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 #include "counttable.h"
  14 #include "sequencecountparser.h"
  15 //**********************************************************************************************************************
  16 vector<string> ChimeraPerseusCommand::setParameters(){
  17         try {
  18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","chimera-accnos",false,true,true); parameters.push_back(pfasta);
  19                 CommandParameter pname("name", "InputTypes", "", "", "NameCount", "NameCount", "none","",false,false,true); parameters.push_back(pname);
  20         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "NameCount", "none","",false,false,true); parameters.push_back(pcount);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "none", "none","",false,false,true); parameters.push_back(pgroup);
  22                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
  23         CommandParameter pdups("dereplicate", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pdups);
  24
  25                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  26                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  27                 CommandParameter pcutoff("cutoff", "Number", "", "0.5", "", "", "","",false,false); parameters.push_back(pcutoff);
  28                 CommandParameter palpha("alpha", "Number", "", "-5.54", "", "", "","",false,false); parameters.push_back(palpha);
  29                 CommandParameter pbeta("beta", "Number", "", "0.33", "", "", "","",false,false); parameters.push_back(pbeta);
  30
  31                 vector<string> myArray;
  32                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  33                 return myArray;
  34         }
  35         catch(exception& e) {
  36                 m->errorOut(e, "ChimeraPerseusCommand", "setParameters");
  37                 exit(1);
  38         }
  39 }
  40 //**********************************************************************************************************************
  41 string ChimeraPerseusCommand::getHelpString(){
  42         try {
  43                 string helpString = "";
  44                 helpString += "The chimera.perseus command reads a fastafile and namefile or countfile and outputs potentially chimeric sequences.\n";
  45                 helpString += "The chimera.perseus command parameters are fasta, name, group, cutoff, processors, dereplicate, alpha and beta.\n";
  46                 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";
  47                 helpString += "The name parameter allows you to provide a name file associated with your fasta file.\n";
  48         helpString += "The count parameter allows you to provide a count file associated with your fasta file. A count or name file is required. \n";
  49                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  50                 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";
  51                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  52         helpString += "If the dereplicate parameter is false, then if one group finds the seqeunce to be chimeric, then all groups find it to be chimeric, default=f.\n";
  53                 helpString += "The alpha parameter ....  The default is -5.54. \n";
  54                 helpString += "The beta parameter ....  The default is 0.33. \n";
  55                 helpString += "The cutoff parameter ....  The default is 0.50. \n";
  56                 helpString += "The chimera.perseus command should be in the following format: \n";
  57                 helpString += "chimera.perseus(fasta=yourFastaFile, name=yourNameFile) \n";
  58                 helpString += "Example: chimera.perseus(fasta=AD.align, name=AD.names) \n";
  59                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  60                 return helpString;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "ChimeraPerseusCommand", "getHelpString");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 string ChimeraPerseusCommand::getOutputPattern(string type) {
  69     try {
  70         string pattern = "";
  71
  72         if (type == "chimera") {  pattern = "[filename],perseus.chimeras"; }
  73         else if (type == "accnos") {  pattern = "[filename],perseus.accnos"; }
  74         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  75
  76         return pattern;
  77     }
  78     catch(exception& e) {
  79         m->errorOut(e, "ChimeraPerseusCommand", "getOutputPattern");
  80         exit(1);
  81     }
  82 }
  83 //**********************************************************************************************************************
  84 ChimeraPerseusCommand::ChimeraPerseusCommand(){
  85         try {
  86                 abort = true; calledHelp = true;
  87                 setParameters();
  88                 vector<string> tempOutNames;
  89                 outputTypes["chimera"] = tempOutNames;
  90                 outputTypes["accnos"] = tempOutNames;
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
  94                 exit(1);
  95         }
  96 }
  97 //***************************************************************************************************************
  98 ChimeraPerseusCommand::ChimeraPerseusCommand(string option)  {
  99         try {
 100                 abort = false; calledHelp = false;
 101         hasCount = false;
 102         hasName = false;
 103
 104                 //allow user to run help
 105                 if(option == "help") { help(); abort = true; calledHelp = true; }
 106                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 107
 108                 else {
 109                         vector<string> myArray = setParameters();
 110
 111                         OptionParser parser(option);
 112                         map<string,string> parameters = parser.getParameters();
 113
 114                         ValidParameters validParameter("chimera.perseus");
 115                         map<string,string>::iterator it;
 116
 117                         //check to make sure all parameters are valid for command
 118                         for (it = parameters.begin(); it != parameters.end(); it++) {
 119                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 120                         }
 121
 122                         vector<string> tempOutNames;
 123                         outputTypes["chimera"] = tempOutNames;
 124                         outputTypes["accnos"] = tempOutNames;
 125
 126                         //if the user changes the input directory command factory will send this info to us in the output parameter
 127                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 128                         if (inputDir == "not found"){   inputDir = "";          }
 129
 130                         //check for required parameters
 131                         fastafile = validParameter.validFile(parameters, "fasta", false);
 132                         if (fastafile == "not found") {
 133                                 //if there is a current fasta file, use it
 134                                 string filename = m->getFastaFile();
 135                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 136                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 137                         }else {
 138                                 m->splitAtDash(fastafile, fastaFileNames);
 139
 140                                 //go through files and make sure they are good, if not, then disregard them
 141                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 142
 143                                         bool ignore = false;
 144                                         if (fastaFileNames[i] == "current") {
 145                                                 fastaFileNames[i] = m->getFastaFile();
 146                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 147                                                 else {
 148                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 149                                                         //erase from file list
 150                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 151                                                         i--;
 152                                                 }
 153                                         }
 154
 155                                         if (!ignore) {
 156
 157                                                 if (inputDir != "") {
 158                                                         string path = m->hasPath(fastaFileNames[i]);
 159                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 161                                                 }
 162
 163                                                 int ableToOpen;
