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. When you use a count file with group info and dereplicate=T, mothur will create a *.pick.count_table file containing seqeunces after chimeras are removed.\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 if (type == "count") {  pattern = "[filename],perseus.pick.count_table"; }
  75         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  76
  77         return pattern;
  78     }
  79     catch(exception& e) {
  80         m->errorOut(e, "ChimeraPerseusCommand", "getOutputPattern");
  81         exit(1);
  82     }
  83 }
  84 //**********************************************************************************************************************
  85 ChimeraPerseusCommand::ChimeraPerseusCommand(){
  86         try {
  87                 abort = true; calledHelp = true;
  88                 setParameters();
  89                 vector<string> tempOutNames;
  90                 outputTypes["chimera"] = tempOutNames;
  91                 outputTypes["accnos"] = tempOutNames;
  92         outputTypes["count"] = tempOutNames;
  93         }
  94         catch(exception& e) {
  95                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
  96                 exit(1);
  97         }
  98 }
  99 //***************************************************************************************************************
 100 ChimeraPerseusCommand::ChimeraPerseusCommand(string option)  {
 101         try {
 102                 abort = false; calledHelp = false;
 103         hasCount = false;
 104         hasName = false;
 105
 106                 //allow user to run help
 107                 if(option == "help") { help(); abort = true; calledHelp = true; }
 108                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 109
 110                 else {
 111                         vector<string> myArray = setParameters();
 112
 113                         OptionParser parser(option);
 114                         map<string,string> parameters = parser.getParameters();
 115
 116                         ValidParameters validParameter("chimera.perseus");
 117                         map<string,string>::iterator it;
 118
 119                         //check to make sure all parameters are valid for command
 120                         for (it = parameters.begin(); it != parameters.end(); it++) {
 121                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 122                         }
 123
 124                         vector<string> tempOutNames;
 125                         outputTypes["chimera"] = tempOutNames;
 126                         outputTypes["accnos"] = tempOutNames;
 127             outputTypes["count"] = tempOutNames;
 128
 129                         //if the user changes the input directory command factory will send this info to us in the output parameter
 130                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 131                         if (inputDir == "not found"){   inputDir = "";          }
 132
 133                         //check for required parameters
 134                         fastafile = validParameter.validFile(parameters, "fasta", false);
 135                         if (fastafile == "not found") {
 136                                 //if there is a current fasta file, use it
 137                                 string filename = m->getFastaFile();
 138                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 139                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 140                         }else {
 141                                 m->splitAtDash(fastafile, fastaFileNames);
 142
 143                                 //go through files and make sure they are good, if not, then disregard them
 144                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 145
 146                                         bool ignore = false;
 147                                         if (fastaFileNames[i] == "current") {
 148                                                 fastaFileNames[i] = m->getFastaFile();
 149                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 150                                                 else {
 151                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 152                                                         //erase from file list
 153                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 154                                                         i--;
 155                                                 }
 156                                         }
 157
 158                                         if (!ignore) {
 159
 160                                                 if (inputDir != "") {
 161                                                         string path = m->hasPath(fastaFileNames[i]);
 162                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 164                                                 }
 165
 166                                                 int ableToOpen;
 167                                                 ifstream in;
 168
 169                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 170
 171                                                 //if you can't open it, try default location
 172                                                 if (ableToOpen == 1) {
 173                                                         if (m->getDefaultPath() != "") { //default path is set
 174                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 175                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 176                                                                 ifstream in2;
 177                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 178                                                                 in2.close();
 179                                                                 fastaFileNames[i] = tryPath;
 180                                                         }
 181                                                 }
 182
 183                                                 if (ableToOpen == 1) {
 184                                                         if (m->getOutputDir() != "") { //default path is set
 185                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 186                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 187                                                                 ifstream in2;
 188                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 189                                                                 in2.close();
 190                                                                 fastaFileNames[i] = tryPath;
 191                                                         }
 192                                                 }
 193
 194                                                 in.close();
 195
 196                                                 if (ableToOpen == 1) {
 197                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 198                                                         //erase from file list
 199                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 200                                                         i--;
 201                                                 }else {
 202                                                         m->setFastaFile(fastaFileNames[i]);
 203                                                 }
 204                                         }
 205                                 }
 206
 207                                 //make sure there is at least one valid file left
 208                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 209                         }
 210
 211
 212                         //check for required parameters
 213                         namefile = validParameter.validFile(parameters, "name", false);
 214                         if (namefile == "not found") { namefile = "";   }
 215                         else {
 216                                 m->splitAtDash(namefile, nameFileNames);
 217
 218                                 //go through files and make sure they are good, if not, then disregard them
 219                                 for (int i = 0; i < nameFileNames.size(); i++) {
 220
 221                                         bool ignore = false;
 222                                         if (nameFileNames[i] == "current") {
 223                                                 nameFileNames[i] = m->getNameFile();
 224                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 225                                                 else {
 226                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 227                                                         //erase from file list
 228                                                         nameFileNames.erase(nameFileNames.begin()+i);
 229                                                         i--;
 230                                                 }
 231                                         }
 232
 233                                         if (!ignore) {
 234
 235                                                 if (inputDir != "") {
 236                                                         string path = m->hasPath(nameFileNames[i]);
 237                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 238                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 239                                                 }
 240
 241                                                 int ableToOpen;
 242                                                 ifstream in;
 243
 244                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 245
 246                                                 //if you can't open it, try default location
 247                                                 if (ableToOpen == 1) {
 248                                                         if (m->getDefaultPath() != "") { //default path is set
