chimeraperseuscommand.cpp

   1 /*
   2  *  chimeraperseuscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 10/26/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraperseuscommand.h"
  11 #include "deconvolutecommand.h"
  12 #include "sequence.hpp"
  13 //**********************************************************************************************************************
  14 vector<string> ChimeraPerseusCommand::setParameters(){
  15         try {
  16                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  17                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pname);
  18                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  19                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  20                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  22                 CommandParameter pcutoff("cutoff", "Number", "", "0.5", "", "", "",false,false); parameters.push_back(pcutoff);
  23                 CommandParameter palpha("alpha", "Number", "", "-5.54", "", "", "",false,false); parameters.push_back(palpha);
  24                 CommandParameter pbeta("beta", "Number", "", "0.33", "", "", "",false,false); parameters.push_back(pbeta);
  25
  26                 vector<string> myArray;
  27                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  28                 return myArray;
  29         }
  30         catch(exception& e) {
  31                 m->errorOut(e, "ChimeraPerseusCommand", "setParameters");
  32                 exit(1);
  33         }
  34 }
  35 //**********************************************************************************************************************
  36 string ChimeraPerseusCommand::getHelpString(){
  37         try {
  38                 string helpString = "";
  39                 helpString += "The chimera.perseus command reads a fastafile and namefile and outputs potentially chimeric sequences.\n";
  40                 helpString += "The chimera.perseus command parameters are fasta, name, group, cutoff, processors, alpha and beta.\n";
  41                 helpString += "The fasta parameter allows you to enter the fasta file containing your potentially chimeric sequences, and is required, unless you have a valid current fasta file. \n";
  42                 helpString += "The name parameter allows you to provide a name file associated with your fasta file. It is required. \n";
  43                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  44                 helpString += "The group parameter allows you to provide a group file.  When checking sequences, only sequences from the same group as the query sequence will be used as the reference. \n";
  45                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  46                 helpString += "The alpha parameter ....  The default is -5.54. \n";
  47                 helpString += "The beta parameter ....  The default is 0.33. \n";
  48                 helpString += "The cutoff parameter ....  The default is 0.50. \n";
  49                 helpString += "The chimera.perseus command should be in the following format: \n";
  50                 helpString += "chimera.perseus(fasta=yourFastaFile, name=yourNameFile) \n";
  51                 helpString += "Example: chimera.perseus(fasta=AD.align, name=AD.names) \n";
  52                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  53                 return helpString;
  54         }
  55         catch(exception& e) {
  56                 m->errorOut(e, "ChimeraPerseusCommand", "getHelpString");
  57                 exit(1);
  58         }
  59 }
  60 //**********************************************************************************************************************
  61 string ChimeraPerseusCommand::getOutputFileNameTag(string type, string inputName=""){
  62         try {
  63         string outputFileName = "";
  64                 map<string, vector<string> >::iterator it;
  65
  66         //is this a type this command creates
  67         it = outputTypes.find(type);
  68         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  69         else {
  70             if (type == "chimera") {  outputFileName =  "perseus.chimeras"; }
  71             else if (type == "accnos") {  outputFileName =  "perseus.accnos"; }
  72             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  73         }
  74         return outputFileName;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "ChimeraPerseusCommand", "getOutputFileNameTag");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82 ChimeraPerseusCommand::ChimeraPerseusCommand(){
  83         try {
  84                 abort = true; calledHelp = true;
  85                 setParameters();
  86                 vector<string> tempOutNames;
  87                 outputTypes["chimera"] = tempOutNames;
  88                 outputTypes["accnos"] = tempOutNames;
  89         }
  90         catch(exception& e) {
  91                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
  92                 exit(1);
  93         }
  94 }
  95 //***************************************************************************************************************
  96 ChimeraPerseusCommand::ChimeraPerseusCommand(string option)  {
  97         try {
  98                 abort = false; calledHelp = false;
  99
 100                 //allow user to run help
 101                 if(option == "help") { help(); abort = true; calledHelp = true; }
 102                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 103
 104                 else {
 105                         vector<string> myArray = setParameters();
 106
 107                         OptionParser parser(option);
 108                         map<string,string> parameters = parser.getParameters();
 109
 110                         ValidParameters validParameter("chimera.uchime");
 111                         map<string,string>::iterator it;
 112
 113                         //check to make sure all parameters are valid for command
 114                         for (it = parameters.begin(); it != parameters.end(); it++) {
 115                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 116                         }
 117
 118                         vector<string> tempOutNames;
 119                         outputTypes["chimera"] = tempOutNames;
 120                         outputTypes["accnos"] = tempOutNames;
 121
 122                         //if the user changes the input directory command factory will send this info to us in the output parameter
 123                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 124                         if (inputDir == "not found"){   inputDir = "";          }
 125
 126                         //check for required parameters
 127                         fastafile = validParameter.validFile(parameters, "fasta", false);
 128                         if (fastafile == "not found") {
 129                                 //if there is a current fasta file, use it
 130                                 string filename = m->getFastaFile();
 131                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 132                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 133                         }else {
 134                                 m->splitAtDash(fastafile, fastaFileNames);
 135
 136                                 //go through files and make sure they are good, if not, then disregard them
 137                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 138
 139                                         bool ignore = false;
 140                                         if (fastaFileNames[i] == "current") {
 141                                                 fastaFileNames[i] = m->getFastaFile();
 142                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 143                                                 else {
