chimeraslayercommand.cpp

   1 /*
   2  *  chimeraslayercommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 3/31/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimeraslayercommand.h"
  11 #include "chimeraslayer.h"
  12 #include "deconvolutecommand.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> ChimeraSlayerCommand::setParameters(){
  16         try {
  17                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  19                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  20                 CommandParameter pwindow("window", "Number", "", "50", "", "", "",false,false); parameters.push_back(pwindow);
  21                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  22                 CommandParameter pmatch("match", "Number", "", "5.0", "", "", "",false,false); parameters.push_back(pmatch);
  23                 CommandParameter pmismatch("mismatch", "Number", "", "-4.0", "", "", "",false,false); parameters.push_back(pmismatch);
  24                 CommandParameter pminsim("minsim", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminsim);
  25                 CommandParameter pmincov("mincov", "Number", "", "70", "", "", "",false,false); parameters.push_back(pmincov);
  26                 CommandParameter pminsnp("minsnp", "Number", "", "10", "", "", "",false,false); parameters.push_back(pminsnp);
  27                 CommandParameter pminbs("minbs", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminbs);
  28                 CommandParameter psearch("search", "Multiple", "kmer-blast", "blast", "", "", "",false,false); parameters.push_back(psearch);
  29                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  30                 CommandParameter prealign("realign", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(prealign);
  31                 CommandParameter ptrim("trim", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ptrim);
  32                 CommandParameter psplit("split", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psplit);
  33                 CommandParameter pnumwanted("numwanted", "Number", "", "15", "", "", "",false,false); parameters.push_back(pnumwanted);
  34                 CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  35                 CommandParameter pdivergence("divergence", "Number", "", "1.007", "", "", "",false,false); parameters.push_back(pdivergence);
  36                 CommandParameter pparents("parents", "Number", "", "3", "", "", "",false,false); parameters.push_back(pparents);
  37                 CommandParameter pincrement("increment", "Number", "", "5", "", "", "",false,false); parameters.push_back(pincrement);
  38                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  39                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  40
  41                 vector<string> myArray;
  42                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  43                 return myArray;
  44         }
  45         catch(exception& e) {
  46                 m->errorOut(e, "ChimeraSlayerCommand", "setParameters");
  47                 exit(1);
  48         }
  49 }
  50 //**********************************************************************************************************************
  51 string ChimeraSlayerCommand::getHelpString(){
  52         try {
  53                 string helpString = "";
  54                 helpString += "The chimera.slayer command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  55                 helpString += "This command was modeled after the chimeraSlayer written by the Broad Institute.\n";
  56                 helpString += "The chimera.slayer command parameters are fasta, name, template, processors, trim, ksize, window, match, mismatch, divergence. minsim, mincov, minbs, minsnp, parents, search, iters, increment, numwanted, and realign.\n";
  57                 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";
  58                 helpString += "The name parameter allows you to provide a name file, if you are using template=self. \n";
  59                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  60                 helpString += "The reference parameter allows you to enter a reference file containing known non-chimeric sequences, and is required. You may also set template=self, in this case the abundant sequences will be used as potential parents. \n";
  61                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  62 #ifdef USE_MPI
  63                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  64 #endif
  65                 helpString += "The trim parameter allows you to output a new fasta file containing your sequences with the chimeric ones trimmed to include only their longest piece, default=F. \n";
  66                 helpString += "The split parameter allows you to check both pieces of non-chimeric sequence for chimeras, thus looking for trimeras and quadmeras. default=F. \n";
  67                 helpString += "The window parameter allows you to specify the window size for searching for chimeras, default=50. \n";
  68                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n";
  69                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n";
  70                 helpString += "The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n";
  71                 helpString += "The match parameter allows you to reward matched bases in blast search, default is 5. \n";
  72                 helpString += "The parents parameter allows you to select the number of potential parents to investigate from the numwanted best matches after rating them, default is 3. \n";
  73                 helpString += "The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n";
  74                 helpString += "The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n";
  75                 helpString += "The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=1000.\n";
  76                 helpString += "The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n";
  77                 helpString += "The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n";
  78                 helpString += "The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n";
  79                 helpString += "The minsnp parameter allows you to specify percent of SNPs to sample on each side of breakpoint for computing bootstrap support (default: 10) \n";
  80                 helpString += "The search parameter allows you to specify search method for finding the closest parent. Choices are blast, and kmer, default blast. \n";