 164                                                 ifstream in;
 165
 166                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 167
 168                                                 //if you can't open it, try default location
 169                                                 if (ableToOpen == 1) {
 170                                                         if (m->getDefaultPath() != "") { //default path is set
 171                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 172                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 173                                                                 ifstream in2;
 174                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 175                                                                 in2.close();
 176                                                                 fastaFileNames[i] = tryPath;
 177                                                         }
 178                                                 }
 179
 180                                                 if (ableToOpen == 1) {
 181                                                         if (m->getOutputDir() != "") { //default path is set
 182                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 183                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 184                                                                 ifstream in2;
 185                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 186                                                                 in2.close();
 187                                                                 fastaFileNames[i] = tryPath;
 188                                                         }
 189                                                 }
 190
 191                                                 in.close();
 192
 193                                                 if (ableToOpen == 1) {
 194                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 195                                                         //erase from file list
 196                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 197                                                         i--;
 198                                                 }else {
 199                                                         m->setFastaFile(fastaFileNames[i]);
 200                                                 }
 201                                         }
 202                                 }
 203
 204                                 //make sure there is at least one valid file left
 205                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 206                         }
 207
 208
 209                         //check for required parameters
 210                         namefile = validParameter.validFile(parameters, "name", false);
 211                         if (namefile == "not found") { namefile = "";   }
 212                         else {
 213                                 m->splitAtDash(namefile, nameFileNames);
 214
 215                                 //go through files and make sure they are good, if not, then disregard them
 216                                 for (int i = 0; i < nameFileNames.size(); i++) {
 217
 218                                         bool ignore = false;
 219                                         if (nameFileNames[i] == "current") {
 220                                                 nameFileNames[i] = m->getNameFile();
 221                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 222                                                 else {
 223                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 224                                                         //erase from file list
 225                                                         nameFileNames.erase(nameFileNames.begin()+i);
 226                                                         i--;
 227                                                 }
 228                                         }
 229
 230                                         if (!ignore) {
 231
 232                                                 if (inputDir != "") {
 233                                                         string path = m->hasPath(nameFileNames[i]);
 234                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 235                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 236                                                 }
 237
 238                                                 int ableToOpen;
 239                                                 ifstream in;
 240
 241                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 242
 243                                                 //if you can't open it, try default location
 244                                                 if (ableToOpen == 1) {
 245                                                         if (m->getDefaultPath() != "") { //default path is set
 246                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 247                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 248                                                                 ifstream in2;
 249                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 250                                                                 in2.close();
 251                                                                 nameFileNames[i] = tryPath;
 252                                                         }
 253                                                 }
 254
 255                                                 if (ableToOpen == 1) {
 256                                                         if (m->getOutputDir() != "") { //default path is set
 257                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 258                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 259                                                                 ifstream in2;
 260                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 261                                                                 in2.close();
 262                                                                 nameFileNames[i] = tryPath;
 263                                                         }
 264                                                 }
 265
 266                                                 in.close();
 267
 268                                                 if (ableToOpen == 1) {
 269                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 270                                                         //erase from file list
 271                                                         nameFileNames.erase(nameFileNames.begin()+i);
 272                                                         i--;
 273                                                 }else {
 274                                                         m->setNameFile(nameFileNames[i]);
 275                                                 }
 276                                         }
 277                                 }
 278                         }
 279
 280             if (nameFileNames.size() != 0) { hasName = true; }
 281
 282             //check for required parameters
 283             vector<string> countfileNames;
 284                         countfile = validParameter.validFile(parameters, "count", false);
 285                         if (countfile == "not found") {
 286                 countfile = "";
 287                         }else {
 288                                 m->splitAtDash(countfile, countfileNames);
 289
 290                                 //go through files and make sure they are good, if not, then disregard them
 291                                 for (int i = 0; i < countfileNames.size(); i++) {
 292
 293                                         bool ignore = false;
 294                                         if (countfileNames[i] == "current") {
 295                                                 countfileNames[i] = m->getCountTableFile();
 296                                                 if (countfileNames[i] != "") {  m->mothurOut("Using " + countfileNames[i] + " as input file for the count parameter where you had given current."); m->mothurOutEndLine(); }
 297                                                 else {
 298                                                         m->mothurOut("You have no current count file, ignoring current."); m->mothurOutEndLine(); ignore=true;
 299                                                         //erase from file list
 300                                                         countfileNames.erase(countfileNames.begin()+i);
 301                                                         i--;