 249                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 250                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 251                                                                 ifstream in2;
 252                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 253                                                                 in2.close();
 254                                                                 nameFileNames[i] = tryPath;
 255                                                         }
 256                                                 }
 257
 258                                                 if (ableToOpen == 1) {
 259                                                         if (m->getOutputDir() != "") { //default path is set
 260                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 261                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 262                                                                 ifstream in2;
 263                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 264                                                                 in2.close();
 265                                                                 nameFileNames[i] = tryPath;
 266                                                         }
 267                                                 }
 268
 269                                                 in.close();
 270
 271                                                 if (ableToOpen == 1) {
 272                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 273                                                         //erase from file list
 274                                                         nameFileNames.erase(nameFileNames.begin()+i);
 275                                                         i--;
 276                                                 }else {
 277                                                         m->setNameFile(nameFileNames[i]);
 278                                                 }
 279                                         }
 280                                 }
 281                         }
 282
 283             if (nameFileNames.size() != 0) { hasName = true; }
 284
 285             //check for required parameters
 286             vector<string> countfileNames;
 287                         countfile = validParameter.validFile(parameters, "count", false);
 288                         if (countfile == "not found") {
 289                 countfile = "";
 290                         }else {
 291                                 m->splitAtDash(countfile, countfileNames);
 292
 293                                 //go through files and make sure they are good, if not, then disregard them
 294                                 for (int i = 0; i < countfileNames.size(); i++) {
 295
 296                                         bool ignore = false;
 297                                         if (countfileNames[i] == "current") {
 298                                                 countfileNames[i] = m->getCountTableFile();
 299                                                 if (countfileNames[i] != "") {  m->mothurOut("Using " + countfileNames[i] + " as input file for the count parameter where you had given current."); m->mothurOutEndLine(); }
 300                                                 else {
 301                                                         m->mothurOut("You have no current count file, ignoring current."); m->mothurOutEndLine(); ignore=true;
 302                                                         //erase from file list
 303                                                         countfileNames.erase(countfileNames.begin()+i);
 304                                                         i--;
 305                                                 }
 306                                         }
 307
 308                                         if (!ignore) {
 309
 310                                                 if (inputDir != "") {
 311                                                         string path = m->hasPath(countfileNames[i]);
 312                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 313                                                         if (path == "") {       countfileNames[i] = inputDir + countfileNames[i];               }
 314                                                 }
 315
 316                                                 int ableToOpen;
 317                                                 ifstream in;
 318
 319                                                 ableToOpen = m->openInputFile(countfileNames[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(countfileNames[i]);
 325                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 326                                                                 ifstream in2;
 327                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 328                                                                 in2.close();
 329                                                                 countfileNames[i] = tryPath;
 330                                                         }
 331                                                 }
 332
 333                                                 if (ableToOpen == 1) {
 334                                                         if (m->getOutputDir() != "") { //default path is set
 335                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(countfileNames[i]);
 336                                                                 m->mothurOut("Unable to open " + countfileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 337                                                                 ifstream in2;
 338                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 339                                                                 in2.close();
 340                                                                 countfileNames[i] = tryPath;
 341                                                         }
 342                                                 }
 343
 344                                                 in.close();
 345
 346                                                 if (ableToOpen == 1) {
 347                                                         m->mothurOut("Unable to open " + countfileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 348                                                         //erase from file list
 349                                                         countfileNames.erase(countfileNames.begin()+i);
 350                                                         i--;
 351                                                 }else {
 352                                                         m->setCountTableFile(countfileNames[i]);
 353                                                 }
 354                                         }
 355                                 }
 356                         }
 357
 358             if (countfileNames.size() != 0) { hasCount = true; }
 359
 360                         //make sure there is at least one valid file left
 361             if (hasName && hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
 362
 363             if (!hasName && !hasCount) {
 364                 //if there is a current name file, use it, else look for current count file
 365                                 string filename = m->getNameFile();
 366                                 if (filename != "") { hasName = true; nameFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the name parameter."); m->mothurOutEndLine(); }
 367                                 else {
 368                     filename = m->getCountTableFile();
 369                     if (filename != "") { hasCount = true; countfileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the count parameter."); m->mothurOutEndLine(); }
 370                     else { m->mothurOut("[ERROR]: You must provide a count or name file."); m->mothurOutEndLine(); abort = true;  }
 371                 }
 372             }
 373             if (!hasName && hasCount) { nameFileNames = countfileNames; }
 374
 375                         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; }
 376
 377                         bool hasGroup = true;
 378                         groupfile = validParameter.validFile(parameters, "group", false);
 379                         if (groupfile == "not found") { groupfile = "";  hasGroup = false; }
 380                         else {
 381                                 m->splitAtDash(groupfile, groupFileNames);
 382
 383                                 //go through files and make sure they are good, if not, then disregard them
 384                                 for (int i = 0; i < groupFileNames.size(); i++) {
 385
 386                                         bool ignore = false;
 387                                         if (groupFileNames[i] == "current") {
 388                                                 groupFileNames[i] = m->getGroupFile();
 389                                                 if (groupFileNames[i] != "") {  m->mothurOut("Using " + groupFileNames[i] + " as input file for the group parameter where you had given current."); m->mothurOutEndLine(); }
 390                                                 else {
 391                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 392                                                         //erase from file list
 393                                                         groupFileNames.erase(groupFileNames.begin()+i);
 394                                                         i--;
 395                                                 }
 396                                         }
 397
 398                                         if (!ignore) {
 399
 400                                                 if (inputDir != "") {
 401                                                         string path = m->hasPath(groupFileNames[i]);
