611                                 }
1612                                 break;
1613                         }
1614                 }
1615                 m->gobble(file);
1616                 return intVector;
1617         }
1618         catch(exception& e) {
1619                 m->errorOut(e, "SffInfoCommand", "parseHeaderLineToIntVector");
1620                 exit(1);
1621         }
1622 }
1623 //***************************************************************************************************************
1624
1625 bool SffInfoCommand::readOligos(string oligoFile){
1626         try {
1627         filehandles.clear();
1628         numSplitReads.clear();
1629         filehandlesHeaders.clear();
1630
1631                 ifstream inOligos;
1632                 m->openInputFile(oligoFile, inOligos);
1633
1634                 string type, oligo, group;
1635
1636                 int indexPrimer = 0;
1637                 int indexBarcode = 0;
1638
1639                 while(!inOligos.eof()){
1640
1641                         inOligos >> type;
1642
1643                         if(type[0] == '#'){
1644                                 while (!inOligos.eof()) {       char c = inOligos.get();  if (c == 10 || c == 13){      break;  }       } // get rest of line if there's any crap there
1645                                 m->gobble(inOligos);
1646                         }
1647                         else{
1648                                 m->gobble(inOligos);
1649                                 //make type case insensitive
1650                                 for(int i=0;i<type.length();i++){       type[i] = toupper(type[i]);  }
1651
1652                                 inOligos >> oligo;
1653
1654                                 for(int i=0;i<oligo.length();i++){
1655                                         oligo[i] = toupper(oligo[i]);
1656                                         if(oligo[i] == 'U')     {       oligo[i] = 'T'; }
1657                                 }
1658
1659                                 if(type == "FORWARD"){
1660                                         group = "";
1661
1662                                         // get rest of line in case there is a primer name
1663                                         while (!inOligos.eof()) {
1664                                                 char c = inOligos.get();
1665                                                 if (c == 10 || c == 13){        break;  }
1666                                                 else if (c == 32 || c == 9){;} //space or tab
1667                                                 else {  group += c;  }
1668                                         }
1669
1670                                         //check for repeat barcodes
1671                                         map<string, int>::iterator itPrime = primers.find(oligo);
1672                                         if (itPrime != primers.end()) { m->mothurOut("primer " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
1673
1674                                         primers[oligo]=indexPrimer; indexPrimer++;
1675                                         primerNameVector.push_back(group);
1676                                 }else if(type == "REVERSE"){
1677                                         //Sequence oligoRC("reverse", oligo);
1678                                         //oligoRC.reverseComplement();
1679                     string oligoRC = reverseOligo(oligo);
1680                                         revPrimer.push_back(oligoRC);
1681                                 }
1682                                 else if(type == "BARCODE"){
1683                                         inOligos >> group;
1684
1685                     //barcode lines can look like   BARCODE   atgcatgc   groupName  - for 454 seqs
1686                     //or                            BARCODE   atgcatgc   atgcatgc    groupName  - for illumina data that has forward and reverse info
1687                                         string temp = "";
1688                     while (!inOligos.eof())     {
1689                                                 char c = inOligos.get();
1690                                                 if (c == 10 || c == 13){        break;  }
1691                                                 else if (c == 32 || c == 9){;} //space or tab
1692                                                 else {  temp += c;  }
1693                                         }
1694
1695                     //then this is illumina data with 4 columns
1696                     if (temp != "") {
1697                         string reverseBarcode = reverseOligo(group); //reverse barcode
1698                         group = temp;
1699
1700                         //check for repeat barcodes
1701                         map<string, int>::iterator itBar = rbarcodes.find(reverseBarcode);
1702                         if (itBar != rbarcodes.end()) { m->mothurOut("barcode " + reverseBarcode + " is in your oligos file already."); m->mothurOutEndLine();  }
1703
1704                         rbarcodes[reverseBarcode]=indexBarcode;
1705                     }
1706
1707                                         //check for repeat barcodes
1708                                         map<string, int>::iterator itBar = barcodes.find(oligo);
1709                                         if (itBar != barcodes.end()) { m->mothurOut("barcode " + oligo + " is in your oligos file already."); m->mothurOutEndLine();  }
1710
1711                                         barcodes[oligo]=indexBarcode; indexBarcode++;
1712                                         barcodeNameVector.push_back(group);
1713
1714                                 }else if(type == "LINKER"){
1715                                         linker.push_back(oligo);
1716                                 }else if(type == "SPACER"){
1717                                         spacer.push_back(oligo);
1718                                 }
1719                                 else{   m->mothurOut("[WARNING]: " + type + " is not recognized as a valid type. Choices are forward, reverse, and barcode. Ignoring " + oligo + "."); m->mothurOutEndLine(); }
1720                         }
1721                         m->gobble(inOligos);
1722                 }
1723                 inOligos.close();
1724
1725                 if(barcodeNameVector.size() == 0 && primerNameVector[0] == ""){ split = 1;      }
1726
1727                 //add in potential combos
1728                 if(barcodeNameVector.size() == 0){
1729                         barcodes[""] = 0;
1730                         barcodeNameVector.push_back("");
1731                 }