 144                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 145                                                         //erase from file list
 146                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 147                                                         i--;
 148                                                 }
 149                                         }
 150
 151                                         if (!ignore) {
 152
 153                                                 if (inputDir != "") {
 154                                                         string path = m->hasPath(fastaFileNames[i]);
 155                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 156                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 157                                                 }
 158
 159                                                 int ableToOpen;
 160                                                 ifstream in;
 161
 162                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 163
 164                                                 //if you can't open it, try default location
 165                                                 if (ableToOpen == 1) {
 166                                                         if (m->getDefaultPath() != "") { //default path is set
 167                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 168                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 169                                                                 ifstream in2;
 170                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 171                                                                 in2.close();
 172                                                                 fastaFileNames[i] = tryPath;
 173                                                         }
 174                                                 }
 175
 176                                                 if (ableToOpen == 1) {
 177                                                         if (m->getOutputDir() != "") { //default path is set
 178                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 179                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 180                                                                 ifstream in2;
 181                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 182                                                                 in2.close();
 183                                                                 fastaFileNames[i] = tryPath;
 184                                                         }
 185                                                 }
 186
 187                                                 in.close();
 188
 189                                                 if (ableToOpen == 1) {
 190                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 191                                                         //erase from file list
 192                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 193                                                         i--;
 194                                                 }else {
 195                                                         m->setFastaFile(fastaFileNames[i]);
 196                                                 }
 197                                         }
 198                                 }
 199
 200                                 //make sure there is at least one valid file left
 201                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 202                         }
 203
 204
 205                         //check for required parameters
 206                         bool hasName = true;
 207                         namefile = validParameter.validFile(parameters, "name", false);
 208                         if (namefile == "not found") {
 209                                 //if there is a current fasta file, use it
 210                                 string filename = m->getNameFile();
 211                                 if (filename != "") { nameFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the name parameter."); m->mothurOutEndLine(); }
 212                                 else {  m->mothurOut("You have no current namefile and the name parameter is required."); m->mothurOutEndLine(); abort = true; }
 213                                 hasName = false;
 214                         }else {
 215                                 m->splitAtDash(namefile, nameFileNames);
 216
 217                                 //go through files and make sure they are good, if not, then disregard them
 218                                 for (int i = 0; i < nameFileNames.size(); i++) {
 219
 220                                         bool ignore = false;
 221                                         if (nameFileNames[i] == "current") {
 222                                                 nameFileNames[i] = m->getNameFile();
 223                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 224                                                 else {
 225                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 226                                                         //erase from file list
 227                                                         nameFileNames.erase(nameFileNames.begin()+i);
 228                                                         i--;
 229                                                 }
 230                                         }
 231
 232                                         if (!ignore) {
 233
 234                                                 if (inputDir != "") {
 235                                                         string path = m->hasPath(nameFileNames[i]);
 236                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 237                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 238                                                 }
 239
 240                                                 int ableToOpen;
 241                                                 ifstream in;
 242
 243                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 244
 245                                                 //if you can't open it, try default location
 246                                                 if (ableToOpen == 1) {
 247                                                         if (m->getDefaultPath() != "") { //default path is set
 248                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 249                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 250                                                                 ifstream in2;
 251                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 252                                                                 in2.close();
 253                                                                 nameFileNames[i] = tryPath;
 254                                                         }
 255                                                 }
 256
 257                                                 if (ableToOpen == 1) {
 258                                                         if (m->getOutputDir() != "") { //default path is set
 259                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 260                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 261                                                                 ifstream in2;
 262                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 263                                                                 in2.close();
 264                                                                 nameFileNames[i] = tryPath;
 265                                                         }
 266                                                 }
 267