  81                 helpString += "The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default true.  \n";
  82                 helpString += "The chimera.slayer command should be in the following format: \n";
  83                 helpString += "chimera.slayer(fasta=yourFastaFile, reference=yourTemplate, search=yourSearch) \n";
  84                 helpString += "Example: chimera.slayer(fasta=AD.align, reference=core_set_aligned.imputed.fasta, search=kmer) \n";
  85                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  86                 return helpString;
  87         }
  88         catch(exception& e) {
  89                 m->errorOut(e, "ChimeraSlayerCommand", "getHelpString");
  90                 exit(1);
  91         }
  92 }
  93 //**********************************************************************************************************************
  94 ChimeraSlayerCommand::ChimeraSlayerCommand(){
  95         try {
  96                 abort = true; calledHelp = true;
  97                 setParameters();
  98                 vector<string> tempOutNames;
  99                 outputTypes["chimera"] = tempOutNames;
 100                 outputTypes["accnos"] = tempOutNames;
 101                 outputTypes["fasta"] = tempOutNames;
 102         }
 103         catch(exception& e) {
 104                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 105                 exit(1);
 106         }
 107 }
 108 //***************************************************************************************************************
 109 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
 110         try {
 111                 abort = false; calledHelp = false;
 112
 113                 //allow user to run help
 114                 if(option == "help") { help(); abort = true; calledHelp = true; }
 115
 116                 else {
 117                         vector<string> myArray = setParameters();
 118
 119                         OptionParser parser(option);
 120                         map<string,string> parameters = parser.getParameters();
 121
 122                         ValidParameters validParameter("chimera.slayer");
 123                         map<string,string>::iterator it;
 124
 125                         //check to make sure all parameters are valid for command
 126                         for (it = parameters.begin(); it != parameters.end(); it++) {
 127                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 128                         }
 129
 130                         vector<string> tempOutNames;
 131                         outputTypes["chimera"] = tempOutNames;
 132                         outputTypes["accnos"] = tempOutNames;
 133                         outputTypes["fasta"] = tempOutNames;
 134
 135                         //if the user changes the input directory command factory will send this info to us in the output parameter
 136                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 137                         if (inputDir == "not found"){   inputDir = "";          }
 138
 139                         //check for required parameters
 140                         fastafile = validParameter.validFile(parameters, "fasta", false);
 141                         if (fastafile == "not found") {
 142                                 //if there is a current fasta file, use it
 143                                 string filename = m->getFastaFile();
 144                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 145                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 146                         }else {
 147                                 m->splitAtDash(fastafile, fastaFileNames);
 148
 149                                 //go through files and make sure they are good, if not, then disregard them
 150                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 151
 152                                         bool ignore = false;
 153                                         if (fastaFileNames[i] == "current") {
 154                                                 fastaFileNames[i] = m->getFastaFile();
 155                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 156                                                 else {
 157                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 158                                                         //erase from file list
 159                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 160                                                         i--;
 161                                                 }
 162                                         }
 163
 164                                         if (!ignore) {
 165
 166                                                 if (inputDir != "") {
 167                                                         string path = m->hasPath(fastaFileNames[i]);
 168                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 169                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 170                                                 }
 171
 172                                                 int ableToOpen;
 173                                                 ifstream in;
 174
 175                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 176
 177                                                 //if you can't open it, try default location
 178                                                 if (ableToOpen == 1) {
 179                                                         if (m->getDefaultPath() != "") { //default path is set
 180                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 181                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 182                                                                 ifstream in2;
 183                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 184                                                                 in2.close();
 185                                                                 fastaFileNames[i] = tryPath;
 186                                                         }
 187                                                 }
 188
 189                                                 if (ableToOpen == 1) {
 190                                                         if (m->getOutputDir() != "") { //default path is set
 191                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 192                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 193                                                                 ifstream in2;