 302                                                 }
 303                                         }
 304
 305                                         if (!ignore) {
 306
 307                                                 if (inputDir != "") {
 308                                                         string path = m->hasPath(countfileNames[i]);
 309                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 310                                                         if (path == "") {       countfileNames[i] = inputDir + countfileNames[i];               }
 311                                                 }
 312
 313                                                 int ableToOpen;
 314                                                 ifstream in;
 315
 316                                                 ableToOpen = m->openInputFile(countfileNames[i], in, "noerror");
 317
 318                                                 //if you can't open it, try default location
 319                                                 if (ableToOpen == 1) {
 320                                                         if (m->getDefaultPath() != "") { //default path is set
 321                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(countfileNames[i]);
 322                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 323                                                                 ifstream in2;
 324                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 325                                                                 in2.close();
 326                                                                 countfileNames[i] = tryPath;
 327                                                         }
 328                                                 }
 329
 330                                                 if (ableToOpen == 1) {
 331                                                         if (m->getOutputDir() != "") { //default path is set
 332                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(countfileNames[i]);
 333                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 334                                                                 ifstream in2;
 335                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 336                                                                 in2.close();
 337                                                                 countfileNames[i] = tryPath;
 338                                                         }
 339                                                 }
 340
 341                                                 in.close();
 342
 343                                                 if (ableToOpen == 1) {
 344                                                         m->mothurOut("Unable to open " + countfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 345                                                         //erase from file list
 346                                                         countfileNames.erase(countfileNames.begin()+i);
 347                                                         i--;
 348                                                 }else {
 349                                                         m->setCountTableFile(countfileNames[i]);
 350                                                 }
 351                                         }
 352                                 }
 353                         }
 354
 355             if (countfileNames.size() != 0) { hasCount = true; }
 356
 357                         //make sure there is at least one valid file left
 358             if (hasName && hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 359
 360             if (!hasName && !hasCount) {
 361                 //if there is a current name file, use it, else look for current count file
 362                                 string filename = m->getNameFile();
 363                                 if (filename != "") { hasName = true; nameFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the name parameter."); m->mothurOutEndLine(); }
 364                                 else {
 365                     filename = m->getCountTableFile();
 366                     if (filename != "") { hasCount = true; countfileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the count parameter."); m->mothurOutEndLine(); }
 367                     else { m->mothurOut("[ERROR]: You must provide a count or name file."); m->mothurOutEndLine(); abort = true;  }
 368                 }
 369             }
 370             if (!hasName && hasCount) { nameFileNames = countfileNames; }
 371
 372                         if (nameFileNames.size() != fastaFileNames.size()) { m->mothurOut("[ERROR]: The number of name or count files does not match the number of fastafiles, please correct."); m->mothurOutEndLine(); abort=true; }
 373
 374                         bool hasGroup = true;
 375                         groupfile = validParameter.validFile(parameters, "group", false);
 376                         if (groupfile == "not found") { groupfile = "";  hasGroup = false; }
 377                         else {
 378                                 m->splitAtDash(groupfile, groupFileNames);
 379
 380                                 //go through files and make sure they are good, if not, then disregard them
 381                                 for (int i = 0; i < groupFileNames.size(); i++) {
 382
 383                                         bool ignore = false;
 384                                         if (groupFileNames[i] == "current") {
 385                                                 groupFileNames[i] = m->getGroupFile();
 386                                                 if (groupFileNames[i] != "") {  m->mothurOut("Using " + groupFileNames[i] + " as input file for the group parameter where you had given current."); m->mothurOutEndLine(); }
 387                                                 else {
 388                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 389                                                         //erase from file list
 390                                                         groupFileNames.erase(groupFileNames.begin()+i);
 391                                                         i--;
 392                                                 }
 393                                         }
 394
 395                                         if (!ignore) {
 396
 397                                                 if (inputDir != "") {
 398                                                         string path = m->hasPath(groupFileNames[i]);
 399                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 400                                                         if (path == "") {       groupFileNames[i] = inputDir + groupFileNames[i];               }
 401                                                 }
 402
 403                                                 int ableToOpen;
 404                                                 ifstream in;
 405
 406                                                 ableToOpen = m->openInputFile(groupFileNames[i], in, "noerror");
 407
 408                                                 //if you can't open it, try default location
 409                                                 if (ableToOpen == 1) {
 410                                                         if (m->getDefaultPath() != "") { //default path is set
 411                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(groupFileNames[i]);
 412                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 413                                                                 ifstream in2;
 414                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 415                                                                 in2.close();
 416                                                                 groupFileNames[i] = tryPath;
 417                                                         }
 418                                                 }
 419
 420                                                 if (ableToOpen == 1) {
 421                                                         if (m->getOutputDir() != "") { //default path is set
 422                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(groupFileNames[i]);
 423                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 424                                                                 ifstream in2;
 425                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 426                                                                 in2.close();
 427                                                                 groupFileNames[i] = tryPath;
 428                                                         }