 268                                                 in.close();
 269
 270                                                 if (ableToOpen == 1) {
 271                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 272                                                         //erase from file list
 273                                                         nameFileNames.erase(nameFileNames.begin()+i);
 274                                                         i--;
 275                                                 }else {
 276                                                         m->setNameFile(nameFileNames[i]);
 277                                                 }
 278                                         }
 279                                 }
 280
 281                                 //make sure there is at least one valid file left
 282                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 283                         }
 284
 285                         if (hasName && (nameFileNames.size() != fastaFileNames.size())) { m->mothurOut("[ERROR]: The number of namefiles does not match the number of fastafiles, please correct."); m->mothurOutEndLine(); abort=true; }
 286
 287                         bool hasGroup = true;
 288                         groupfile = validParameter.validFile(parameters, "group", false);
 289                         if (groupfile == "not found") { groupfile = "";  hasGroup = false; }
 290                         else {
 291                                 m->splitAtDash(groupfile, groupFileNames);
 292
 293                                 //go through files and make sure they are good, if not, then disregard them
 294                                 for (int i = 0; i < groupFileNames.size(); i++) {
 295
 296                                         bool ignore = false;
 297                                         if (groupFileNames[i] == "current") {
 298                                                 groupFileNames[i] = m->getGroupFile();
 299                                                 if (groupFileNames[i] != "") {  m->mothurOut("Using " + groupFileNames[i] + " as input file for the group parameter where you had given current."); m->mothurOutEndLine(); }
 300                                                 else {
 301                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 302                                                         //erase from file list
 303                                                         groupFileNames.erase(groupFileNames.begin()+i);
 304                                                         i--;
 305                                                 }
 306                                         }
 307
 308                                         if (!ignore) {
 309
 310                                                 if (inputDir != "") {
 311                                                         string path = m->hasPath(groupFileNames[i]);
 312                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 313                                                         if (path == "") {       groupFileNames[i] = inputDir + groupFileNames[i];               }
 314                                                 }
 315
 316                                                 int ableToOpen;
 317                                                 ifstream in;
 318
 319                                                 ableToOpen = m->openInputFile(groupFileNames[i], in, "noerror");
 320
 321                                                 //if you can't open it, try default location
 322                                                 if (ableToOpen == 1) {
 323                                                         if (m->getDefaultPath() != "") { //default path is set
 324                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(groupFileNames[i]);
 325                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 326                                                                 ifstream in2;
 327                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 328                                                                 in2.close();
 329                                                                 groupFileNames[i] = tryPath;
 330                                                         }
 331                                                 }
 332
 333                                                 if (ableToOpen == 1) {
 334                                                         if (m->getOutputDir() != "") { //default path is set
 335                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(groupFileNames[i]);
 336                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 337                                                                 ifstream in2;
 338                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 339                                                                 in2.close();
 340                                                                 groupFileNames[i] = tryPath;
 341                                                         }
 342                                                 }
 343
 344                                                 in.close();
 345
 346                                                 if (ableToOpen == 1) {
 347                                                         m->mothurOut("Unable to open " + groupFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 348                                                         //erase from file list
 349                                                         groupFileNames.erase(groupFileNames.begin()+i);
 350                                                         i--;
 351                                                 }else {
 352                                                         m->setGroupFile(groupFileNames[i]);
 353                                                 }
 354                                         }
 355                                 }
 356
 357                                 //make sure there is at least one valid file left
 358                                 if (groupFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid group files."); m->mothurOutEndLine(); abort = true; }
 359                         }
 360
 361                         if (hasGroup && (groupFileNames.size() != fastaFileNames.size())) { m->mothurOut("[ERROR]: The number of groupfiles does not match the number of fastafiles, please correct."); m->mothurOutEndLine(); abort=true; }
 362
 363
 364                         //if the user changes the output directory command factory will send this info to us in the output parameter
 365                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 366
 367                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 368                         m->setProcessors(temp);
 369                         m->mothurConvert(temp, processors);
 370
 371                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.50";  }
 372                         m->mothurConvert(temp, cutoff);
 373
 374                         temp = validParameter.validFile(parameters, "alpha", false);    if (temp == "not found"){       temp = "-5.54"; }
 375                         m->mothurConvert(temp, alpha);
 376
 377                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.33";  }
 378                         m->mothurConvert(temp, beta);
 379                 }
 380         }
 381         catch(exception& e) {
 382                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
 383                 exit(1);
 384         }
 385 }
 386 //***************************************************************************************************************
 387
 388 int ChimeraPerseusCommand::execute(){
 389         try{
 390                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 391