 194                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 195                                                                 in2.close();
 196                                                                 fastaFileNames[i] = tryPath;
 197                                                         }
 198                                                 }
 199
 200                                                 in.close();
 201
 202                                                 if (ableToOpen == 1) {
 203                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 204                                                         //erase from file list
 205                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 206                                                         i--;
 207                                                 }
 208                                         }
 209                                 }
 210
 211                                 //make sure there is at least one valid file left
 212                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 213                         }
 214
 215
 216                         //check for required parameters
 217                         bool hasName = true;
 218                         namefile = validParameter.validFile(parameters, "name", false);
 219                         if (namefile == "not found") { namefile = "";  hasName = false; }
 220                         else {
 221                                 m->splitAtDash(namefile, nameFileNames);
 222
 223                                 //go through files and make sure they are good, if not, then disregard them
 224                                 for (int i = 0; i < nameFileNames.size(); i++) {
 225
 226                                         bool ignore = false;
 227                                         if (nameFileNames[i] == "current") {
 228                                                 nameFileNames[i] = m->getNameFile();
 229                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 230                                                 else {
 231                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 232                                                         //erase from file list
 233                                                         nameFileNames.erase(nameFileNames.begin()+i);
 234                                                         i--;
 235                                                 }
 236                                         }
 237
 238                                         if (!ignore) {
 239
 240                                                 if (inputDir != "") {
 241                                                         string path = m->hasPath(nameFileNames[i]);
 242                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 243                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 244                                                 }
 245
 246                                                 int ableToOpen;
 247                                                 ifstream in;
 248
 249                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 250
 251                                                 //if you can't open it, try default location
 252                                                 if (ableToOpen == 1) {
 253                                                         if (m->getDefaultPath() != "") { //default path is set
 254                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 255                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 256                                                                 ifstream in2;
 257                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 258                                                                 in2.close();
 259                                                                 nameFileNames[i] = tryPath;
 260                                                         }
 261                                                 }
 262
 263                                                 if (ableToOpen == 1) {
 264                                                         if (m->getOutputDir() != "") { //default path is set
 265                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 266                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 267                                                                 ifstream in2;
 268                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 269                                                                 in2.close();
 270                                                                 nameFileNames[i] = tryPath;
 271                                                         }
 272                                                 }
 273
 274                                                 in.close();
 275
 276                                                 if (ableToOpen == 1) {
 277                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 278                                                         //erase from file list
 279                                                         nameFileNames.erase(nameFileNames.begin()+i);
 280                                                         i--;
 281                                                 }
 282                                         }
 283                                 }
 284
 285                                 //make sure there is at least one valid file left
 286                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 287                         }
 288
 289                         if (hasName && (nameFileNames.size() != fastaFileNames.size())) { m->mothurOut("[ERROR]: The number of namefiles does not match the number of fastafiles, please correct."); m->mothurOutEndLine(); abort=true; }
 290
 291                         //if the user changes the output directory command factory will send this info to us in the output parameter
 292                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 293
 294
 295                         string path;
 296                         it = parameters.find("reference");
 297                         //user has given a template file
 298                         if(it != parameters.end()){
 299                                 if (it->second == "self") { templatefile = "self"; }
 300                                 else {
 301                                         path = m->hasPath(it->second);
 302                                         //if the user has not given a path then, add inputdir. else leave path alone.