 402                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 403                                                         if (path == "") {       groupFileNames[i] = inputDir + groupFileNames[i];               }
 404                                                 }
 405
 406                                                 int ableToOpen;
 407                                                 ifstream in;
 408
 409                                                 ableToOpen = m->openInputFile(groupFileNames[i], in, "noerror");
 410
 411                                                 //if you can't open it, try default location
 412                                                 if (ableToOpen == 1) {
 413                                                         if (m->getDefaultPath() != "") { //default path is set
 414                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(groupFileNames[i]);
 415                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 416                                                                 ifstream in2;
 417                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 418                                                                 in2.close();
 419                                                                 groupFileNames[i] = tryPath;
 420                                                         }
 421                                                 }
 422
 423                                                 if (ableToOpen == 1) {
 424                                                         if (m->getOutputDir() != "") { //default path is set
 425                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(groupFileNames[i]);
 426                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 427                                                                 ifstream in2;
 428                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 429                                                                 in2.close();
 430                                                                 groupFileNames[i] = tryPath;
 431                                                         }
 432                                                 }
 433
 434                                                 in.close();
 435
 436                                                 if (ableToOpen == 1) {
 437                                                         m->mothurOut("Unable to open " + groupFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 438                                                         //erase from file list
 439                                                         groupFileNames.erase(groupFileNames.begin()+i);
 440                                                         i--;
 441                                                 }else {
 442                                                         m->setGroupFile(groupFileNames[i]);
 443                                                 }
 444                                         }
 445                                 }
 446
 447                                 //make sure there is at least one valid file left
 448                                 if (groupFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid group files."); m->mothurOutEndLine(); abort = true; }
 449                         }
 450
 451                         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; }
 452
 453             if (hasGroup && hasCount) { m->mothurOut("[ERROR]: You must enter ONLY ONE of the following: count or group."); m->mothurOutEndLine(); abort = true; }
 454
 455                         //if the user changes the output directory command factory will send this info to us in the output parameter
 456                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 457
 458                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 459                         m->setProcessors(temp);
 460                         m->mothurConvert(temp, processors);
 461
 462                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.50";  }
 463                         m->mothurConvert(temp, cutoff);
 464
 465                         temp = validParameter.validFile(parameters, "alpha", false);    if (temp == "not found"){       temp = "-5.54"; }
 466                         m->mothurConvert(temp, alpha);
 467
 468                         temp = validParameter.validFile(parameters, "cutoff", false);   if (temp == "not found"){       temp = "0.33";  }
 469                         m->mothurConvert(temp, beta);
 470
 471                         temp = validParameter.validFile(parameters, "dereplicate", false);
 472                         if (temp == "not found") {
 473                                 if (groupfile != "")    {  temp = "false";                                      }
 474                                 else                    {  temp = "true";       }
 475                         }
 476                         dups = m->isTrue(temp);
 477                 }
 478         }
 479         catch(exception& e) {
 480                 m->errorOut(e, "ChimeraPerseusCommand", "ChimeraPerseusCommand");
 481                 exit(1);
 482         }
 483 }
 484 //***************************************************************************************************************
 485
 486 int ChimeraPerseusCommand::execute(){
 487         try{
 488                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 489
 490
 491                 //process each file
 492                 for (int s = 0; s < fastaFileNames.size(); s++) {
 493
 494                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 495
 496                         int start = time(NULL);
 497                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 498                         map<string, string> variables;
 499                         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]));
 500                         string outputFileName = getOutputFileName("chimera", variables);
 501                         string accnosFileName = getOutputFileName("accnos", variables);
 502             string newCountFile = "";
 503
 504                         //string newFasta = m->getRootName(fastaFileNames[s]) + "temp";
 505
 506                         //you provided a groupfile
 507                         string groupFile = "";
 508                         if (groupFileNames.size() != 0) { groupFile = groupFileNames[s]; }
 509
 510                         string nameFile = "";
 511                         if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 512                                 nameFile = nameFileNames[s];
 513                         }else { nameFile = getNamesFile(fastaFileNames[s]); }
 514
 515                         if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        } return 0;     }
 516
 517                         int numSeqs = 0;
 518                         int numChimeras = 0;
 519
 520             if (hasCount) {
 521                 CountTable* ct = new CountTable();
 522                 ct->readTable(nameFile);
 523
 524                 if (ct->hasGroupInfo()) {
 525                     cparser = new SequenceCountParser(fastaFileNames[s], *ct);
 526                     variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(nameFile));
 527                     newCountFile = getOutputFileName("count", variables);
 528
 529                     vector<string> groups = cparser->getNamesOfGroups();
 530
 531                     if (m->control_pressed) { delete ct; delete cparser; for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        }  return 0; }
 532
 533                     //clears files
 534                     ofstream out, out1, out2;
 535                     m->openOutputFile(outputFileName, out); out.close();
 536                     m->openOutputFile(accnosFileName, out1); out1.close();
 537
 538                     if(processors == 1) {       numSeqs = driverGroups(outputFileName, accnosFileName, newCountFile, 0, groups.size(), groups);
 539                         if (dups) {
 540                             CountTable c; c.readTable(nameFile);
 541                             if (!m->isBlank(newCountFile)) {
 542                                 ifstream in2;
 543                                 m->openInputFile(newCountFile, in2);
 544
 545                                 string name, group;
 546                                 while (!in2.eof()) {
 547                                     in2 >> name >> group; m->gobble(in2);
 548                                     c.setAbund(name, group, 0);
 549                                 }
 550                                 in2.close();
 551                             }
 552                             m->mothurRemove(newCountFile);
 553                             c.printTable(newCountFile);
 554                         }
 555
 556                     }
 557                     else                                {       numSeqs = createProcessesGroups(outputFileName, accnosFileName, newCountFile, groups, groupFile, fastaFileNames[s], nameFile);                  }
 558
 559                     if (m->control_pressed) {  delete ct; delete cparser; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;    }
 560                     map<string, string> uniqueNames = cparser->getAllSeqsMap();
 561                     if (!dups) {
 562                         numChimeras = deconvoluteResults(uniqueNames, outputFileName, accnosFileName);
 563                     }else {
 564                         set<string> doNotRemove;
 565                         CountTable c; c.readTable(newCountFile);
 566                         vector<string> namesInTable = c.getNamesOfSeqs();
 567                         for (int i = 0; i < namesInTable.size(); i++) {
 568                             int temp = c.getNumSeqs(namesInTable[i]);
 569                             if (temp == 0) {  c.remove(namesInTable[i]);  }
 570                             else { doNotRemove.insert((namesInTable[i])); }
 571                         }
 572                         //remove names we want to keep from accnos file.
