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 "deconvolutecommand.h"
  12 #include "referencedb.h"
  13 #include "sequenceparser.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> ChimeraSlayerCommand::setParameters(){
  17         try {
  18                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  20                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pgroup);
  22                 CommandParameter pwindow("window", "Number", "", "50", "", "", "",false,false); parameters.push_back(pwindow);
  23                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  24                 CommandParameter pmatch("match", "Number", "", "5.0", "", "", "",false,false); parameters.push_back(pmatch);
  25                 CommandParameter pmismatch("mismatch", "Number", "", "-4.0", "", "", "",false,false); parameters.push_back(pmismatch);
  26                 CommandParameter pminsim("minsim", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminsim);
  27                 CommandParameter pmincov("mincov", "Number", "", "70", "", "", "",false,false); parameters.push_back(pmincov);
  28                 CommandParameter pminsnp("minsnp", "Number", "", "10", "", "", "",false,false); parameters.push_back(pminsnp);
  29                 CommandParameter pminbs("minbs", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminbs);
  30                 CommandParameter psearch("search", "Multiple", "kmer-blast", "blast", "", "", "",false,false); parameters.push_back(psearch);
  31                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  32                 CommandParameter prealign("realign", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(prealign);
  33                 CommandParameter ptrim("trim", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ptrim);
  34                 CommandParameter psplit("split", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psplit);
  35                 CommandParameter pnumwanted("numwanted", "Number", "", "15", "", "", "",false,false); parameters.push_back(pnumwanted);
  36                 CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  37                 CommandParameter pdivergence("divergence", "Number", "", "1.007", "", "", "",false,false); parameters.push_back(pdivergence);
  38                 CommandParameter pparents("parents", "Number", "", "3", "", "", "",false,false); parameters.push_back(pparents);
  39                 CommandParameter pincrement("increment", "Number", "", "5", "", "", "",false,false); parameters.push_back(pincrement);
  40                 CommandParameter pblastlocation("blastlocation", "String", "", "", "", "", "",false,false); parameters.push_back(pblastlocation);
  41                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  42                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  43                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  44
  45                 vector<string> myArray;
  46                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  47                 return myArray;
  48         }
  49         catch(exception& e) {
  50                 m->errorOut(e, "ChimeraSlayerCommand", "setParameters");
  51                 exit(1);
  52         }
  53 }
  54 //**********************************************************************************************************************
  55 string ChimeraSlayerCommand::getHelpString(){
  56         try {
  57                 string helpString = "";
  58                 helpString += "The chimera.slayer command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  59                 helpString += "This command was modeled after the chimeraSlayer written by the Broad Institute.\n";
  60                 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, blastlocation and realign.\n";
  61                 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";
  62                 helpString += "The name parameter allows you to provide a name file, if you are using reference=self. \n";
  63                 helpString += "The group parameter allows you to provide a group file. The group file can be used with a namesfile and reference=self. When checking sequences, only sequences from the same group as the query sequence will be used as the reference. \n";
  64                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  65                 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";
  66                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  67 #ifdef USE_MPI
  68                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  69 #endif
  70                 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";
  71                 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";
  72                 helpString += "The window parameter allows you to specify the window size for searching for chimeras, default=50. \n";
  73                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n";
  74                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n";
  75                 helpString += "The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n";
  76                 helpString += "The match parameter allows you to reward matched bases in blast search, default is 5. \n";
  77                 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";
  78                 helpString += "The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n";
  79                 helpString += "The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n";
  80                 helpString += "The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=1000.\n";
  81                 helpString += "The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n";
  82                 helpString += "The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n";
  83                 helpString += "The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n";
  84                 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";
  85                 helpString += "The search parameter allows you to specify search method for finding the closest parent. Choices are blast, and kmer, default blast. \n";
  86                 helpString += "The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default true.  \n";
  87                 helpString += "The blastlocation parameter allows you to specify the location of your blast executable. By default mothur will look in ./blast/bin relative to mothur's executable.  \n";
  88                 helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
  89                 helpString += "The chimera.slayer command should be in the following format: \n";
  90                 helpString += "chimera.slayer(fasta=yourFastaFile, reference=yourTemplate, search=yourSearch) \n";
  91                 helpString += "Example: chimera.slayer(fasta=AD.align, reference=core_set_aligned.imputed.fasta, search=kmer) \n";
  92                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  93                 return helpString;
  94         }
  95         catch(exception& e) {
  96                 m->errorOut(e, "ChimeraSlayerCommand", "getHelpString");
  97                 exit(1);
  98         }
  99 }