 303                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 304
 305                                         templatefile = validParameter.validFile(parameters, "reference", true);
 306                                         if (templatefile == "not open") { abort = true; }
 307                                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("reference is a required parameter for the chimera.slayer command."); m->mothurOutEndLine(); abort = true;  }
 308                                 }
 309                         }
 310
 311                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 312                         m->setProcessors(temp);
 313                         convert(temp, processors);
 314
 315                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 316                         convert(temp, ksize);
 317
 318                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 319                         convert(temp, window);
 320
 321                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 322                         convert(temp, match);
 323
 324                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 325                         convert(temp, mismatch);
 326
 327                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 328                         convert(temp, divR);
 329
 330                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 331                         convert(temp, minSimilarity);
 332
 333                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 334                         convert(temp, minCoverage);
 335
 336                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 337                         convert(temp, minBS);
 338
 339                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "10"; }
 340                         convert(temp, minSNP);
 341
 342                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 343                         convert(temp, parents);
 344
 345                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "t"; }
 346                         realign = m->isTrue(temp);
 347
 348                         temp = validParameter.validFile(parameters, "trim", false);                             if (temp == "not found") { temp = "f"; }
 349                         trim = m->isTrue(temp);
 350
 351                         temp = validParameter.validFile(parameters, "split", false);                    if (temp == "not found") { temp = "f"; }
 352                         trimera = m->isTrue(temp);
 353
 354                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "blast"; }
 355
 356                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 357                         convert(temp, iters);
 358
 359                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 360                         convert(temp, increment);
 361
 362                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 363                         convert(temp, numwanted);
 364
 365                         if ((search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 366                 }
 367         }
 368         catch(exception& e) {
 369                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 370                 exit(1);
 371         }
 372 }
 373 //***************************************************************************************************************
 374
 375 int ChimeraSlayerCommand::execute(){
 376         try{
 377                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 378
 379                 for (int s = 0; s < fastaFileNames.size(); s++) {
 380
 381                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 382
 383                         int start = time(NULL);
 384
 385                         if (templatefile != "self") { //you want to run slayer with a refernce template
 386                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 387                         }else {
 388                                 if (processors != 1) { m->mothurOut("When using template=self, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 389                                 string nameFile = "";
 390                                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 391                                         nameFile = nameFileNames[s];
 392                                 }else {
 393                                         m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 394
 395                                         //use unique.seqs to create new name and fastafile
 396                                         string inputString = "fasta=" + fastaFileNames[s];
 397                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 398                                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 399
 400                                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 401                                         uniqueCommand->execute();
 402
 403                                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 404
 405                                         delete uniqueCommand;
 406
 407                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 408
 409                                         nameFile = filenames["name"][0];
 410                                         fastaFileNames[s] = filenames["fasta"][0];
 411                                 }
 412
 413                                 //sort fastafile by abundance, returns new sorted fastafile name
 414                                 m->mothurOut("Sorting fastafile according to abundance..."); cout.flush();
 415                                 map<string, int> priority = sortFastaFile(fastaFileNames[s], nameFile);
 416                                 m->mothurOut("Done."); m->mothurOutEndLine();
 417
 418                                 if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); }  return 0;    }
 419
 420                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, priority, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign);
 421                         }
 422
 423                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 424                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimera";
 425                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 426                         string trimFastaFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.fasta";
 427
 428                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0;    }
 429
 430                         if (chimera->getUnaligned()) {
 431                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 432                                 delete chimera;
 433                                 return 0;
 434                         }
 435                         templateSeqsLength = chimera->getLength();
 436
 437                 #ifdef USE_MPI
 438                         int pid, numSeqsPerProcessor;
 439                                 int tag = 2001;
 440                                 vector<unsigned long int> MPIPos;
 441
 442                                 MPI_Status status;
 443                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 444                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 445
 446                                 MPI_File inMPI;