 392
 393                 //process each file
 394                 for (int s = 0; s < fastaFileNames.size(); s++) {
 395
 396                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 397
 398                         int start = time(NULL);
 399                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 400                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("chimera");
 401                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + getOutputFileNameTag("accnos");
 402
 403                         //string newFasta = m->getRootName(fastaFileNames[s]) + "temp";
 404
 405                         //you provided a groupfile
 406                         string groupFile = "";
 407                         if (groupFileNames.size() != 0) { groupFile = groupFileNames[s]; }
 408
 409                         string nameFile = "";
 410                         if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 411                                 nameFile = nameFileNames[s];
 412                         }else { nameFile = getNamesFile(fastaFileNames[s]); }
 413
 414                         if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } return 0;     }
 415
 416                         int numSeqs = 0;
 417                         int numChimeras = 0;
 418
 419                         if (groupFile != "") {
 420                                 //Parse sequences by group
 421                                 SequenceParser parser(groupFile, fastaFileNames[s], nameFile);
 422                                 vector<string> groups = parser.getNamesOfGroups();
 423
 424                                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0; }
 425
 426                                 //clears files
 427                                 ofstream out, out1, out2;
 428                                 m->openOutputFile(outputFileName, out); out.close();
 429                                 m->openOutputFile(accnosFileName, out1); out1.close();
 430
 431                                 if(processors == 1)     {       numSeqs = driverGroups(parser, outputFileName, accnosFileName, 0, groups.size(), groups);       }
 432                                 else                            {       numSeqs = createProcessesGroups(parser, outputFileName, accnosFileName, groups, groupFile, fastaFileNames[s], nameFile);                        }
 433
 434                                 if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        }  return 0;    }
 435
 436                                 numChimeras = deconvoluteResults(parser, outputFileName, accnosFileName);
 437
 438                                 m->mothurOut("The number of sequences checked may be larger than the number of unique sequences because some sequences are found in several samples."); m->mothurOutEndLine();
 439
 440                                 if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        }  return 0;    }
 441
 442                         }else{
 443                                 if (processors != 1) { m->mothurOut("Without a groupfile, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 444
 445                                 //read sequences and store sorted by frequency
 446                                 vector<seqData> sequences = readFiles(fastaFileNames[s], nameFile);
 447
 448                                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 449
 450                                 numSeqs = driver(outputFileName, sequences, accnosFileName, numChimeras);
 451                         }
 452
 453                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 454
 455                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences. " + toString(numChimeras) + " chimeras were found.");      m->mothurOutEndLine();
 456                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 457                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 458                 }
 459
 460                 //set accnos file as new current accnosfile
 461                 string current = "";
 462                 itTypes = outputTypes.find("accnos");
 463                 if (itTypes != outputTypes.end()) {
 464                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 465                 }
 466
 467                 m->mothurOutEndLine();
 468                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 469                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 470                 m->mothurOutEndLine();
 471
 472                 return 0;
 473
 474         }
 475         catch(exception& e) {
 476                 m->errorOut(e, "ChimeraPerseusCommand", "execute");
 477                 exit(1);
 478         }
 479 }
 480 //**********************************************************************************************************************
 481 string ChimeraPerseusCommand::getNamesFile(string& inputFile){
 482         try {
 483                 string nameFile = "";
 484
 485                 m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 486
 487                 //use unique.seqs to create new name and fastafile
 488                 string inputString = "fasta=" + inputFile;
 489                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 490                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 491                 m->mothurCalling = true;
 492
 493                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 494                 uniqueCommand->execute();
 495
 496                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 497
 498                 delete uniqueCommand;
 499                 m->mothurCalling = false;
 500                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 501
 502                 nameFile = filenames["name"][0];
 503                 inputFile = filenames["fasta"][0];
 504
 505                 return nameFile;
 506         }
 507         catch(exception& e) {
 508                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 509                 exit(1);
 510         }
 511 }
 512 //**********************************************************************************************************************
 513 int ChimeraPerseusCommand::driverGroups(SequenceParser& parser, string outputFName, string accnos, int start, int end, vector<string> groups){
 514         try {
 515
 516                 int totalSeqs = 0;
 517                 int numChimeras = 0;
 518
 519                 for (int i = start; i < end; i++) {
 520
 521                         m->mothurOutEndLine(); m->mothurOut("Checking sequences from group " + groups[i] + "...");      m->mothurOutEndLine();
 522
 523                         int start = time(NULL);  if (m->control_pressed) {  return 0; }
 524
 525                         vector<seqData> sequences = loadSequences(parser, groups[i]);
 526
 527                         if (m->control_pressed) { return 0; }
 528
 529                         int numSeqs = driver((outputFName + groups[i]), sequences, (accnos+groups[i]), numChimeras);
 530                         totalSeqs += numSeqs;
 531
 532                         if (m->control_pressed) { return 0; }
 533
 534                         //append files
 535                         m->appendFiles((outputFName+groups[i]), outputFName); m->mothurRemove((outputFName+groups[i]));
 536                         m->appendFiles((accnos+groups[i]), accnos); m->mothurRemove((accnos+groups[i]));
 537
 538                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences from group " + groups[i] + ".");    m->mothurOutEndLine();
 539                 }
 540
 541                 return totalSeqs;
 542
 543         }
 544         catch(exception& e) {
 545                 m->errorOut(e, "ChimeraPerseusCommand", "driverGroups");
 546                 exit(1);