 100 //**********************************************************************************************************************
 101 ChimeraSlayerCommand::ChimeraSlayerCommand(){
 102         try {
 103                 abort = true; calledHelp = true;
 104                 setParameters();
 105                 vector<string> tempOutNames;
 106                 outputTypes["chimera"] = tempOutNames;
 107                 outputTypes["accnos"] = tempOutNames;
 108                 outputTypes["fasta"] = tempOutNames;
 109         }
 110         catch(exception& e) {
 111                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 112                 exit(1);
 113         }
 114 }
 115 //***************************************************************************************************************
 116 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
 117         try {
 118                 abort = false; calledHelp = false;
 119                 ReferenceDB* rdb = ReferenceDB::getInstance();
 120
 121                 //allow user to run help
 122                 if(option == "help") { help(); abort = true; calledHelp = true; }
 123                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 124
 125                 else {
 126                         vector<string> myArray = setParameters();
 127
 128                         OptionParser parser(option);
 129                         map<string,string> parameters = parser.getParameters();
 130
 131                         ValidParameters validParameter("chimera.slayer");
 132                         map<string,string>::iterator it;
 133
 134                         //check to make sure all parameters are valid for command
 135                         for (it = parameters.begin(); it != parameters.end(); it++) {
 136                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 137                         }
 138
 139                         vector<string> tempOutNames;
 140                         outputTypes["chimera"] = tempOutNames;
 141                         outputTypes["accnos"] = tempOutNames;
 142                         outputTypes["fasta"] = tempOutNames;
 143
 144                         //if the user changes the input directory command factory will send this info to us in the output parameter
 145                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 146                         if (inputDir == "not found"){   inputDir = "";          }
 147
 148                         //check for required parameters
 149                         fastafile = validParameter.validFile(parameters, "fasta", false);
 150                         if (fastafile == "not found") {
 151                                 //if there is a current fasta file, use it
 152                                 string filename = m->getFastaFile();
 153                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 154                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 155                         }else {
 156                                 m->splitAtDash(fastafile, fastaFileNames);
 157
 158                                 //go through files and make sure they are good, if not, then disregard them
 159                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 160
 161                                         bool ignore = false;
 162                                         if (fastaFileNames[i] == "current") {
 163                                                 fastaFileNames[i] = m->getFastaFile();
 164                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 165                                                 else {
 166                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 167                                                         //erase from file list
 168                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 169                                                         i--;
 170                                                 }
 171                                         }
 172
 173                                         if (!ignore) {
 174
 175                                                 if (inputDir != "") {
 176                                                         string path = m->hasPath(fastaFileNames[i]);
 177                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 178                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 179                                                 }
 180
 181                                                 int ableToOpen;
 182                                                 ifstream in;
 183
 184                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 185
 186                                                 //if you can't open it, try default location
 187                                                 if (ableToOpen == 1) {
 188                                                         if (m->getDefaultPath() != "") { //default path is set
 189                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 190                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 191                                                                 ifstream in2;
 192                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 193                                                                 in2.close();
 194                                                                 fastaFileNames[i] = tryPath;
 195                                                         }
 196                                                 }
 197
 198                                                 if (ableToOpen == 1) {
 199                                                         if (m->getOutputDir() != "") { //default path is set
 200                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 201                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 202                                                                 ifstream in2;
 203                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 204                                                                 in2.close();
 205                                                                 fastaFileNames[i] = tryPath;
 206                                                         }
 207                                                 }
 208
 209                                                 in.close();
 210
 211                                                 if (ableToOpen == 1) {
 212                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 213                                                         //erase from file list
 214                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 215                                                         i--;
 216                                                 }else {
 217                                                         m->setFastaFile(fastaFileNames[i]);
 218                                                 }
 219                                         }
 220                                 }
 221
 222                                 //make sure there is at least one valid file left
 223                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 224                         }
 225
 226
 227                         //check for required parameters