 573                         set<string> accnosNames = m->readAccnos(accnosFileName);
 574                         ofstream out2;
 575                         m->openOutputFile(accnosFileName, out2);
 576                         for (set<string>::iterator it = accnosNames.begin(); it != accnosNames.end(); it++) {
 577                             if (doNotRemove.count(*it) == 0) {  out2 << (*it) << endl; }
 578                         }
 579                         out2.close();
 580                         c.printTable(newCountFile);
 581                         outputNames.push_back(newCountFile); outputTypes["count"].push_back(newCountFile);
 582
 583                     }
 584                     delete cparser;
 585
 586                     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();
 587
 588                     if (m->control_pressed) {  delete ct; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;  }
 589
 590                 }else {
 591                     if (processors != 1) { m->mothurOut("Your count file does not contain group information, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 592
 593                     //read sequences and store sorted by frequency
 594                     vector<seqData> sequences = readFiles(fastaFileNames[s], ct);
 595
 596                     if (m->control_pressed) { delete ct; for (int j = 0; j < outputNames.size(); j++) { m->mothurRemove(outputNames[j]);        } return 0; }
 597
 598                     numSeqs = driver(outputFileName, sequences, accnosFileName, numChimeras);
 599                 }
 600                 delete ct;
 601             }else {
 602                 if (groupFile != "") {
 603                     //Parse sequences by group
 604                     parser = new SequenceParser(groupFile, fastaFileNames[s], nameFile);
 605                     vector<string> groups = parser->getNamesOfGroups();
 606
 607                     if (m->control_pressed) { delete parser; for (int j = 0; j < outputNames.size(); j++) {     m->mothurRemove(outputNames[j]);        }  return 0; }
 608
 609                     //clears files
 610                     ofstream out, out1, out2;
 611                     m->openOutputFile(outputFileName, out); out.close();
 612                     m->openOutputFile(accnosFileName, out1); out1.close();
 613
 614                     if(processors == 1) {       numSeqs = driverGroups(outputFileName, accnosFileName, "", 0, groups.size(), groups);   }
 615                     else                                {       numSeqs = createProcessesGroups(outputFileName, accnosFileName, "", groups, groupFile, fastaFileNames[s], nameFile);                    }
 616
 617                     if (m->control_pressed) {  delete parser; for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        }  return 0;    }
 618                     map<string, string> uniqueNames = parser->getAllSeqsMap();
 619                     if (!dups) {
 620                         numChimeras = deconvoluteResults(uniqueNames, outputFileName, accnosFileName);
 621                     }
 622                     delete parser;
 623
 624                     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();
 625
 626                     if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  return 0;  }
 627                 }else{
 628                     if (processors != 1) { m->mothurOut("Without a groupfile, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 629
 630                     //read sequences and store sorted by frequency
 631                     vector<seqData> sequences = readFiles(fastaFileNames[s], nameFile);
 632
 633                     if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {    m->mothurRemove(outputNames[j]);        } return 0; }
 634
 635                     numSeqs = driver(outputFileName, sequences, accnosFileName, numChimeras);
 636                 }
 637                         }
 638
 639                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        } return 0; }
 640
 641                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences. " + toString(numChimeras) + " chimeras were found.");      m->mothurOutEndLine();
 642                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 643                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 644                 }
 645
 646                 //set accnos file as new current accnosfile
 647                 string current = "";
 648                 itTypes = outputTypes.find("accnos");
 649                 if (itTypes != outputTypes.end()) {
 650                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 651                 }
 652
 653                 m->mothurOutEndLine();
 654                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 655                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 656                 m->mothurOutEndLine();
 657
 658                 return 0;
 659
 660         }
 661         catch(exception& e) {
 662                 m->errorOut(e, "ChimeraPerseusCommand", "execute");
 663                 exit(1);
 664         }
 665 }
 666 //**********************************************************************************************************************
 667 string ChimeraPerseusCommand::getNamesFile(string& inputFile){
 668         try {
 669                 string nameFile = "";
 670
 671                 m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 672
 673                 //use unique.seqs to create new name and fastafile
 674                 string inputString = "fasta=" + inputFile;
 675                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 676                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 677                 m->mothurCalling = true;
 678
 679                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
 680                 uniqueCommand->execute();
 681
 682                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 683
 684                 delete uniqueCommand;
 685                 m->mothurCalling = false;
 686                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 687
 688                 nameFile = filenames["name"][0];
 689                 inputFile = filenames["fasta"][0];
 690
 691                 return nameFile;
 692         }
 693         catch(exception& e) {
 694                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 695                 exit(1);
 696         }
 697 }
 698 //**********************************************************************************************************************
 699 int ChimeraPerseusCommand::driverGroups(string outputFName, string accnos, string countlist, int start, int end, vector<string> groups){
 700         try {
 701
 702                 int totalSeqs = 0;
 703                 int numChimeras = 0;
 704
 705         ofstream outCountList;
 706         if (hasCount && dups) { m->openOutputFile(countlist, outCountList); }
 707
 708                 for (int i = start; i < end; i++) {
 709
 710                         m->mothurOutEndLine(); m->mothurOut("Checking sequences from group " + groups[i] + "...");      m->mothurOutEndLine();
 711
 712                         int start = time(NULL);  if (m->control_pressed) {  return 0; }
 713
 714                         vector<seqData> sequences = loadSequences(groups[i]);
 715
 716                         if (m->control_pressed) { return 0; }
 717
 718                         int numSeqs = driver((outputFName + groups[i]), sequences, (accnos+groups[i]), numChimeras);
 719                         totalSeqs += numSeqs;
 720
 721                         if (m->control_pressed) { return 0; }
 722
 723             if (dups) {
 724                 if (!m->isBlank(accnos+groups[i])) {
 725                     ifstream in;
 726                     m->openInputFile(accnos+groups[i], in);
 727                     string name;
 728                     if (hasCount) {
 729                         while (!in.eof()) {
 730                             in >> name; m->gobble(in);
 731                             outCountList << name << '\t' << groups[i] << endl;
 732                         }
 733                         in.close();
 734                     }else {
 735                         map<string, string> thisnamemap = parser->getNameMap(groups[i]);
 736                         map<string, string>::iterator itN;
 737                         ofstream out;