 447                                 MPI_File outMPI;
 448                                 MPI_File outMPIAccnos;
 449                                 MPI_File outMPIFasta;
 450
 451                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 452                                 int inMode=MPI_MODE_RDONLY;
 453
 454                                 char outFilename[1024];
 455                                 strcpy(outFilename, outputFileName.c_str());
 456
 457                                 char outAccnosFilename[1024];
 458                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 459
 460                                 char outFastaFilename[1024];
 461                                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 462
 463                                 char inFileName[1024];
 464                                 strcpy(inFileName, fastaFileNames[s].c_str());
 465
 466                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 467                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 468                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 469                                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 470
 471                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) {  MPI_File_close(&outMPIFasta);  } MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }   delete chimera; return 0;  }
 472
 473                                 if (pid == 0) { //you are the root process
 474                                         m->mothurOutEndLine();
 475                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 476                                         m->mothurOutEndLine();
 477
 478                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 479
 480                                         //print header
 481                                         int length = outTemp.length();
 482                                         char* buf2 = new char[length];
 483                                         memcpy(buf2, outTemp.c_str(), length);
 484
 485                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 486                                         delete buf2;
 487
 488                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 489
 490                                         if (templatefile != "self") { //if template=self we can only use 1 processor
 491                                                 //send file positions to all processes
 492                                                 for(int i = 1; i < processors; i++) {
 493                                                         MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 494                                                         MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 495                                                 }
 496                                         }
 497                                         //figure out how many sequences you have to align
 498                                         numSeqsPerProcessor = numSeqs / processors;
 499                                         int startIndex =  pid * numSeqsPerProcessor;
 500                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 501
 502                                         if (templatefile == "self") { //if template=self we can only use 1 processor
 503                                                 startIndex = 0;
 504                                                 numSeqsPerProcessor = numSeqs;
 505                                         }
 506
 507                                         //do your part
 508                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 509
 510                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  remove(outputFileName.c_str());  remove(accnosFileName.c_str());  delete chimera; return 0;  }
 511
 512                                 }else{ //you are a child process
 513                                         if (templatefile != "self") { //if template=self we can only use 1 processor
 514                                                 MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 515                                                 MPIPos.resize(numSeqs+1);
 516                                                 MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 517
 518                                                 //figure out how many sequences you have to align
 519                                                 numSeqsPerProcessor = numSeqs / processors;
 520                                                 int startIndex =  pid * numSeqsPerProcessor;
 521                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 522
 523                                                 //do your part
 524                                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 525
 526                                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   remove(outputNames[j].c_str()); }  delete chimera; return 0;  }
 527
 528                                         }
 529                                 }
 530
 531                                 //close files
 532                                 MPI_File_close(&inMPI);
 533                                 MPI_File_close(&outMPI);
 534                                 MPI_File_close(&outMPIAccnos);
 535                                 if (trim) { MPI_File_close(&outMPIFasta); }
 536                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 537
 538                 #else
 539                         ofstream outHeader;
 540                         string tempHeader = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimeras.tempHeader";
 541                         m->openOutputFile(tempHeader, outHeader);
 542
 543                         chimera->printHeader(outHeader);
 544                         outHeader.close();
 545
 546                         vector<unsigned long int> positions = m->divideFile(fastaFileNames[s], processors);
 547
 548                         for (int i = 0; i < (positions.size()-1); i++) {
 549                                 lines.push_back(new linePair(positions[i], positions[(i+1)]));
 550                         }
 551
 552                         //break up file
 553                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 554                                 if(processors == 1){
 555                                         numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 556
 557                                         if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {      remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 558
 559                                 }else{
 560                                         processIDS.resize(0);
 561
 562                                         numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 563
 564                                         rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 565                                         rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 566                                         if (trim) {  rename((trimFastaFileName + toString(processIDS[0]) + ".temp").c_str(), trimFastaFileName.c_str()); }
 567
 568                                         //append output files
 569                                         for(int i=1;i<processors;i++){