 429                                                 }
 430
 431                                                 in.close();
 432
 433                                                 if (ableToOpen == 1) {
 434                                                         m->mothurOut("Unable to open " + groupFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 435                                                         //erase from file list
 436                                                         groupFileNames.erase(groupFileNames.begin()+i);
 437                                                         i--;
 438                                                 }else {
 439                                                         m->setGroupFile(groupFileNames[i]);
 440                                                 }
 441                                         }
 442                                 }
 443
 444                                 //make sure there is at least one valid file left
 445                                 if (groupFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid group files."); m->mothurOutEndLine(); abort = true; }
 446                         }
 447
 448                         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; }
 449
 450             if (hasGroup && hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or group."); m->mothurOutEndLine(); abort = true; }
 451
 452                         //if the user changes the output directory command factory will send this info to us in the output parameter
 453                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 454
 455                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 456                         m->setProcessors(temp);
 457                         m->mothurConvert(temp, processors);
 458
 459                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.50";  }
 460                         m->mothurConvert(temp, cutoff);
 461
 462                         temp = validParameter.validFile(parameters, "alpha", false);    if (temp == "not found"){       temp = "-5.54"; }
 463                         m->mothurConvert(temp, alpha);
 464
 465                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.33";  }
 466                         m->mothurConvert(temp, beta);
 467
 468                         temp = validParameter.validFile(parameters, "dereplicate", false);
 469                         if (temp == "not found") {
 470                                 if (groupfile != "")    {  temp = "false";                                      }
 471                                 else                    {  temp = "true";       }
 472                         }
 473                         dups = m->isTrue(temp);
 474                 }
 475         }
 476         catch(exception& e) {
 477                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
 478                 exit(1);
 479         }
 480 }
 481 //***************************************************************************************************************
 482
 483 int ChimeraPerseusCommand::execute(){
 484         try{
 485                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 486
 487
 488                 //process each file
 489                 for (int s = 0; s < fastaFileNames.size(); s++) {
 490
 491                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 492
 493                         int start = time(NULL);
 494                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 495                         map<string, string> variables;
 496                         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]));
 497                         string outputFileName = getOutputFileName("chimera", variables);
 498                         string accnosFileName = getOutputFileName("accnos", variables);
 499
 500                         //string newFasta = m->getRootName(fastaFileNames[s]) + "temp";
 501
 502                         //you provided a groupfile
 503                         string groupFile = "";
 504                         if (groupFileNames.size() != 0) { groupFile = groupFileNames[s]; }
 505
 506                         string nameFile = "";
 507                         if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 508                                 nameFile = nameFileNames[s];
 509                         }else { nameFile = getNamesFile(fastaFileNames[s]); }
 510
 511                         if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } return 0;     }
 512
 513                         int numSeqs = 0;
 514                         int numChimeras = 0;
 515
 516             if (hasCount) {
 517                 CountTable* ct = new CountTable();
 518                 ct->readTable(nameFile);
 519
 520                 if (ct->hasGroupInfo()) {
 521                     cparser = new SequenceCountParser(fastaFileNames[s], *ct);
 522
 523                     vector<string> groups = cparser->getNamesOfGroups();
 524
 525                     if (m->control_pressed) { delete ct; delete cparser; for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        }  return 0; }
 526
 527                     //clears files
 528                     ofstream out, out1, out2;
 529                     m->openOutputFile(outputFileName, out); out.close();
 530                     m->openOutputFile(accnosFileName, out1); out1.close();
 531
 532                     if(processors == 1) {       numSeqs = driverGroups(outputFileName, accnosFileName, 0, groups.size(), groups);       }
 533                     else                                {       numSeqs = createProcessesGroups(outputFileName, accnosFileName, groups, groupFile, fastaFileNames[s], nameFile);                        }
 534
 535                     if (m->control_pressed) {  delete ct; delete cparser; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;    }
 536                     map<string, string> uniqueNames = cparser->getAllSeqsMap();
 537                     if (!dups) {
 538                         numChimeras = deconvoluteResults(uniqueNames, outputFileName, accnosFileName);
 539                     }
 540                     delete cparser;
 541
 542                     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();
 543
 544                     if (m->control_pressed) {  delete ct; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;  }
 545
 546                 }else {
 547                     if (processors != 1) { m->mothurOut("Your count file does not contain group information, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 548
 549                     //read sequences and store sorted by frequency
 550                     vector<seqData> sequences = readFiles(fastaFileNames[s], ct);
 551
 552                     if (m->control_pressed) { delete ct; for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        } return 0; }
 553
 554                     numSeqs = driver(outputFileName, sequences, accnosFileName, numChimeras);
 555                 }
 556                 delete ct;
 557             }else {
 558                 if (groupFile != "") {
 559                     //Parse sequences by group
 560                     parser = new SequenceParser(groupFile, fastaFileNames[s], nameFile);
 561                     vector<string> groups = parser->getNamesOfGroups();
 562
 563                     if (m->control_pressed) { delete parser; for (int j = 0; j < outputNames.size(); j++) {     m->mothurRemove(outputNames[j]);        }  return 0; }
 564
 565                     //clears files
 566                     ofstream out, out1, out2;
 567                     m->openOutputFile(outputFileName, out); out.close();
 568                     m->openOutputFile(accnosFileName, out1); out1.close();
 569
 570                     if(processors == 1) {       numSeqs = driverGroups(outputFileName, accnosFileName, 0, groups.size(), groups);       }
 571                     else                                {       numSeqs = createProcessesGroups(outputFileName, accnosFileName, groups, groupFile, fastaFileNames[s], nameFile);                        }
 572
 573                     if (m->control_pressed) {  delete parser; for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        }  return 0;    }
 574                     map<string, string> uniqueNames = parser->getAllSeqsMap();
 575                     if (!dups) {
 576                         numChimeras = deconvoluteResults(uniqueNames, outputFileName, accnosFileName);
 577                     }
 578                     delete parser;
 579
 580                     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();
 581
 582                     if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  return 0;  }
 583                 }else{