 738                         m->openOutputFile(accnos+groups[i]+".temp", out);
 739                         while (!in.eof()) {
 740                             in >> name; m->gobble(in);
 741                             itN = thisnamemap.find(name);
 742                             if (itN != thisnamemap.end()) {
 743                                 vector<string> tempNames; m->splitAtComma(itN->second, tempNames);
 744                                 for (int j = 0; j < tempNames.size(); j++) { out << tempNames[j] << endl; }
 745
 746                             }else { m->mothurOut("[ERROR]: parsing cannot find " + name + ".\n"); m->control_pressed = true; }
 747                         }
 748                         out.close();
 749                         in.close();
 750                         m->renameFile(accnos+groups[i]+".temp", accnos+groups[i]);
 751                     }
 752
 753                 }
 754             }
 755
 756                         //append files
 757                         m->appendFiles((outputFName+groups[i]), outputFName); m->mothurRemove((outputFName+groups[i]));
 758                         m->appendFiles((accnos+groups[i]), accnos); m->mothurRemove((accnos+groups[i]));
 759
 760                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences from group " + groups[i] + ".");    m->mothurOutEndLine();
 761                 }
 762
 763         if (hasCount && dups) { outCountList.close(); }
 764
 765                 return totalSeqs;
 766
 767         }
 768         catch(exception& e) {
 769                 m->errorOut(e, "ChimeraPerseusCommand", "driverGroups");
 770                 exit(1);
 771         }
 772 }
 773 //**********************************************************************************************************************
 774 vector<seqData> ChimeraPerseusCommand::loadSequences(string group){
 775         try {
 776         bool error = false;
 777                 alignLength = 0;
 778         vector<seqData> sequences;
 779         if (hasCount) {
 780             vector<Sequence> thisGroupsSeqs = cparser->getSeqs(group);
 781             map<string, int> counts = cparser->getCountTable(group);
 782             map<string, int>::iterator it;
 783
 784             for (int i = 0; i < thisGroupsSeqs.size(); i++) {
 785
 786                 if (m->control_pressed) {  return sequences; }
 787
 788                 it = counts.find(thisGroupsSeqs[i].getName());
 789                 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(); }
 790                 else {
 791                     thisGroupsSeqs[i].setAligned(removeNs(thisGroupsSeqs[i].getUnaligned()));
 792                     sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), it->second));
 793                     if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
 794                 }
 795             }
 796         }else{
 797             vector<Sequence> thisGroupsSeqs = parser->getSeqs(group);
 798             map<string, string> nameMap = parser->getNameMap(group);
 799             map<string, string>::iterator it;
 800
 801             for (int i = 0; i < thisGroupsSeqs.size(); i++) {
 802
 803                 if (m->control_pressed) {  return sequences; }
 804
 805                 it = nameMap.find(thisGroupsSeqs[i].getName());
 806                 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(); }
 807                 else {
 808                     int num = m->getNumNames(it->second);
 809                     thisGroupsSeqs[i].setAligned(removeNs(thisGroupsSeqs[i].getUnaligned()));
 810                     sequences.push_back(seqData(thisGroupsSeqs[i].getName(), thisGroupsSeqs[i].getUnaligned(), num));
 811                     if (thisGroupsSeqs[i].getUnaligned().length() > alignLength) { alignLength = thisGroupsSeqs[i].getUnaligned().length(); }
 812                 }
 813             }
 814
 815                 }
 816
 817         if (error) { m->control_pressed = true; }
 818                 //sort by frequency
 819                 sort(sequences.rbegin(), sequences.rend());
 820
 821                 return sequences;
 822         }
 823         catch(exception& e) {
 824                 m->errorOut(e, "ChimeraPerseusCommand", "loadSequences");
 825                 exit(1);
 826         }
 827 }
 828
 829 //**********************************************************************************************************************
 830 vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, string name){
 831         try {
 832                 map<string, int>::iterator it;
 833                 map<string, int> nameMap = m->readNames(name);
 834
 835                 //read fasta file and create sequenceData structure - checking for file mismatches
 836                 vector<seqData> sequences;
 837                 bool error = false;
 838                 ifstream in;
 839                 m->openInputFile(inputFile, in);
 840                 alignLength = 0;
 841
 842                 while (!in.eof()) {
 843
 844                         if (m->control_pressed) { in.close(); return sequences; }
 845
 846                         Sequence temp(in); m->gobble(in);
 847
 848                         it = nameMap.find(temp.getName());
 849                         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(); }
 850                         else {
 851                 temp.setAligned(removeNs(temp.getUnaligned()));
 852                                 sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), it->second));
 853                 if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
 854                         }
 855                 }
 856                 in.close();
 857
 858                 if (error) { m->control_pressed = true; }
 859
 860                 //sort by frequency
 861                 sort(sequences.rbegin(), sequences.rend());
 862
 863                 return sequences;
 864         }
 865         catch(exception& e) {
 866                 m->errorOut(e, "ChimeraPerseusCommand", "readFiles");
 867                 exit(1);
 868         }
 869 }
 870 //**********************************************************************************************************************
 871 string ChimeraPerseusCommand::removeNs(string seq){
 872         try {
 873         string newSeq = "";
 874         for (int i = 0; i < seq.length(); i++) {
 875             if (seq[i] != 'N') {  newSeq += seq[i]; }
 876         }
 877         return newSeq;
 878     }
 879         catch(exception& e) {
 880                 m->errorOut(e, "ChimeraPerseusCommand", "removeNs");
 881                 exit(1);
 882         }
 883 }
 884 //**********************************************************************************************************************
 885 vector<seqData> ChimeraPerseusCommand::readFiles(string inputFile, CountTable* ct){
 886         try {
 887                 //read fasta file and create sequenceData structure - checking for file mismatches
 888                 vector<seqData> sequences;
 889                 ifstream in;
 890                 m->openInputFile(inputFile, in);
 891                 alignLength = 0;
 892
 893                 while (!in.eof()) {
 894             Sequence temp(in); m->gobble(in);
 895
 896                         int count = ct->getNumSeqs(temp.getName());
 897                         if (m->control_pressed) { break; }
 898                         else {
 899                 temp.setAligned(removeNs(temp.getUnaligned()));
 900                                 sequences.push_back(seqData(temp.getName(), temp.getUnaligned(), count));
 901                 if (temp.getUnaligned().length() > alignLength) { alignLength = temp.getUnaligned().length(); }
 902                         }
 903                 }
 904                 in.close();
 905
 906                 //sort by frequency
 907                 sort(sequences.rbegin(), sequences.rend());
 908
 909                 return sequences;
 910         }
 911         catch(exception& e) {
 912                 m->errorOut(e, "ChimeraPerseusCommand", "getNamesFile");
 913                 exit(1);
 914         }
 915 }
 916 //**********************************************************************************************************************
 917 int ChimeraPerseusCommand::driver(string chimeraFileName, vector<seqData>& sequences, string accnosFileName, int& numChimeras){
 918         try {
 919
 920                 vector<vector<double> > correctModel(4);        //could be an option in the future to input own model matrix