 547         }
 548 }
 549 //**********************************************************************************************************************
 550 vector<seqData> ChimeraPerseusCommand::loadSequences(SequenceParser& parser, string group){
 551         try {
 552
 553                 vector<Sequence> thisGroupsSeqs = parser.getSeqs(group);
 554                 map<string, string> nameMap = parser.getNameMap(group);
 555                 map<string, string>::iterator it;
 556
 557                 vector<seqData> sequences;
 558                 bool error = false;
 559         alignLength = 0;
 560
 561                 for (int i = 0; i < thisGroupsSeqs.size(); i++) {
 562
 563                         if (m->control_pressed) {  return sequences; }
 564
 565                         it = nameMap.find(thisGroupsSeqs[i].getName());
 566                         if (it == nameMap.end()) { error = true; m->mothurOut("[ERROR]: " + thisGroupsSeqs[i].getName() + " is in your fasta file and not in your namefile, please correct."); m->mothurOutEndLine(); }
 567                         else {
 568                                 int num = m->getNumNames(it->second);
 569                                 sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), num));
 570                 if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
 571                         }
 572                 }
 573
 574                 if (error) { m->control_pressed = true; }
 575
 576                 //sort by frequency
 577                 sort(sequences.rbegin(), sequences.rend());
 578
 579                 return sequences;
 580         }
 581         catch(exception& e) {
 582                 m->errorOut(e, "ChimeraPerseusCommand", "loadSequences");
 583                 exit(1);
 584         }
 585 }
 586
 587 //**********************************************************************************************************************
 588 vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, string name){
 589         try {
 590                 map<string, int>::iterator it;
 591                 map<string, int> nameMap = m->readNames(name);
 592
 593                 //read fasta file and create sequenceData structure - checking for file mismatches
 594                 vector<seqData> sequences;
 595                 bool error = false;
 596                 ifstream in;
 597                 m->openInputFile(inputFile, in);
 598                 alignLength = 0;
 599
 600                 while (!in.eof()) {
 601
 602                         if (m->control_pressed) { in.close(); return sequences; }
 603
 604                         Sequence temp(in); m->gobble(in);
 605
 606                         it = nameMap.find(temp.getName());
 607                         if (it == nameMap.end()) { error = true; m->mothurOut("[ERROR]: " + temp.getName() + " is in your fasta file and not in your namefile, please correct."); m->mothurOutEndLine(); }
 608                         else {
 609                                 sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), it->second));
 610                 if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
 611                         }
 612                 }
 613                 in.close();
 614
 615                 if (error) { m->control_pressed = true; }
 616
 617                 //sort by frequency
 618                 sort(sequences.rbegin(), sequences.rend());
 619
 620                 return sequences;
 621         }
 622         catch(exception& e) {
 623                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 624                 exit(1);
 625         }
 626 }
 627 //**********************************************************************************************************************
 628 int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& sequences, string accnosFileName, int& numChimeras){
 629         try {
 630
 631                 vector<vector<double> > correctModel(4);        //could be an option in the future to input own model matrix
 632                 for(int i=0;i<4;i++){   correctModel[i].resize(4);      }
 633
 634                 correctModel[0][0] = 0.000000;  //AA
 635                 correctModel[1][0] = 11.619259; //CA
 636                 correctModel[2][0] = 11.694004; //TA
 637                 correctModel[3][0] = 7.748623;  //GA
 638
 639                 correctModel[1][1] = 0.000000;  //CC
 640                 correctModel[2][1] = 7.619657;  //TC
 641                 correctModel[3][1] = 12.852562; //GC
 642
 643                 correctModel[2][2] = 0.000000;  //TT
 644                 correctModel[3][2] = 10.964048; //TG
 645
 646                 correctModel[3][3] = 0.000000;  //GG
 647
 648                 for(int i=0;i<4;i++){
 649                         for(int j=0;j<i;j++){
 650                                 correctModel[j][i] = correctModel[i][j];
 651                         }
 652                 }
 653
 654                 int numSeqs = sequences.size();
 655                 //int alignLength = sequences[0].sequence.size();
 656
 657                 ofstream chimeraFile;
 658                 ofstream accnosFile;
 659                 m->openOutputFile(chimeraFileName, chimeraFile);
 660                 m->openOutputFile(accnosFileName, accnosFile);
 661
 662                 Perseus myPerseus;
 663                 vector<vector<double> > binMatrix = myPerseus.binomial(alignLength);
 664
 665                 chimeraFile << "SequenceIndex\tName\tDiffsToBestMatch\tBestMatchIndex\tBestMatchName\tDiffstToChimera\tIndexofLeftParent\tIndexOfRightParent\tNameOfLeftParent\tNameOfRightParent\tDistanceToBestMatch\tcIndex\t(cIndex - singleDist)\tloonIndex\tMismatchesToChimera\tMismatchToTrimera\tChimeraBreakPoint\tLogisticProbability\tTypeOfSequence\n";
 666
 667                 vector<bool> chimeras(numSeqs, 0);
 668
 669                 for(int i=0;i<numSeqs;i++){
 670                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 671
 672                         vector<bool> restricted = chimeras;
 673
 674                         vector<vector<int> > leftDiffs(numSeqs);
 675                         vector<vector<int> > leftMaps(numSeqs);
 676                         vector<vector<int> > rightDiffs(numSeqs);
 677                         vector<vector<int> > rightMaps(numSeqs);
 678
 679                         vector<int> singleLeft, bestLeft;
 680                         vector<int> singleRight, bestRight;
 681
 682                         int bestSingleIndex, bestSingleDiff;
 683                         vector<pwAlign> alignments(numSeqs);
 684
 685                         int comparisons = myPerseus.getAlignments(i, sequences, alignments, leftDiffs, leftMaps, rightDiffs, rightMaps, bestSingleIndex, bestSingleDiff, restricted);
 686                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 687
 688                         int minMismatchToChimera, leftParentBi, rightParentBi, breakPointBi;
 689
 690                         string dummyA, dummyB;
 691
 692             if (sequences[i].sequence.size() < 3) {
 693                 chimeraFile << i << '\t' << sequences[i].seqName << "\t0\t0\tNull\t0\t0\t0\tNull\tNull\t0.0\t0.0\t0.0\t0\t0\t0\t0.0\t0.0\tgood" << endl;
 694             }else if(comparisons >= 2){
 695                                 minMismatchToChimera = myPerseus.getChimera(sequences, leftDiffs, rightDiffs, leftParentBi, rightParentBi, breakPointBi, singleLeft, bestLeft, singleRight, bestRight, restricted);
 696                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 697
 698                                 int minMismatchToTrimera = numeric_limits<int>::max();
 699                                 int leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB;
 700
 701                                 if(minMismatchToChimera >= 3 && comparisons >= 3){