 570                                                 m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 571                                                 remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 572                                         }
 573
 574                                         //append output files
 575                                         for(int i=1;i<processors;i++){
 576                                                 m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 577                                                 remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str());
 578                                         }
 579
 580                                         if (trim) {
 581                                                 for(int i=1;i<processors;i++){
 582                                                         m->appendFiles((trimFastaFileName + toString(processIDS[i]) + ".temp"), trimFastaFileName);
 583                                                         remove((trimFastaFileName + toString(processIDS[i]) + ".temp").c_str());
 584                                                 }
 585                                         }
 586
 587                                         if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 588                                 }
 589
 590                         #else
 591                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 592
 593                                 if (m->control_pressed) { outputTypes.clear(); if (trim) { remove(trimFastaFileName.c_str()); } remove(outputFileName.c_str()); remove(tempHeader.c_str()); remove(accnosFileName.c_str()); for (int j = 0; j < outputNames.size(); j++) {      remove(outputNames[j].c_str()); } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 594
 595                         #endif
 596
 597                         m->appendFiles(outputFileName, tempHeader);
 598
 599                         remove(outputFileName.c_str());
 600                         rename(tempHeader.c_str(), outputFileName.c_str());
 601
 602                 #endif
 603                         delete chimera;
 604
 605
 606                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 607
 608                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 609                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 610                         if (trim) {  outputNames.push_back(trimFastaFileName); outputTypes["fasta"].push_back(trimFastaFileName); }
 611
 612                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 613                 }
 614
 615                 //set accnos file as new current accnosfile
 616                 string current = "";
 617                 itTypes = outputTypes.find("accnos");
 618                 if (itTypes != outputTypes.end()) {
 619                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 620                 }
 621
 622                 if (trim) {
 623                         itTypes = outputTypes.find("fasta");
 624                         if (itTypes != outputTypes.end()) {
 625                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 626                         }
 627                 }
 628
 629                 m->mothurOutEndLine();
 630                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 631                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 632                 m->mothurOutEndLine();
 633
 634                 return 0;
 635
 636         }
 637         catch(exception& e) {
 638                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 639                 exit(1);
 640         }
 641 }
 642 //**********************************************************************************************************************
 643
 644 int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos, string fasta){
 645         try {
 646                 ofstream out;
 647                 m->openOutputFile(outputFName, out);
 648
 649                 ofstream out2;
 650                 m->openOutputFile(accnos, out2);
 651
 652                 ofstream out3;
 653                 if (trim) {  m->openOutputFile(fasta, out3); }
 654
 655                 ifstream inFASTA;
 656                 m->openInputFile(filename, inFASTA);
 657
 658                 inFASTA.seekg(filePos->start);
 659
 660                 bool done = false;
 661                 int count = 0;
 662
 663                 while (!done) {
 664
 665                         if (m->control_pressed) {       out.close(); out2.close(); if (trim) { out3.close(); } inFASTA.close(); return 1;       }
 666
 667                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 668                         string candidateAligned = candidateSeq->getAligned();
 669
 670                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 671
 672                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 673                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 674                                 }else{
 675                                         //find chimeras
 676                                         chimera->getChimeras(candidateSeq);
 677
 678                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 679
 680                                         //if you are not chimeric, then check each half
 681                                         data_results wholeResults = chimera->getResults();
 682
 683                                         //determine if we need to split
 684                                         bool isChimeric = false;
 685
 686                                         if (wholeResults.flag == "yes") {
 687                                                 string chimeraFlag = "no";
 688                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
 689                                                    ||
 690                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
 691
 692
 693                                                 if (chimeraFlag == "yes") {
 694                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
 695                                                 }
 696                                         }
 697
 698                                         if ((!isChimeric) && trimera) {
 699
 700                                                 //split sequence in half by bases
 701                                                 string leftQuery, rightQuery;
 702                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
 703                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
 704
 705                                                 //run chimeraSlayer on each piece
 706                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
 707                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
 708
 709                                                 //find chimeras
 710                                                 chimera->getChimeras(left);
 711                                                 data_results leftResults = chimera->getResults();
 712
 713                                                 chimera->getChimeras(right);
 714                                                 data_results rightResults = chimera->getResults();
 715
 716                                                 //if either piece is chimeric then report