 921                 for(int i=0;i<4;i++){   correctModel[i].resize(4);      }
 922
 923                 correctModel[0][0] = 0.000000;  //AA
 924                 correctModel[1][0] = 11.619259; //CA
 925                 correctModel[2][0] = 11.694004; //TA
 926                 correctModel[3][0] = 7.748623;  //GA
 927
 928                 correctModel[1][1] = 0.000000;  //CC
 929                 correctModel[2][1] = 7.619657;  //TC
 930                 correctModel[3][1] = 12.852562; //GC
 931
 932                 correctModel[2][2] = 0.000000;  //TT
 933                 correctModel[3][2] = 10.964048; //TG
 934
 935                 correctModel[3][3] = 0.000000;  //GG
 936
 937                 for(int i=0;i<4;i++){
 938                         for(int j=0;j<i;j++){
 939                                 correctModel[j][i] = correctModel[i][j];
 940                         }
 941                 }
 942
 943                 int numSeqs = sequences.size();
 944                 //int alignLength = sequences[0].sequence.size();
 945
 946                 ofstream chimeraFile;
 947                 ofstream accnosFile;
 948                 m->openOutputFile(chimeraFileName, chimeraFile);
 949                 m->openOutputFile(accnosFileName, accnosFile);
 950
 951                 Perseus myPerseus;
 952                 vector<vector<double> > binMatrix = myPerseus.binomial(alignLength);
 953
 954                 chimeraFile << "SequenceIndex\tName\tDiffsToBestMatch\tBestMatchIndex\tBestMatchName\tDiffstToChimera\tIndexofLeftParent\tIndexOfRightParent\tNameOfLeftParent\tNameOfRightParent\tDistanceToBestMatch\tcIndex\t(cIndex - singleDist)\tloonIndex\tMismatchesToChimera\tMismatchToTrimera\tChimeraBreakPoint\tLogisticProbability\tTypeOfSequence\n";
 955
 956                 vector<bool> chimeras(numSeqs, 0);
 957
 958                 for(int i=0;i<numSeqs;i++){
 959                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 960
 961                         vector<bool> restricted = chimeras;
 962
 963                         vector<vector<int> > leftDiffs(numSeqs);
 964                         vector<vector<int> > leftMaps(numSeqs);
 965                         vector<vector<int> > rightDiffs(numSeqs);
 966                         vector<vector<int> > rightMaps(numSeqs);
 967
 968                         vector<int> singleLeft, bestLeft;
 969                         vector<int> singleRight, bestRight;
 970
 971                         int bestSingleIndex, bestSingleDiff;
 972                         vector<pwAlign> alignments(numSeqs);
 973
 974                         int comparisons = myPerseus.getAlignments(i, sequences, alignments, leftDiffs, leftMaps, rightDiffs, rightMaps, bestSingleIndex, bestSingleDiff, restricted);
 975                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 976
 977                         int minMismatchToChimera, leftParentBi, rightParentBi, breakPointBi;
 978
 979                         string dummyA, dummyB;
 980
 981             if (sequences[i].sequence.size() < 3) {
 982                 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;
 983             }else if(comparisons >= 2){
 984                                 minMismatchToChimera = myPerseus.getChimera(sequences, leftDiffs, rightDiffs, leftParentBi, rightParentBi, breakPointBi, singleLeft, bestLeft, singleRight, bestRight, restricted);
 985                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 986
 987                                 int minMismatchToTrimera = numeric_limits<int>::max();
 988                                 int leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB;
 989
 990                                 if(minMismatchToChimera >= 3 && comparisons >= 3){
 991                                         minMismatchToTrimera = myPerseus.getTrimera(sequences, leftDiffs, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, singleLeft, bestLeft, singleRight, bestRight, restricted);
 992                                         if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 993                                 }
 994
 995                                 double singleDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, sequences[bestSingleIndex].sequence, dummyA, dummyB, correctModel);
 996
 997                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
 998
 999                                 string type;
1000                                 string chimeraRefSeq;
1001
1002                                 if(minMismatchToChimera - minMismatchToTrimera >= 3){
1003                                         type = "trimera";
1004                                         chimeraRefSeq = myPerseus.stitchTrimera(alignments, leftParentTri, middleParentTri, rightParentTri, breakPointTriA, breakPointTriB, leftMaps, rightMaps);
1005                                 }
1006                                 else{
1007                                         type = "chimera";
1008                                         chimeraRefSeq = myPerseus.stitchBimera(alignments, leftParentBi, rightParentBi, breakPointBi, leftMaps, rightMaps);
1009                                 }
1010
1011                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
1012
1013                                 double chimeraDist = myPerseus.modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq, dummyA, dummyB, correctModel);
1014
1015                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
1016
1017                                 double cIndex = chimeraDist;//modeledPairwiseAlignSeqs(sequences[i].sequence, chimeraRefSeq);
1018                                 double loonIndex = myPerseus.calcLoonIndex(sequences[i].sequence, sequences[leftParentBi].sequence, sequences[rightParentBi].sequence, breakPointBi, binMatrix);
1019
1020                                 if (m->control_pressed) { chimeraFile.close(); m->mothurRemove(chimeraFileName); accnosFile.close(); m->mothurRemove(accnosFileName); return 0; }
1021
1022                                 chimeraFile << i << '\t' << sequences[i].seqName << '\t' << bestSingleDiff << '\t' << bestSingleIndex << '\t' << sequences[bestSingleIndex].seqName << '\t';
1023                                 chimeraFile << minMismatchToChimera << '\t' << leftParentBi << '\t' << rightParentBi << '\t' << sequences[leftParentBi].seqName << '\t' << sequences[rightParentBi].seqName << '\t';
1024                                 chimeraFile << singleDist << '\t' << cIndex << '\t' << (cIndex - singleDist) << '\t' << loonIndex << '\t';
1025                                 chimeraFile << minMismatchToChimera << '\t' << minMismatchToTrimera << '\t' << breakPointBi << '\t';
1026
1027                                 double probability = myPerseus.classifyChimera(singleDist, cIndex, loonIndex, alpha, beta);
1028
1029                                 chimeraFile << probability << '\t';
1030
1031                                 if(probability > cutoff){
1032                                         chimeraFile << type << endl;
1033                                         accnosFile << sequences[i].seqName << endl;
1034                                         chimeras[i] = 1;
1035                                         numChimeras++;
1036                                 }
1037                                 else{
1038                                         chimeraFile << "good" << endl;
1039                                 }
1040
1041                         }
1042                         else{
1043                                 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;
1044                         }
1045
1046                         //report progress
1047                         if((i+1) % 100 == 0){   m->mothurOut("Processing sequence: " + toString(i+1) + "\n");           }
1048                 }
1049
1050                 if((numSeqs) % 100 != 0){       m->mothurOut("Processing sequence: " + toString(numSeqs) + "\n");               }
1051
1052                 chimeraFile.close();
1053                 accnosFile.close();
1054
1055                 return numSeqs;
1056         }
1057         catch(exception& e) {
1058                 m->errorOut(e, "ChimeraPerseusCommand", "driver");
1059                 exit(1);
1060         }
1061 }
1062 /**************************************************************************************************/
1063 int ChimeraPerseusCommand::createProcessesGroups(string outputFName, string accnos, string newCountFile, vector<string> groups, string group, string fasta, string name) {