 702                                         minMismatchToTrimera = myPerseus.getTrimera(sequences, leftDiffs, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, singleLeft, bestLeft, singleRight, bestRight, restricted);
 703                                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 704                                 }
 705
 706                                 double singleDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, sequences[bestSingleIndex].sequence, dummyA, dummyB, correctModel);
 707
 708                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 709
 710                                 string type;
 711                                 string chimeraRefSeq;
 712
 713                                 if(minMismatchToChimera - minMismatchToTrimera >= 3){
 714                                         type = "trimera";
 715                                         chimeraRefSeq = myPerseus.stitchTrimera(alignments, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, leftMaps, rightMaps);
 716                                 }
 717                                 else{
 718                                         type = "chimera";
 719                                         chimeraRefSeq = myPerseus.stitchBimera(alignments, leftParentBi, rightParentBi, breakPointBi, leftMaps, rightMaps);
 720                                 }
 721                                 ;
 722                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 723
 724                                 double chimeraDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq, dummyA, dummyB, correctModel);
 725
 726                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 727
 728                                 double cIndex = chimeraDist;//modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq);
 729                                 double loonIndex = myPerseus.calcLoonIndex(sequences[i].sequence, sequences[leftParentBi].sequence, sequences[rightParentBi].sequence, breakPointBi, binMatrix);
 730
 731                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 732
 733                                 chimeraFile << i << '\t' << sequences[i].seqName << '\t' << bestSingleDiff << '\t' << bestSingleIndex << '\t' << sequences[bestSingleIndex].seqName << '\t';
 734                                 chimeraFile << minMismatchToChimera << '\t' << leftParentBi << '\t' << rightParentBi << '\t' << sequences[leftParentBi].seqName << '\t' << sequences[rightParentBi].seqName << '\t';
 735                                 chimeraFile << singleDist << '\t' << cIndex << '\t' << (cIndex - singleDist) << '\t' << loonIndex << '\t';
 736                                 chimeraFile << minMismatchToChimera << '\t' << minMismatchToTrimera << '\t' << breakPointBi << '\t';
 737
 738                                 double probability = myPerseus.classifyChimera(singleDist, cIndex, loonIndex, alpha, beta);
 739
 740                                 chimeraFile << probability << '\t';
 741
 742                                 if(probability > cutoff){
 743                                         chimeraFile << type << endl;
 744                                         accnosFile << sequences[i].seqName << endl;
 745                                         chimeras[i] = 1;
 746                                         numChimeras++;
 747                                 }
 748                                 else{
 749                                         chimeraFile << "good" << endl;
 750                                 }
 751
 752                         }
 753                         else{
 754                                 chimeraFile << i << '\t' << sequences[i].seqName << "\t0\t0\tNull\t0\t0\t0\tNull\tNull\t0.0\t0.0\t0.0\t0\t0\t0\t0.0\t0.0\tgood" << endl;
 755                         }
 756
 757                         //report progress
 758                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1) + "\n");           }
 759                 }
 760
 761                 if((numSeqs) % 100 != 0){       m->mothurOut("Processing sequence: " + toString(numSeqs) + "\n");               }
 762
 763                 chimeraFile.close();
 764                 accnosFile.close();
 765
 766                 return numSeqs;
 767         }
 768         catch(exception& e) {
 769                 m->errorOut(e, "ChimeraPerseusCommand", "driver");
 770                 exit(1);
 771         }
 772 }
 773 /**************************************************************************************************/
 774 int ChimeraPerseusCommand::createProcessesGroups(SequenceParser& parser, string outputFName, string accnos, vector<string> groups, string group, string fasta, string name) {
 775         try {
 776
 777                 vector<int> processIDS;
 778                 int process = 1;
 779                 int num = 0;
 780
 781                 //sanity check
 782                 if (groups.size() < processors) { processors = groups.size(); }
 783
 784                 //divide the groups between the processors
 785                 vector<linePair> lines;
 786                 int numGroupsPerProcessor = groups.size() / processors;
 787                 for (int i = 0; i < processors; i++) {
 788                         int startIndex =  i * numGroupsPerProcessor;
 789                         int endIndex = (i+1) * numGroupsPerProcessor;
 790                         if(i == (processors - 1)){      endIndex = groups.size();       }
 791                         lines.push_back(linePair(startIndex, endIndex));
 792                 }
 793
 794 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
 795
 796                 //loop through and create all the processes you want
 797                 while (process != processors) {
 798                         int pid = fork();
 799
 800                         if (pid > 0) {
 801                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 802                                 process++;
 803                         }else if (pid == 0){
 804                                 num = driverGroups(parser, outputFName + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", lines[process].start, lines[process].end, groups);
 805
 806                                 //pass numSeqs to parent
 807                                 ofstream out;
 808                                 string tempFile = outputFName + toString(getpid()) + ".num.temp";
 809                                 m->openOutputFile(tempFile, out);
 810                                 out << num << endl;
 811                                 out.close();
 812
 813                                 exit(0);
 814                         }else {
 815                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 816                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 817                                 exit(0);
 818                         }
 819                 }
 820
 821                 //do my part
 822                 num = driverGroups(parser, outputFName, accnos, lines[0].start, lines[0].end, groups);
 823
 824                 //force parent to wait until all the processes are done
 825                 for (int i=0;i<processIDS.size();i++) {
 826                         int temp = processIDS[i];
 827                         wait(&temp);
 828                 }
 829
 830                 for (int i = 0; i < processIDS.size(); i++) {
 831                         ifstream in;
 832                         string tempFile =  outputFName + toString(processIDS[i]) + ".num.temp";
 833                         m->openInputFile(tempFile, in);
 834                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 835                         in.close(); m->mothurRemove(tempFile);
 836                 }
 837
 838 #else
 839                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 840                 //Windows version shared memory, so be careful when passing variables through the preClusterData struct.