 584                     if (processors != 1) { m->mothurOut("Without a groupfile, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 585
 586                     //read sequences and store sorted by frequency
 587                     vector<seqData> sequences = readFiles(fastaFileNames[s], nameFile);
 588
 589                     if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        } return 0; }
 590
 591                     numSeqs = driver(outputFileName, sequences, accnosFileName, numChimeras);
 592                 }
 593                         }
 594
 595                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 596
 597                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences. " + toString(numChimeras) + " chimeras were found.");      m->mothurOutEndLine();
 598                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 599                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 600                 }
 601
 602                 //set accnos file as new current accnosfile
 603                 string current = "";
 604                 itTypes = outputTypes.find("accnos");
 605                 if (itTypes != outputTypes.end()) {
 606                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 607                 }
 608
 609                 m->mothurOutEndLine();
 610                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 611                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 612                 m->mothurOutEndLine();
 613
 614                 return 0;
 615
 616         }
 617         catch(exception& e) {
 618                 m->errorOut(e, "ChimeraPerseusCommand", "execute");
 619                 exit(1);
 620         }
 621 }
 622 //**********************************************************************************************************************
 623 string ChimeraPerseusCommand::getNamesFile(string& inputFile){
 624         try {
 625                 string nameFile = "";
 626
 627                 m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 628
 629                 //use unique.seqs to create new name and fastafile
 630                 string inputString = "fasta=" + inputFile;
 631                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 632                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 633                 m->mothurCalling = true;
 634
 635                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 636                 uniqueCommand->execute();
 637
 638                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 639
 640                 delete uniqueCommand;
 641                 m->mothurCalling = false;
 642                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 643
 644                 nameFile = filenames["name"][0];
 645                 inputFile = filenames["fasta"][0];
 646
 647                 return nameFile;
 648         }
 649         catch(exception& e) {
 650                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 651                 exit(1);
 652         }
 653 }
 654 //**********************************************************************************************************************
 655 int ChimeraPerseusCommand::driverGroups(string outputFName, string accnos, int start, int end, vector<string> groups){
 656         try {
 657
 658                 int totalSeqs = 0;
 659                 int numChimeras = 0;
 660
 661                 for (int i = start; i < end; i++) {
 662
 663                         m->mothurOutEndLine(); m->mothurOut("Checking sequences from group " + groups[i] + "...");      m->mothurOutEndLine();
 664
 665                         int start = time(NULL);  if (m->control_pressed) {  return 0; }
 666
 667                         vector<seqData> sequences = loadSequences(groups[i]);
 668
 669                         if (m->control_pressed) { return 0; }
 670
 671                         int numSeqs = driver((outputFName + groups[i]), sequences, (accnos+groups[i]), numChimeras);
 672                         totalSeqs += numSeqs;
 673
 674                         if (m->control_pressed) { return 0; }
 675
 676                         //append files
 677                         m->appendFiles((outputFName+groups[i]), outputFName); m->mothurRemove((outputFName+groups[i]));
 678                         m->appendFiles((accnos+groups[i]), accnos); m->mothurRemove((accnos+groups[i]));
 679
 680                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences from group " + groups[i] + ".");    m->mothurOutEndLine();
 681                 }
 682
 683                 return totalSeqs;
 684
 685         }
 686         catch(exception& e) {
 687                 m->errorOut(e, "ChimeraPerseusCommand", "driverGroups");
 688                 exit(1);
 689         }
 690 }
 691 //**********************************************************************************************************************
 692 vector<seqData> ChimeraPerseusCommand::loadSequences(string group){
 693         try {
 694         bool error = false;
 695                 alignLength = 0;
 696         vector<seqData> sequences;
 697         if (hasCount) {
 698             vector<Sequence> thisGroupsSeqs = cparser->getSeqs(group);
 699             map<string, int> counts = cparser->getCountTable(group);
 700             map<string, int>::iterator it;
 701
 702             for (int i = 0; i < thisGroupsSeqs.size(); i++) {
 703
 704                 if (m->control_pressed) {  return sequences; }
 705
 706                 it = counts.find(thisGroupsSeqs[i].getName());
 707                 if (it == counts.end()) { error = true; m->mothurOut("[ERROR]: " + thisGroupsSeqs[i].getName() + " is in your fasta file and not in your count file, please correct."); m->mothurOutEndLine(); }
 708                 else {
 709                     thisGroupsSeqs[i].setAligned(removeNs(thisGroupsSeqs[i].getUnaligned()));
 710                     sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), it->second));
 711                     if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
 712                 }
 713             }
 714         }else{
 715             vector<Sequence> thisGroupsSeqs = parser->getSeqs(group);
 716             map<string, string> nameMap = parser->getNameMap(group);
 717             map<string, string>::iterator it;
 718
 719             for (int i = 0; i < thisGroupsSeqs.size(); i++) {
 720
 721                 if (m->control_pressed) {  return sequences; }
 722
 723                 it = nameMap.find(thisGroupsSeqs[i].getName());
 724                 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(); }
 725                 else {
 726                     int num = m->getNumNames(it->second);
 727                     thisGroupsSeqs[i].setAligned(removeNs(thisGroupsSeqs[i].getUnaligned()));
 728                     sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), num));
 729                     if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
 730                 }
 731             }
 732
 733                 }
 734
 735         if (error) { m->control_pressed = true; }
 736                 //sort by frequency
 737                 sort(sequences.rbegin(), sequences.rend());
 738
 739                 return sequences;
 740         }
 741         catch(exception& e) {
 742                 m->errorOut(e, "ChimeraPerseusCommand", "loadSequences");
 743                 exit(1);
 744         }
 745 }
 746
 747 //**********************************************************************************************************************
 748 vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, string name){
 749         try {
 750                 map<string, int>::iterator it;
 751                 map<string, int> nameMap = m->readNames(name);
 752
 753                 //read fasta file and create sequenceData structure - checking for file mismatches
 754                 vector<seqData> sequences;
 755                 bool error = false;
 756                 ifstream in;
 757                 m->openInputFile(inputFile, in);
 758                 alignLength = 0;
 759
 760                 while (!in.eof()) {
 761
 762                         if (m->control_pressed) { in.close(); return sequences; }
 763
 764                         Sequence temp(in); m->gobble(in);
 765
 766                         it = nameMap.find(temp.getName());
 767                         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(); }
 768                         else {
 769                 temp.setAligned(removeNs(temp.getUnaligned()));
 770                                 sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), it->second));
 771                 if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
 772                         }
 773                 }
 774                 in.close();
 775