 717                                                 Sequence* trimmed = chimera->print(out, out2, leftResults, rightResults);
 718                                                 if (trim) { trimmed->printSequence(out3); delete trimmed; }
 719
 720                                                 delete left; delete right;
 721
 722                                         }else { //already chimeric
 723                                                 //print results
 724                                                 Sequence* trimmed = chimera->print(out, out2);
 725                                                 if (trim) { trimmed->printSequence(out3); delete trimmed; }
 726                                         }
 727
 728
 729                                 }
 730                                 count++;
 731                         }
 732
 733                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 734                                 unsigned long int pos = inFASTA.tellg();
 735                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 736                         #else
 737                                 if (inFASTA.eof()) { break; }
 738                         #endif
 739
 740                         delete candidateSeq;
 741                         //report progress
 742                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 743                 }
 744                 //report progress
 745                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 746
 747                 out.close();
 748                 out2.close();
 749                 if (trim) { out3.close(); }
 750                 inFASTA.close();
 751
 752                 return count;
 753         }
 754         catch(exception& e) {
 755                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 756                 exit(1);
 757         }
 758 }
 759 //**********************************************************************************************************************
 760 #ifdef USE_MPI
 761 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, MPI_File& outFastaMPI, vector<unsigned long int>& MPIPos){
 762         try {
 763                 MPI_Status status;
 764                 int pid;
 765                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 766
 767                 for(int i=0;i<num;i++){
 768
 769                         if (m->control_pressed) {       return 1;       }
 770
 771                         //read next sequence
 772                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 773
 774                         char* buf4 = new char[length];
 775                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 776
 777                         string tempBuf = buf4;
 778                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 779                         istringstream iss (tempBuf,istringstream::in);
 780
 781                         delete buf4;
 782
 783                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 784                         string candidateAligned = candidateSeq->getAligned();
 785
 786                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 787
 788                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 789                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 790                                 }else{
 791
 792                                         //find chimeras
 793                                         chimera->getChimeras(candidateSeq);
 794
 795                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 796
 797                                         //if you are not chimeric, then check each half
 798                                         data_results wholeResults = chimera->getResults();
 799
 800                                         //determine if we need to split
 801                                         bool isChimeric = false;
 802
 803                                         if (wholeResults.flag == "yes") {
 804                                                 string chimeraFlag = "no";
 805                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
 806                                                    ||
 807                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
 808
 809
 810                                                 if (chimeraFlag == "yes") {
 811                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
 812                                                 }
 813                                         }
 814
 815                                         if ((!isChimeric) && trimera) {
 816                                                 //split sequence in half by bases
 817                                                 string leftQuery, rightQuery;
 818                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
 819                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
 820
 821                                                 //run chimeraSlayer on each piece
 822                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
 823                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
 824
 825                                                 //find chimeras
 826                                                 chimera->getChimeras(left);
 827                                                 data_results leftResults = chimera->getResults();
 828
 829                                                 chimera->getChimeras(right);
 830                                                 data_results rightResults = chimera->getResults();
 831
 832                                                 //if either piece is chimeric then report
 833                                                 Sequence* trimmed = chimera->print(outMPI, outAccMPI, leftResults, rightResults);
 834                                                 if (trim) {
 835                                                         string outputString = ">" + trimmed->getName() + "\n" + trimmed->getAligned() + "\n";
 836                                                         delete trimmed;
 837
 838                                                         //write to accnos file
 839                                                         int length = outputString.length();
 840                                                         char* buf2 = new char[length];
 841                                                         memcpy(buf2, outputString.c_str(), length);
 842
 843                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 844                                                         delete buf2;
 845                                                 }
 846
 847                                                 delete left; delete right;
 848
 849                                         }else {
 850                                                 //print results
 851                                                 Sequence* trimmed = chimera->print(outMPI, outAccMPI);
 852
 853                                                 if (trim) {
 854                                                         string outputString = ">" + trimmed->getName() + "\n" + trimmed->getAligned() + "\n";
 855                                                         delete trimmed;
 856
 857                                                         //write to accnos file
 858                                                         int length = outputString.length();