 841                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
 842                 //////////////////////////////////////////////////////////////////////////////////////////////////////
 843
 844                 vector<perseusData*> pDataArray;
 845                 DWORD   dwThreadIdArray[processors-1];
 846                 HANDLE  hThreadArray[processors-1];
 847
 848                 //Create processor worker threads.
 849                 for( int i=1; i<processors; i++ ){
 850                         // Allocate memory for thread data.
 851                         string extension = toString(i) + ".temp";
 852
 853                         perseusData* tempPerseus = new perseusData(alpha, beta, cutoff, outputFName+extension, fasta, name, group, accnos+extension, groups, m, lines[i].start, lines[i].end, i);
 854
 855                         pDataArray.push_back(tempPerseus);
 856                         processIDS.push_back(i);
 857
 858                         //MyPerseusThreadFunction is in header. It must be global or static to work with the threads.
 859                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
 860                         hThreadArray[i-1] = CreateThread(NULL, 0, MyPerseusThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
 861                 }
 862
 863
 864                 //using the main process as a worker saves time and memory
 865                 num = driverGroups(parser, outputFName, accnos, lines[0].start, lines[0].end, groups);
 866
 867                 //Wait until all threads have terminated.
 868                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
 869
 870                 //Close all thread handles and free memory allocations.
 871                 for(int i=0; i < pDataArray.size(); i++){
 872                         num += pDataArray[i]->count;
 873                         CloseHandle(hThreadArray[i]);
 874                         delete pDataArray[i];
 875                 }
 876 #endif
 877
 878
 879                 //append output files
 880                 for(int i=0;i<processIDS.size();i++){
 881                         m->appendFiles((outputFName + toString(processIDS[i]) + ".temp"), outputFName);
 882                         m->mothurRemove((outputFName + toString(processIDS[i]) + ".temp"));
 883
 884                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
 885                         m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));
 886                 }
 887
 888                 return num;
 889
 890         }
 891         catch(exception& e) {
 892                 m->errorOut(e, "ChimeraPerseusCommand", "createProcessesGroups");
 893                 exit(1);
 894         }
 895 }
 896 //**********************************************************************************************************************
 897 int ChimeraPerseusCommand::deconvoluteResults(SequenceParser& parser, string outputFileName, string accnosFileName){
 898         try {
 899                 map<string, string> uniqueNames = parser.getAllSeqsMap();
 900                 map<string, string>::iterator itUnique;
 901                 int total = 0;
 902
 903                 //edit accnos file
 904                 ifstream in2;
 905                 m->openInputFile(accnosFileName, in2);
 906
 907                 ofstream out2;
 908                 m->openOutputFile(accnosFileName+".temp", out2);
 909
 910                 string name;
 911                 set<string> namesInFile; //this is so if a sequence is found to be chimera in several samples we dont write it to the results file more than once
 912                 set<string>::iterator itNames;
 913                 set<string> chimerasInFile;
 914                 set<string>::iterator itChimeras;
 915
 916
 917                 while (!in2.eof()) {
 918                         if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(outputFileName); m->mothurRemove((accnosFileName+".temp")); return 0; }
 919
 920                         in2 >> name; m->gobble(in2);
 921
 922                         //find unique name
 923                         itUnique = uniqueNames.find(name);
 924
 925                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing accnos results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
 926                         else {
 927                                 itChimeras = chimerasInFile.find((itUnique->second));
 928
 929                                 if (itChimeras == chimerasInFile.end()) {
 930                                         out2 << itUnique->second << endl;
 931                                         chimerasInFile.insert((itUnique->second));
 932                                         total++;
 933                                 }
 934                         }
 935                 }
 936                 in2.close();
 937                 out2.close();
 938
 939                 m->mothurRemove(accnosFileName);
 940                 rename((accnosFileName+".temp").c_str(), accnosFileName.c_str());
 941
 942                 //edit chimera file
 943                 ifstream in;
 944                 m->openInputFile(outputFileName, in);
 945
 946                 ofstream out;
 947                 m->openOutputFile(outputFileName+".temp", out); out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 948
 949                 int DiffsToBestMatch, BestMatchIndex, DiffstToChimera, IndexofLeftParent, IndexOfRightParent;
 950                 float temp1,temp2, temp3, temp4, temp5, temp6, temp7, temp8;
 951                 string index, BestMatchName, parent1, parent2, flag;
 952                 name = "";
 953                 namesInFile.clear();
 954                 //assumptions - in file each read will always look like
 955                 /*
 956                  SequenceIndex  Name    DiffsToBestMatch        BestMatchIndex  BestMatchName   DiffstToChimera IndexofLeftParent       IndexOfRightParent      NameOfLeftParent        NameOfRightParent       DistanceToBestMatch     cIndex  (cIndex - singleDist)   loonIndex       MismatchesToChimera     MismatchToTrimera       ChimeraBreakPoint       LogisticProbability     TypeOfSequence