 776                 if (error) { m->control_pressed = true; }
 777
 778                 //sort by frequency
 779                 sort(sequences.rbegin(), sequences.rend());
 780
 781                 return sequences;
 782         }
 783         catch(exception& e) {
 784                 m->errorOut(e, "ChimeraPerseusCommand", "readFiles");
 785                 exit(1);
 786         }
 787 }
 788 //**********************************************************************************************************************
 789 string ChimeraPerseusCommand::removeNs(string seq){
 790         try {
 791         string newSeq = "";
 792         for (int i = 0; i < seq.length(); i++) {
 793             if (seq[i] != 'N') {  newSeq += seq[i]; }
 794         }
 795         return newSeq;
 796     }
 797         catch(exception& e) {
 798                 m->errorOut(e, "ChimeraPerseusCommand", "removeNs");
 799                 exit(1);
 800         }
 801 }
 802 //**********************************************************************************************************************
 803 vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, CountTable* ct){
 804         try {
 805                 //read fasta file and create sequenceData structure - checking for file mismatches
 806                 vector<seqData> sequences;
 807                 ifstream in;
 808                 m->openInputFile(inputFile, in);
 809                 alignLength = 0;
 810
 811                 while (!in.eof()) {
 812             Sequence temp(in); m->gobble(in);
 813
 814                         int count = ct->getNumSeqs(temp.getName());
 815                         if (m->control_pressed) { break; }
 816                         else {
 817                 temp.setAligned(removeNs(temp.getUnaligned()));
 818                                 sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), count));
 819                 if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
 820                         }
 821                 }
 822                 in.close();
 823
 824                 //sort by frequency
 825                 sort(sequences.rbegin(), sequences.rend());
 826
 827                 return sequences;
 828         }
 829         catch(exception& e) {
 830                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 831                 exit(1);
 832         }
 833 }
 834 //**********************************************************************************************************************
 835 int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& sequences, string accnosFileName, int& numChimeras){
 836         try {
 837
 838                 vector<vector<double> > correctModel(4);        //could be an option in the future to input own model matrix
 839                 for(int i=0;i<4;i++){   correctModel[i].resize(4);      }
 840
 841                 correctModel[0][0] = 0.000000;  //AA
 842                 correctModel[1][0] = 11.619259; //CA
 843                 correctModel[2][0] = 11.694004; //TA
 844                 correctModel[3][0] = 7.748623;  //GA
 845
 846                 correctModel[1][1] = 0.000000;  //CC
 847                 correctModel[2][1] = 7.619657;  //TC
 848                 correctModel[3][1] = 12.852562; //GC
 849
 850                 correctModel[2][2] = 0.000000;  //TT
 851                 correctModel[3][2] = 10.964048; //TG
 852
 853                 correctModel[3][3] = 0.000000;  //GG
 854
 855                 for(int i=0;i<4;i++){
 856                         for(int j=0;j<i;j++){
 857                                 correctModel[j][i] = correctModel[i][j];
 858                         }
 859                 }
 860
 861                 int numSeqs = sequences.size();
 862                 //int alignLength = sequences[0].sequence.size();
 863
 864                 ofstream chimeraFile;
 865                 ofstream accnosFile;
 866                 m->openOutputFile(chimeraFileName, chimeraFile);
 867                 m->openOutputFile(accnosFileName, accnosFile);
 868
 869                 Perseus myPerseus;
 870                 vector<vector<double> > binMatrix = myPerseus.binomial(alignLength);
 871
 872                 chimeraFile << "SequenceIndex\tName\tDiffsToBestMatch\tBestMatchIndex\tBestMatchName\tDiffstToChimera\tIndexofLeftParent\tIndexOfRightParent\tNameOfLeftParent\tNameOfRightParent\tDistanceToBestMatch\tcIndex\t(cIndex - singleDist)\tloonIndex\tMismatchesToChimera\tMismatchToTrimera\tChimeraBreakPoint\tLogisticProbability\tTypeOfSequence\n";
 873
 874                 vector<bool> chimeras(numSeqs, 0);
 875
 876                 for(int i=0;i<numSeqs;i++){
 877                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 878
 879                         vector<bool> restricted = chimeras;
 880
 881                         vector<vector<int> > leftDiffs(numSeqs);
 882                         vector<vector<int> > leftMaps(numSeqs);
 883                         vector<vector<int> > rightDiffs(numSeqs);
 884                         vector<vector<int> > rightMaps(numSeqs);
 885
 886                         vector<int> singleLeft, bestLeft;
 887                         vector<int> singleRight, bestRight;
 888
 889                         int bestSingleIndex, bestSingleDiff;
 890                         vector<pwAlign> alignments(numSeqs);
 891
 892                         int comparisons = myPerseus.getAlignments(i, sequences, alignments, leftDiffs, leftMaps, rightDiffs, rightMaps, bestSingleIndex, bestSingleDiff, restricted);
 893                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 894
 895                         int minMismatchToChimera, leftParentBi, rightParentBi, breakPointBi;
 896
 897                         string dummyA, dummyB;
 898
 899             if (sequences[i].sequence.size() < 3) {
 900                 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;
 901             }else if(comparisons >= 2){
 902                                 minMismatchToChimera = myPerseus.getChimera(sequences, leftDiffs, rightDiffs, leftParentBi, rightParentBi, breakPointBi, singleLeft, bestLeft, singleRight, bestRight, restricted);
 903                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 904
 905                                 int minMismatchToTrimera = numeric_limits<int>::max();
 906                                 int leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB;
 907
 908                                 if(minMismatchToChimera >= 3 && comparisons >= 3){
 909                                         minMismatchToTrimera = myPerseus.getTrimera(sequences, leftDiffs, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, singleLeft, bestLeft, singleRight, bestRight, restricted);
 910                                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 911                                 }
 912
 913                                 double singleDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, sequences[bestSingleIndex].sequence, dummyA, dummyB, correctModel);
 914
 915                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 916
 917                                 string type;
 918                                 string chimeraRefSeq;
 919
 920                                 if(minMismatchToChimera - minMismatchToTrimera >= 3){
 921                                         type = "trimera";
 922                                         chimeraRefSeq = myPerseus.stitchTrimera(alignments, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, leftMaps, rightMaps);
 923                                 }
 924                                 else{
 925                                         type = "chimera";
 926                                         chimeraRefSeq = myPerseus.stitchBimera(alignments, leftParentBi, rightParentBi, breakPointBi, leftMaps, rightMaps);
 927                                 }
 928
 929                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 930
 931                                 double chimeraDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq, dummyA, dummyB, correctModel);
 932
 933                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 934
 935                                 double cIndex = chimeraDist;//modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq);
 936                                 double loonIndex = myPerseus.calcLoonIndex(sequences[i].sequence, sequences[leftParentBi].sequence, sequences[rightParentBi].sequence, breakPointBi, binMatrix);
 937
 938                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 939
 940                                 chimeraFile << i << '\t' << sequences[i].seqName << '\t' << bestSingleDiff << '\t' << bestSingleIndex << '\t' << sequences[bestSingleIndex].seqName << '\t';
 941                                 chimeraFile << minMismatchToChimera << '\t' << leftParentBi << '\t' << rightParentBi << '\t' << sequences[leftParentBi].seqName << '\t' << sequences[rightParentBi].seqName << '\t';