1064         try {
1065
1066                 vector<int> processIDS;
1067                 int process = 1;
1068                 int num = 0;
1069
1070         CountTable newCount;
1071         if (hasCount && dups) { newCount.readTable(name); }
1072
1073                 //sanity check
1074                 if (groups.size() < processors) { processors = groups.size(); }
1075
1076                 //divide the groups between the processors
1077                 vector<linePair> lines;
1078                 int numGroupsPerProcessor = groups.size() / processors;
1079                 for (int i = 0; i < processors; i++) {
1080                         int startIndex =  i * numGroupsPerProcessor;
1081                         int endIndex = (i+1) * numGroupsPerProcessor;
1082                         if(i == (processors - 1)){      endIndex = groups.size();       }
1083                         lines.push_back(linePair(startIndex, endIndex));
1084                 }
1085
1086 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
1087
1088                 //loop through and create all the processes you want
1089                 while (process != processors) {
1090                         int pid = fork();
1091
1092                         if (pid > 0) {
1093                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1094                                 process++;
1095                         }else if (pid == 0){
1096                                 num = driverGroups(outputFName + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", accnos + ".byCount." + toString(getpid()) + ".temp", lines[process].start, lines[process].end, groups);
1097
1098                                 //pass numSeqs to parent
1099                                 ofstream out;
1100                                 string tempFile = outputFName + toString(getpid()) + ".num.temp";
1101                                 m->openOutputFile(tempFile, out);
1102                                 out << num << endl;
1103                                 out.close();
1104
1105                                 exit(0);
1106                         }else {
1107                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1108                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1109                                 exit(0);
1110                         }
1111                 }
1112
1113                 //do my part
1114                 num = driverGroups(outputFName, accnos, accnos + ".byCount", lines[0].start, lines[0].end, groups);
1115
1116                 //force parent to wait until all the processes are done
1117                 for (int i=0;i<processIDS.size();i++) {
1118                         int temp = processIDS[i];
1119                         wait(&temp);
1120                 }
1121
1122                 for (int i = 0; i < processIDS.size(); i++) {
1123                         ifstream in;
1124                         string tempFile =  outputFName + toString(processIDS[i]) + ".num.temp";
1125                         m->openInputFile(tempFile, in);
1126                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
1127                         in.close(); m->mothurRemove(tempFile);
1128                 }
1129
1130 #else
1131                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1132                 //Windows version shared memory, so be careful when passing variables through the preClusterData struct.
1133                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1134                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1135
1136                 vector<perseusData*> pDataArray;
1137                 DWORD   dwThreadIdArray[processors-1];
1138                 HANDLE  hThreadArray[processors-1];
1139
1140                 //Create processor worker threads.
1141                 for( int i=1; i<processors; i++ ){
1142                         // Allocate memory for thread data.
1143                         string extension = toString(i) + ".temp";
1144
1145                         perseusData* tempPerseus = new perseusData(dups, hasName, hasCount, alpha, beta, cutoff, outputFName+extension, fasta, name, group, accnos+extension,  accnos+".byCount."+extension, groups, m, lines[i].start, lines[i].end, i);
1146
1147                         pDataArray.push_back(tempPerseus);
1148                         processIDS.push_back(i);
1149
1150                         //MyPerseusThreadFunction is in header. It must be global or static to work with the threads.
1151                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1152                         hThreadArray[i-1] = CreateThread(NULL, 0, MyPerseusThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
1153                 }
1154
1155
1156                 //using the main process as a worker saves time and memory
1157                 num = driverGroups(outputFName, accnos, accnos + ".byCount", lines[0].start, lines[0].end, groups);
1158
1159                 //Wait until all threads have terminated.
1160                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1161
1162                 //Close all thread handles and free memory allocations.
1163                 for(int i=0; i < pDataArray.size(); i++){
1164                         num += pDataArray[i]->count;
1165                         CloseHandle(hThreadArray[i]);
1166                         delete pDataArray[i];
1167                 }
1168 #endif
1169                 //read my own
1170         if (hasCount && dups) {
1171             if (!m->isBlank(accnos + ".byCount")) {
1172                 ifstream in2;
1173                 m->openInputFile(accnos + ".byCount", in2);
1174
1175                 string name, group;
1176                 while (!in2.eof()) {
1177                     in2 >> name >> group; m->gobble(in2);
1178                     newCount.setAbund(name, group, 0);
1179                 }
1180                 in2.close();
1181             }
1182             m->mothurRemove(accnos + ".byCount");
1183         }
1184
1185
1186                 //append output files
1187                 for(int i=0;i<processIDS.size();i++){
1188                         m->appendFiles((outputFName + toString(processIDS[i]) + ".temp"), outputFName);
1189                         m->mothurRemove((outputFName + toString(processIDS[i]) + ".temp"));
1190
1191                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
1192                         m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));
1193
1194             if (hasCount && dups) {
1195                 if (!m->isBlank(accnos + ".byCount." + toString(processIDS[i]) + ".temp")) {
1196                     ifstream in2;
1197                     m->openInputFile(accnos + ".byCount." + toString(processIDS[i]) + ".temp", in2);
1198
1199                     string name, group;
1200                     while (!in2.eof()) {
1201                         in2 >> name >> group; m->gobble(in2);
1202                         newCount.setAbund(name, group, 0);
1203                     }
1204                     in2.close();
1205                 }
1206                 m->mothurRemove(accnos + ".byCount." + toString(processIDS[i]) + ".temp");
1207             }
1208
1209                 }
1210
1211         //print new *.pick.count_table
1212         if (hasCount && dups) {  newCount.printTable(newCountFile);   }
1213
1214                 return num;
1215
1216         }
1217         catch(exception& e) {
1218                 m->errorOut(e, "ChimeraPerseusCommand", "createProcessesGroups");
1219                 exit(1);
1220         }
1221 }
1222 //**********************************************************************************************************************
1223 int ChimeraPerseusCommand::deconvoluteResults(map<string, string>& uniqueNames, string outputFileName, string accnosFileName){
1224         try {
1225                 map<string, string>::iterator itUnique;
1226                 int total = 0;
1227
1228                 //edit accnos file
1229                 ifstream in2;
1230                 m->openInputFile(accnosFileName, in2);
1231
1232                 ofstream out2;
1233                 m->openOutputFile(accnosFileName+".temp", out2);
1234
1235                 string name;
1236                 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
1237                 set<string>::iterator itNames;
1238                 set<string> chimerasInFile;
1239                 set<string>::iterator itChimeras;