 957                  0      F01QG4L02JVBQY  0       0       Null    0       0       0       Null    Null    0.0     0.0     0.0     0.0     0       0       0       0.0     0.0     good
 958                  1      F01QG4L02ICTC6  0       0       Null    0       0       0       Null    Null    0.0     0.0     0.0     0.0     0       0       0       0.0     0.0     good
 959                  2      F01QG4L02JZOEC  48      0       F01QG4L02JVBQY  47      0       0       F01QG4L02JVBQY  F01QG4L02JVBQY  2.0449  2.03545 -0.00944493     0       47      2147483647      138     0       good
 960                  3      F01QG4L02G7JEC  42      0       F01QG4L02JVBQY  40      1       0       F01QG4L02ICTC6  F01QG4L02JVBQY  1.87477 1.81113 -0.0636404      5.80145 40      2147483647      25      0       good
 961                  */
 962
 963                 //get and print headers
 964                 BestMatchName = m->getline(in); m->gobble(in);
 965                 out << BestMatchName << endl;
 966
 967                 while (!in.eof()) {
 968
 969                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove((outputFileName+".temp")); return 0; }
 970
 971                         bool print = false;
 972                         in >> index;    m->gobble(in);
 973
 974                         if (index != "SequenceIndex") { //if you are not a header line, there will be a header line for each group if group file is given
 975                                 in >> name;             m->gobble(in);
 976                                 in >> DiffsToBestMatch; m->gobble(in);
 977                                 in >> BestMatchIndex; m->gobble(in);
 978                                 in >> BestMatchName; m->gobble(in);
 979                                 in >> DiffstToChimera; m->gobble(in);
 980                                 in >> IndexofLeftParent; m->gobble(in);
 981                                 in >> IndexOfRightParent; m->gobble(in);
 982                                 in >> parent1;  m->gobble(in);
 983                                 in >> parent2;  m->gobble(in);
 984                                 in >> temp1 >> temp2 >> temp3 >> temp4 >> temp5 >> temp6 >> temp7 >> temp8 >> flag; m->gobble(in);
 985
 986                                 //find unique name
 987                                 itUnique = uniqueNames.find(name);
 988
 989                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
 990                                 else {
 991                                         name = itUnique->second;
 992                                         //is this name already in the file
 993                                         itNames = namesInFile.find((name));
 994
 995                                         if (itNames == namesInFile.end()) { //no not in file
 996                                                 if (flag == "good") { //are you really a no??
 997                                                         //is this sequence really not chimeric??
 998                                                         itChimeras = chimerasInFile.find(name);
 999
1000                                                         //then you really are a no so print, otherwise skip
1001                                                         if (itChimeras == chimerasInFile.end()) { print = true; }
1002                                                 }else{ print = true; }
1003                                         }
1004                                 }
1005
1006                                 if (print) {
1007                                         out << index << '\t' << name  << '\t' << DiffsToBestMatch << '\t' << BestMatchIndex << '\t';
1008                                         namesInFile.insert(name);
1009
1010                                         if (BestMatchName != "Null") {
1011                                                 itUnique = uniqueNames.find(BestMatchName);
1012                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find BestMatchName "+ BestMatchName + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1013                                                 else {  out << itUnique->second << '\t';        }
1014                                         }else { out << "Null" << '\t'; }
1015
1016                                         out << DiffstToChimera << '\t' << IndexofLeftParent << '\t' << IndexOfRightParent << '\t';
1017
1018                                         if (parent1 != "Null") {
1019                                                 itUnique = uniqueNames.find(parent1);
1020                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent1 "+ parent1 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1021                                                 else {  out << itUnique->second << '\t';        }
1022                                         }else { out << "Null" << '\t'; }
1023
1024                                         if (parent1 != "Null") {
1025                                                 itUnique = uniqueNames.find(parent2);
1026                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent2 "+ parent2 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1027                                                 else {  out << itUnique->second << '\t';        }
1028                                         }else { out << "Null" << '\t'; }
1029
1030                                         out << temp1 << '\t' << temp2 << '\t' << temp3 << '\t' << temp4 << '\t' << temp5 << '\t' << temp6 << '\t' << temp7 << '\t' << temp8 << '\t' << flag << endl;
1031                                 }
1032                         }else { index = m->getline(in); m->gobble(in); }
1033                 }
1034                 in.close();
1035                 out.close();
1036
1037                 m->mothurRemove(outputFileName);
1038                 rename((outputFileName+".temp").c_str(), outputFileName.c_str());
1039
1040                 return total;
1041         }
1042         catch(exception& e) {
1043                 m->errorOut(e, "ChimeraPerseusCommand", "deconvoluteResults");
1044                 exit(1);
1045         }
1046 }
1047 //**********************************************************************************************************************
1048
1049