 942                                 chimeraFile << singleDist << '\t' << cIndex << '\t' << (cIndex - singleDist) << '\t' << loonIndex << '\t';
 943                                 chimeraFile << minMismatchToChimera << '\t' << minMismatchToTrimera << '\t' << breakPointBi << '\t';
 944
 945                                 double probability = myPerseus.classifyChimera(singleDist, cIndex, loonIndex, alpha, beta);
 946
 947                                 chimeraFile << probability << '\t';
 948
 949                                 if(probability > cutoff){
 950                                         chimeraFile << type << endl;
 951                                         accnosFile << sequences[i].seqName << endl;
 952                                         chimeras[i] = 1;
 953                                         numChimeras++;
 954                                 }
 955                                 else{
 956                                         chimeraFile << "good" << endl;
 957                                 }
 958
 959                         }
 960                         else{
 961                                 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;
 962                         }
 963
 964                         //report progress
 965                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1) + "\n");           }
 966                 }
 967
 968                 if((numSeqs) % 100 != 0){       m->mothurOut("Processing sequence: " + toString(numSeqs) + "\n");               }
 969
 970                 chimeraFile.close();
 971                 accnosFile.close();
 972
 973                 return numSeqs;
 974         }
 975         catch(exception& e) {
 976                 m->errorOut(e, "ChimeraPerseusCommand", "driver");
 977                 exit(1);
 978         }
 979 }
 980 /**************************************************************************************************/
 981 int ChimeraPerseusCommand::createProcessesGroups(string outputFName, string accnos, vector<string> groups, string group, string fasta, string name) {
 982         try {
 983
 984                 vector<int> processIDS;
 985                 int process = 1;
 986                 int num = 0;
 987
 988                 //sanity check
 989                 if (groups.size() < processors) { processors = groups.size(); }
 990
 991                 //divide the groups between the processors
 992                 vector<linePair> lines;
 993                 int numGroupsPerProcessor = groups.size() / processors;
 994                 for (int i = 0; i < processors; i++) {
 995                         int startIndex =  i * numGroupsPerProcessor;
 996                         int endIndex = (i+1) * numGroupsPerProcessor;
 997                         if(i == (processors - 1)){      endIndex = groups.size();       }
 998                         lines.push_back(linePair(startIndex, endIndex));
 999                 }
1000
1001 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1002
1003                 //loop through and create all the processes you want
1004                 while (process != processors) {
1005                         int pid = fork();
1006
1007                         if (pid > 0) {
1008                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1009                                 process++;
1010                         }else if (pid == 0){
1011                                 num = driverGroups(outputFName + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", lines[process].start, lines[process].end, groups);
1012
1013                                 //pass numSeqs to parent
1014                                 ofstream out;
1015                                 string tempFile = outputFName + toString(getpid()) + ".num.temp";
1016                                 m->openOutputFile(tempFile, out);
1017                                 out << num << endl;
1018                                 out.close();
1019
1020                                 exit(0);
1021                         }else {
1022                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1023                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1024                                 exit(0);
1025                         }
1026                 }
1027
1028                 //do my part
1029                 num = driverGroups(outputFName, accnos, lines[0].start, lines[0].end, groups);
1030
1031                 //force parent to wait until all the processes are done
1032                 for (int i=0;i<processIDS.size();i++) {
1033                         int temp = processIDS[i];
1034                         wait(&temp);
1035                 }
1036
1037                 for (int i = 0; i < processIDS.size(); i++) {
1038                         ifstream in;
1039                         string tempFile =  outputFName + toString(processIDS[i]) + ".num.temp";
1040                         m->openInputFile(tempFile, in);
1041                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1042                         in.close(); m->mothurRemove(tempFile);
1043                 }
1044
1045 #else
1046                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1047                 //Windows version shared memory, so be careful when passing variables through the preClusterData struct.
1048                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1049                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1050
1051                 vector<perseusData*> pDataArray;
1052                 DWORD   dwThreadIdArray[processors-1];
1053                 HANDLE  hThreadArray[processors-1];
1054
1055                 //Create processor worker threads.
1056                 for( int i=1; i<processors; i++ ){
1057                         // Allocate memory for thread data.
1058                         string extension = toString(i) + ".temp";
1059
1060                         perseusData* tempPerseus = new perseusData(hasName, hasCount, alpha, beta, cutoff, outputFName+extension, fasta, name, group, accnos+extension, groups, m, lines[i].start, lines[i].end, i);
1061
1062                         pDataArray.push_back(tempPerseus);
1063                         processIDS.push_back(i);
1064
1065                         //MyPerseusThreadFunction is in header. It must be global or static to work with the threads.
1066                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1067                         hThreadArray[i-1] = CreateThread(NULL, 0, MyPerseusThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
1068                 }
1069
1070
1071                 //using the main process as a worker saves time and memory
1072                 num = driverGroups(outputFName, accnos, lines[0].start, lines[0].end, groups);
1073
1074                 //Wait until all threads have terminated.
1075                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1076
1077                 //Close all thread handles and free memory allocations.
1078                 for(int i=0; i < pDataArray.size(); i++){
1079             if (pDataArray[i]->count != pDataArray[i]->end) {
1080                 m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end) + " sequences assigned to it, quitting. \n"); m->control_pressed = true;
1081             }
1082                         num += pDataArray[i]->count;
1083                         CloseHandle(hThreadArray[i]);
1084                         delete pDataArray[i];
1085                 }
1086 #endif
1087
1088
1089                 //append output files
1090                 for(int i=0;i<processIDS.size();i++){
1091                         m->appendFiles((outputFName + toString(processIDS[i]) + ".temp"), outputFName);
1092                         m->mothurRemove((outputFName + toString(processIDS[i]) + ".temp"));
1093
1094                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
1095                         m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));
1096                 }
1097
1098                 return num;
1099
1100         }
1101         catch(exception& e) {
1102                 m->errorOut(e, "ChimeraPerseusCommand", "createProcessesGroups");
1103                 exit(1);
1104         }
1105 }
1106 //**********************************************************************************************************************
1107 int ChimeraPerseusCommand::deconvoluteResults(map<string, string>& uniqueNames, string outputFileName, string accnosFileName){
1108         try {
1109                 map<string, string>::iterator itUnique;
1110                 int total = 0;
1111
1112                 //edit accnos file
1113                 ifstream in2;
1114                 m->openInputFile(accnosFileName, in2);