1732
1733                 if(primerNameVector.size() == 0){
1734                         primers[""] = 0;
1735                         primerNameVector.push_back("");
1736                 }
1737
1738                 filehandles.resize(barcodeNameVector.size());
1739                 for(int i=0;i<filehandles.size();i++){
1740                         filehandles[i].assign(primerNameVector.size(), "");
1741                 }
1742
1743                 if(split > 1){
1744                         set<string> uniqueNames; //used to cleanup outputFileNames
1745                         for(map<string, int>::iterator itBar = barcodes.begin();itBar != barcodes.end();itBar++){
1746                                 for(map<string, int>::iterator itPrimer = primers.begin();itPrimer != primers.end(); itPrimer++){
1747
1748                                         string primerName = primerNameVector[itPrimer->second];
1749                                         string barcodeName = barcodeNameVector[itBar->second];
1750
1751                                         string comboGroupName = "";
1752                                         string fastaFileName = "";
1753                                         string qualFileName = "";
1754                                         string nameFileName = "";
1755
1756                                         if(primerName == ""){
1757                                                 comboGroupName = barcodeNameVector[itBar->second];
1758                                         }
1759                                         else{
1760                                                 if(barcodeName == ""){
1761                                                         comboGroupName = primerNameVector[itPrimer->second];
1762                                                 }
1763                                                 else{
1764                                                         comboGroupName = barcodeNameVector[itBar->second] + "." + primerNameVector[itPrimer->second];
1765                                                 }
1766                                         }
1767
1768                                         ofstream temp;
1769                                         string thisFilename = outputDir + m->getRootName(m->getSimpleName(currentFileName)) + comboGroupName + "." + getOutputFileNameTag("sff");
1770                                         if (uniqueNames.count(thisFilename) == 0) {
1771                                                 outputNames.push_back(thisFilename);
1772                                                 outputTypes["sff"].push_back(thisFilename);
1773                                                 uniqueNames.insert(thisFilename);
1774                                         }
1775
1776                                         filehandles[itBar->second][itPrimer->second] = thisFilename;
1777                                         m->openOutputFile(thisFilename, temp);          temp.close();
1778                                 }
1779                         }
1780                 }
1781                 numFPrimers = primers.size();
1782         numLinkers = linker.size();
1783         numSpacers = spacer.size();
1784                 noMatchFile = outputDir + m->getRootName(m->getSimpleName(currentFileName)) + "scrap." + getOutputFileNameTag("sff");
1785         m->mothurRemove(noMatchFile);
1786
1787                 bool allBlank = true;
1788                 for (int i = 0; i < barcodeNameVector.size(); i++) {
1789                         if (barcodeNameVector[i] != "") {
1790                                 allBlank = false;
1791                                 break;
1792                         }
1793                 }
1794                 for (int i = 0; i < primerNameVector.size(); i++) {
1795                         if (primerNameVector[i] != "") {
1796                                 allBlank = false;
1797                                 break;
1798                         }
1799                 }
1800
1801         filehandlesHeaders.resize(filehandles.size());
1802         numSplitReads.resize(filehandles.size());
1803         for (int i = 0; i < filehandles.size(); i++) {
1804             numSplitReads[i].resize(filehandles[i].size(), 0);
1805             for (int j = 0; j < filehandles[i].size(); j++) {
1806                 filehandlesHeaders[i].push_back(filehandles[i][j]+"headers");
1807             }
1808         }
1809
1810                 if (allBlank) {
1811                         m->mothurOut("[WARNING]: your oligos file does not contain any group names.  mothur will not create a split the sff file."); m->mothurOutEndLine();
1812                         split = 1;
1813                         return false;
1814                 }
1815
1816                 return true;
1817
1818         }
1819         catch(exception& e) {
1820                 m->errorOut(e, "SffInfoCommand", "readOligos");
1821                 exit(1);
1822         }
1823 }
1824 //********************************************************************/
1825 string SffInfoCommand::reverseOligo(string oligo){
1826         try {
1827         string reverse = "";
1828
1829         for(int i=oligo.length()-1;i>=0;i--){
1830
1831             if(oligo[i] == 'A')         {       reverse += 'T'; }
1832             else if(oligo[i] == 'T'){   reverse += 'A'; }
1833             else if(oligo[i] == 'U'){   reverse += 'A'; }
1834
1835             else if(oligo[i] == 'G'){   reverse += 'C'; }
1836             else if(oligo[i] == 'C'){   reverse += 'G'; }
1837
1838             else if(oligo[i] == 'R'){   reverse += 'Y'; }
1839             else if(oligo[i] == 'Y'){   reverse += 'R'; }
1840
1841             else if(oligo[i] == 'M'){   reverse += 'K'; }
1842             else if(oligo[i] == 'K'){   reverse += 'M'; }
1843
1844             else if(oligo[i] == 'W'){   reverse += 'W'; }
1845             else if(oligo[i] == 'S'){   reverse += 'S'; }
1846
1847             else if(oligo[i] == 'B'){   reverse += 'V'; }
1848             else if(oligo[i] == 'V'){   reverse += 'B'; }
1849
1850             else if(oligo[i] == 'D'){   reverse += 'H'; }
1851             else if(oligo[i] == 'H'){   reverse += 'D'; }
1852
1853             else                                                {       reverse += 'N'; }
1854         }
1855
1856
1857         return reverse;
1858     }
1859         catch(exception& e) {
1860                 m->errorOut(e, "SffInfoCommand", "reverseOligo");
1861                 exit(1);
1862         }
1863 }
1864
1865 //**********************************************************************************************************************
1866
1867
1868
1869