1240
1241
1242                 while (!in2.eof()) {
1243                         if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(outputFileName); m->mothurRemove((accnosFileName+".temp")); return 0; }
1244
1245                         in2 >> name; m->gobble(in2);
1246
1247                         //find unique name
1248                         itUnique = uniqueNames.find(name);
1249
1250                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing accnos results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1251                         else {
1252                                 itChimeras = chimerasInFile.find((itUnique->second));
1253
1254                                 if (itChimeras == chimerasInFile.end()) {
1255                                         out2 << itUnique->second << endl;
1256                                         chimerasInFile.insert((itUnique->second));
1257                                         total++;
1258                                 }
1259                         }
1260                 }
1261                 in2.close();
1262                 out2.close();
1263
1264                 m->mothurRemove(accnosFileName);
1265                 rename((accnosFileName+".temp").c_str(), accnosFileName.c_str());
1266
1267                 //edit chimera file
1268                 ifstream in;
1269                 m->openInputFile(outputFileName, in);
1270
1271                 ofstream out;
1272                 m->openOutputFile(outputFileName+".temp", out); out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
1273
1274                 int DiffsToBestMatch, BestMatchIndex, DiffstToChimera, IndexofLeftParent, IndexOfRightParent;
1275                 float temp1,temp2, temp3, temp4, temp5, temp6, temp7, temp8;
1276                 string index, BestMatchName, parent1, parent2, flag;
1277                 name = "";
1278                 namesInFile.clear();
1279                 //assumptions - in file each read will always look like
1280                 /*
1281                  SequenceIndex  Name    DiffsToBestMatch        BestMatchIndex  BestMatchName   DiffstToChimera IndexofLeftParent       IndexOfRightParent      NameOfLeftParent        NameOfRightParent       DistanceToBestMatch     cIndex  (cIndex - singleDist)   loonIndex       MismatchesToChimera     MismatchToTrimera       ChimeraBreakPoint       LogisticProbability     TypeOfSequence
1282                  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
1283                  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
1284                  2      F01QG4L02JZOEC  48      0       F01QG4L02JVBQY  47      0       0       F01QG4L02JVBQY  F01QG4L02JVBQY  2.0449  2.03545 -0.00944493     0       47      2147483647      138     0       good
1285                  3      F01QG4L02G7JEC  42      0       F01QG4L02JVBQY  40      1       0       F01QG4L02ICTC6  F01QG4L02JVBQY  1.87477 1.81113 -0.0636404      5.80145 40      2147483647      25      0       good
1286                  */
1287
1288                 //get and print headers
1289                 BestMatchName = m->getline(in); m->gobble(in);
1290                 out << BestMatchName << endl;
1291
1292                 while (!in.eof()) {
1293
1294                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove((outputFileName+".temp")); return 0; }
1295
1296                         bool print = false;
1297                         in >> index;    m->gobble(in);
1298
1299                         if (index != "SequenceIndex") { //if you are not a header line, there will be a header line for each group if group file is given
1300                                 in >> name;             m->gobble(in);
1301                                 in >> DiffsToBestMatch; m->gobble(in);
1302                                 in >> BestMatchIndex; m->gobble(in);
1303                                 in >> BestMatchName; m->gobble(in);
1304                                 in >> DiffstToChimera; m->gobble(in);
1305                                 in >> IndexofLeftParent; m->gobble(in);
1306                                 in >> IndexOfRightParent; m->gobble(in);
1307                                 in >> parent1;  m->gobble(in);
1308                                 in >> parent2;  m->gobble(in);
1309                                 in >> temp1 >> temp2 >> temp3 >> temp4 >> temp5 >> temp6 >> temp7 >> temp8 >> flag; m->gobble(in);
1310
1311                                 //find unique name
1312                                 itUnique = uniqueNames.find(name);
1313
1314                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1315                                 else {
1316                                         name = itUnique->second;
1317                                         //is this name already in the file
1318                                         itNames = namesInFile.find((name));
1319
1320                                         if (itNames == namesInFile.end()) { //no not in file
1321                                                 if (flag == "good") { //are you really a no??
1322                                                         //is this sequence really not chimeric??
1323                                                         itChimeras = chimerasInFile.find(name);
1324
1325                                                         //then you really are a no so print, otherwise skip
1326                                                         if (itChimeras == chimerasInFile.end()) { print = true; }
1327                                                 }else{ print = true; }
1328                                         }
1329                                 }
1330
1331                                 if (print) {
1332                                         out << index << '\t' << name  << '\t' << DiffsToBestMatch << '\t' << BestMatchIndex << '\t';
1333                                         namesInFile.insert(name);
1334
1335                                         if (BestMatchName != "Null") {
1336                                                 itUnique = uniqueNames.find(BestMatchName);
1337                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find BestMatchName "+ BestMatchName + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1338                                                 else {  out << itUnique->second << '\t';        }
1339                                         }else { out << "Null" << '\t'; }
1340
1341                                         out << DiffstToChimera << '\t' << IndexofLeftParent << '\t' << IndexOfRightParent << '\t';
1342
1343                                         if (parent1 != "Null") {
1344                                                 itUnique = uniqueNames.find(parent1);
1345                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent1 "+ parent1 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1346                                                 else {  out << itUnique->second << '\t';        }
1347                                         }else { out << "Null" << '\t'; }
1348
1349                                         if (parent1 != "Null") {
1350                                                 itUnique = uniqueNames.find(parent2);
1351                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parent2 "+ parent2 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1352                                                 else {  out << itUnique->second << '\t';        }
1353                                         }else { out << "Null" << '\t'; }
1354
1355                                         out << temp1 << '\t' << temp2 << '\t' << temp3 << '\t' << temp4 << '\t' << temp5 << '\t' << temp6 << '\t' << temp7 << '\t' << temp8 << '\t' << flag << endl;
1356                                 }
1357                         }else { index = m->getline(in); m->gobble(in); }
1358                 }
1359                 in.close();
1360                 out.close();
1361
1362                 m->mothurRemove(outputFileName);
1363                 rename((outputFileName+".temp").c_str(), outputFileName.c_str());
1364
1365                 return total;
1366         }
1367         catch(exception& e) {
1368                 m->errorOut(e, "ChimeraPerseusCommand", "deconvoluteResults");
1369                 exit(1);
1370         }
1371 }
1372 //**********************************************************************************************************************
1373
1374