1115
1116                 ofstream out2;
1117                 m->openOutputFile(accnosFileName+".temp", out2);
1118
1119                 string name;
1120                 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
1121                 set<string>::iterator itNames;
1122                 set<string> chimerasInFile;
1123                 set<string>::iterator itChimeras;
1124
1125
1126                 while (!in2.eof()) {
1127                         if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(outputFileName); m->mothurRemove((accnosFileName+".temp")); return 0; }
1128
1129                         in2 >> name; m->gobble(in2);
1130
1131                         //find unique name
1132                         itUnique = uniqueNames.find(name);
1133
1134                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing accnos results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1135                         else {
1136                                 itChimeras = chimerasInFile.find((itUnique->second));
1137
1138                                 if (itChimeras == chimerasInFile.end()) {
1139                                         out2 << itUnique->second << endl;
1140                                         chimerasInFile.insert((itUnique->second));
1141                                         total++;
1142                                 }
1143                         }
1144                 }
1145                 in2.close();
1146                 out2.close();
1147
1148                 m->mothurRemove(accnosFileName);
1149                 rename((accnosFileName+".temp").c_str(), accnosFileName.c_str());
1150
1151                 //edit chimera file
1152                 ifstream in;
1153                 m->openInputFile(outputFileName, in);
1154
1155                 ofstream out;
1156                 m->openOutputFile(outputFileName+".temp", out); out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1157
1158                 int DiffsToBestMatch, BestMatchIndex, DiffstToChimera, IndexofLeftParent, IndexOfRightParent;
1159                 float temp1,temp2, temp3, temp4, temp5, temp6, temp7, temp8;
1160                 string index, BestMatchName, parent1, parent2, flag;
1161                 name = "";
1162                 namesInFile.clear();
1163                 //assumptions - in file each read will always look like
1164                 /*
1165                  SequenceIndex  Name    DiffsToBestMatch        BestMatchIndex  BestMatchName   DiffstToChimera IndexofLeftParent       IndexOfRightParent      NameOfLeftParent        NameOfRightParent       DistanceToBestMatch     cIndex  (cIndex - singleDist)   loonIndex       MismatchesToChimera     MismatchToTrimera       ChimeraBreakPoint       LogisticProbability     TypeOfSequence
1166                  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
1167                  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
1168                  2      F01QG4L02JZOEC  48      0       F01QG4L02JVBQY  47      0       0       F01QG4L02JVBQY  F01QG4L02JVBQY  2.0449  2.03545 -0.00944493     0       47      2147483647      138     0       good
1169                  3      F01QG4L02G7JEC  42      0       F01QG4L02JVBQY  40      1       0       F01QG4L02ICTC6  F01QG4L02JVBQY  1.87477 1.81113 -0.0636404      5.80145 40      2147483647      25      0       good
1170                  */
1171
1172                 //get and print headers
1173                 BestMatchName = m->getline(in); m->gobble(in);
1174                 out << BestMatchName << endl;
1175
1176                 while (!in.eof()) {
1177
1178                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove((outputFileName+".temp")); return 0; }
1179
1180                         bool print = false;
1181                         in >> index;    m->gobble(in);
1182
1183                         if (index != "SequenceIndex") { //if you are not a header line, there will be a header line for each group if group file is given
1184                                 in >> name;             m->gobble(in);
1185                                 in >> DiffsToBestMatch; m->gobble(in);
1186                                 in >> BestMatchIndex; m->gobble(in);
1187                                 in >> BestMatchName; m->gobble(in);
1188                                 in >> DiffstToChimera; m->gobble(in);
1189                                 in >> IndexofLeftParent; m->gobble(in);
1190                                 in >> IndexOfRightParent; m->gobble(in);
1191                                 in >> parent1;  m->gobble(in);
1192                                 in >> parent2;  m->gobble(in);
1193                                 in >> temp1 >> temp2 >> temp3 >> temp4 >> temp5 >> temp6 >> temp7 >> temp8 >> flag; m->gobble(in);
1194
1195                                 //find unique name
1196                                 itUnique = uniqueNames.find(name);
1197
1198                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1199                                 else {
1200                                         name = itUnique->second;
1201                                         //is this name already in the file
1202                                         itNames = namesInFile.find((name));
1203
1204                                         if (itNames == namesInFile.end()) { //no not in file
1205                                                 if (flag == "good") { //are you really a no??
1206                                                         //is this sequence really not chimeric??
1207                                                         itChimeras = chimerasInFile.find(name);
1208
1209                                                         //then you really are a no so print, otherwise skip
1210                                                         if (itChimeras == chimerasInFile.end()) { print = true; }
1211                                                 }else{ print = true; }
1212                                         }
1213                                 }
1214
1215                                 if (print) {
1216                                         out << index << '\t' << name  << '\t' << DiffsToBestMatch << '\t' << BestMatchIndex << '\t';
1217                                         namesInFile.insert(name);
1218
1219                                         if (BestMatchName != "Null") {
1220                                                 itUnique = uniqueNames.find(BestMatchName);
1221                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find BestMatchName "+ BestMatchName + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1222                                                 else {  out << itUnique->second << '\t';        }
1223                                         }else { out << "Null" << '\t'; }
1224
1225                                         out << DiffstToChimera << '\t' << IndexofLeftParent << '\t' << IndexOfRightParent << '\t';
1226
1227                                         if (parent1 != "Null") {
1228                                                 itUnique = uniqueNames.find(parent1);
1229                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent1 "+ parent1 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1230                                                 else {  out << itUnique->second << '\t';        }
1231                                         }else { out << "Null" << '\t'; }
1232
1233                                         if (parent1 != "Null") {
1234                                                 itUnique = uniqueNames.find(parent2);
1235                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent2 "+ parent2 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1236                                                 else {  out << itUnique->second << '\t';        }
1237                                         }else { out << "Null" << '\t'; }
1238
1239                                         out << temp1 << '\t' << temp2 << '\t' << temp3 << '\t' << temp4 << '\t' << temp5 << '\t' << temp6 << '\t' << temp7 << '\t' << temp8 << '\t' << flag << endl;
1240                                 }
1241                         }else { index = m->getline(in); m->gobble(in); }
1242                 }
1243                 in.close();
1244                 out.close();
1245
1246                 m->mothurRemove(outputFileName);
1247                 rename((outputFileName+".temp").c_str(), outputFileName.c_str());
1248
1249                 return total;
1250         }
1251         catch(exception& e) {
1252                 m->errorOut(e, "ChimeraPerseusCommand", "deconvoluteResults");
1253                 exit(1);
1254         }
1255 }
1256 //**********************************************************************************************************************
1257
1258