 859                                                         char* buf2 = new char[length];
 860                                                         memcpy(buf2, outputString.c_str(), length);
 861
 862                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 863                                                         delete buf2;
 864                                                 }
 865                                         }
 866
 867                                 }
 868                         }
 869                         delete candidateSeq;
 870
 871                         //report progress
 872                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 873                 }
 874                 //report progress
 875                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 876
 877
 878                 return 0;
 879         }
 880         catch(exception& e) {
 881                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 882                 exit(1);
 883         }
 884 }
 885 #endif
 886
 887 /**************************************************************************************************/
 888
 889 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos, string fasta) {
 890         try {
 891 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 892                 int process = 0;
 893                 int num = 0;
 894
 895                 //loop through and create all the processes you want
 896                 while (process != processors) {
 897                         int pid = fork();
 898
 899                         if (pid > 0) {
 900                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 901                                 process++;
 902                         }else if (pid == 0){
 903                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp");
 904
 905                                 //pass numSeqs to parent
 906                                 ofstream out;
 907                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 908                                 m->openOutputFile(tempFile, out);
 909                                 out << num << endl;
 910                                 out.close();
 911
 912                                 exit(0);
 913                         }else {
 914                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 915                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 916                                 exit(0);
 917                         }
 918                 }
 919
 920                 //force parent to wait until all the processes are done
 921                 for (int i=0;i<processors;i++) {
 922                         int temp = processIDS[i];
 923                         wait(&temp);
 924                 }
 925
 926                 for (int i = 0; i < processIDS.size(); i++) {
 927                         ifstream in;
 928                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 929                         m->openInputFile(tempFile, in);
 930                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 931                         in.close(); remove(tempFile.c_str());
 932                 }
 933
 934                 return num;
 935 #endif
 936         }
 937         catch(exception& e) {
 938                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
 939                 exit(1);
 940         }
 941 }
 942
 943 /**************************************************************************************************/
 944
 945 int ChimeraSlayerCommand::divideInHalf(Sequence querySeq, string& leftQuery, string& rightQuery) {
 946         try {
 947
 948                 string queryUnAligned = querySeq.getUnaligned();
 949                 int numBases = int(queryUnAligned.length() * 0.5);
 950
 951                 string queryAligned = querySeq.getAligned();
 952                 leftQuery = querySeq.getAligned();
 953                 rightQuery = querySeq.getAligned();
 954
 955                 int baseCount = 0;
 956                 int leftSpot = 0;
 957                 for (int i = 0; i < queryAligned.length(); i++) {
 958                         //if you are a base
 959                         if (isalpha(queryAligned[i])) {
 960                                 baseCount++;
 961                         }
 962
 963                         //if you have half
 964                         if (baseCount >= numBases) {  leftSpot = i; break; } //first half
 965                 }
 966
 967                 //blank out right side
 968                 for (int i = leftSpot; i < leftQuery.length(); i++) { leftQuery[i] = '.'; }
 969
 970                 //blank out left side
 971                 for (int i = 0; i < leftSpot; i++) { rightQuery[i] = '.'; }
 972
 973                 return 0;
 974
 975         }
 976         catch(exception& e) {
 977                 m->errorOut(e, "ChimeraSlayerCommand", "divideInHalf");
 978                 exit(1);
 979         }
 980 }
 981 /**************************************************************************************************/
 982 map<string, int> ChimeraSlayerCommand::sortFastaFile(string fastaFile, string nameFile) {
 983         try {
 984                 map<string, int> nameAbund;
 985
 986                 //read through fastafile and store info
 987                 map<string, string> seqs;
 988                 ifstream in;
 989                 m->openInputFile(fastaFile, in);
 990
 991                 while (!in.eof()) {
 992
 993                         if (m->control_pressed) { in.close(); return nameAbund; }
 994
 995                         Sequence seq(in); m->gobble(in);
 996                         seqs[seq.getName()] = seq.getAligned();
 997                 }
 998
 999                 in.close();
1000
1001                 //read namefile
1002                 vector<seqPriorityNode> nameMapCount;
1003                 int error = m->readNames(nameFile, nameMapCount, seqs);
1004
1005                 if (m->control_pressed) { return nameAbund; }
1006
1007                 if (error == 1) { m->control_pressed = true; return nameAbund; }
1008                 if (seqs.size() != nameMapCount.size()) { m->mothurOut( "The number of sequences in your fastafile does not match the number of sequences in your namefile, aborting."); m->mothurOutEndLine(); m->control_pressed = true; return nameAbund; }
1009
1010                 sort(nameMapCount.begin(), nameMapCount.end(), compareSeqPriorityNodes);
1011
1012                 string newFasta = fastaFile + ".temp";
1013                 ofstream out;
1014                 m->openOutputFile(newFasta, out);
1015
1016                 //print new file in order of
1017                 for (int i = 0; i < nameMapCount.size(); i++) {
1018                         out << ">" << nameMapCount[i].name << endl << nameMapCount[i].seq << endl;
1019                         nameAbund[nameMapCount[i].name] = nameMapCount[i].numIdentical;
1020                 }
1021                 out.close();
1022
1023                 rename(newFasta.c_str(), fastaFile.c_str());
1024
1025                 return nameAbund;
1026
1027         }
1028         catch(exception& e) {
1029                 m->errorOut(e, "ChimeraSlayerCommand", "sortFastaFile");
1030                 exit(1);
1031         }
1032 }
1033
1034 /**************************************************************************************************/
1035