 228                         bool hasName = true;
 229                         namefile = validParameter.validFile(parameters, "name", false);
 230                         if (namefile == "not found") { namefile = "";  hasName = false; }
 231                         else {
 232                                 m->splitAtDash(namefile, nameFileNames);
 233
 234                                 //go through files and make sure they are good, if not, then disregard them
 235                                 for (int i = 0; i < nameFileNames.size(); i++) {
 236
 237                                         bool ignore = false;
 238                                         if (nameFileNames[i] == "current") {
 239                                                 nameFileNames[i] = m->getNameFile();
 240                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 241                                                 else {
 242                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 243                                                         //erase from file list
 244                                                         nameFileNames.erase(nameFileNames.begin()+i);
 245                                                         i--;
 246                                                 }
 247                                         }
 248
 249                                         if (!ignore) {
 250
 251                                                 if (inputDir != "") {
 252                                                         string path = m->hasPath(nameFileNames[i]);
 253                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 254                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 255                                                 }
 256
 257                                                 int ableToOpen;
 258                                                 ifstream in;
 259
 260                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 261
 262                                                 //if you can't open it, try default location
 263                                                 if (ableToOpen == 1) {
 264                                                         if (m->getDefaultPath() != "") { //default path is set
 265                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 266                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 267                                                                 ifstream in2;
 268                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 269                                                                 in2.close();
 270                                                                 nameFileNames[i] = tryPath;
 271                                                         }
 272                                                 }
 273
 274                                                 if (ableToOpen == 1) {
 275                                                         if (m->getOutputDir() != "") { //default path is set
 276                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 277                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 278                                                                 ifstream in2;
 279                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 280                                                                 in2.close();
 281                                                                 nameFileNames[i] = tryPath;
 282                                                         }
 283                                                 }
 284
 285                                                 in.close();
 286
 287                                                 if (ableToOpen == 1) {
 288                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 289                                                         //erase from file list
 290                                                         nameFileNames.erase(nameFileNames.begin()+i);
 291                                                         i--;
 292                                                 }else {
 293                                                         m->setNameFile(nameFileNames[i]);
 294                                                 }
 295                                         }
 296                                 }
 297
 298                                 //make sure there is at least one valid file left
 299                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 300                         }
 301
 302                         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; }
 303
 304                         bool hasGroup = true;
 305                         groupfile = validParameter.validFile(parameters, "group", false);
 306                         if (groupfile == "not found") { groupfile = "";  hasGroup = false; }
 307                         else {
 308                                 m->splitAtDash(groupfile, groupFileNames);
 309
 310                                 //go through files and make sure they are good, if not, then disregard them
 311                                 for (int i = 0; i < groupFileNames.size(); i++) {
 312
 313                                         bool ignore = false;
 314                                         if (groupFileNames[i] == "current") {
 315                                                 groupFileNames[i] = m->getGroupFile();
 316                                                 if (groupFileNames[i] != "") {  m->mothurOut("Using " + groupFileNames[i] + " as input file for the group parameter where you had given current."); m->mothurOutEndLine(); }
 317                                                 else {
 318                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 319                                                         //erase from file list
 320                                                         groupFileNames.erase(groupFileNames.begin()+i);
 321                                                         i--;
 322                                                 }
 323                                         }
 324
 325                                         if (!ignore) {
 326
 327                                                 if (inputDir != "") {
 328                                                         string path = m->hasPath(groupFileNames[i]);
 329                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 330                                                         if (path == "") {       groupFileNames[i] = inputDir + groupFileNames[i];               }
 331                                                 }
 332
 333                                                 int ableToOpen;
 334                                                 ifstream in;
 335
 336                                                 ableToOpen = m->openInputFile(groupFileNames[i], in, "noerror");
 337
 338                                                 //if you can't open it, try default location
 339                                                 if (ableToOpen == 1) {
 340                                                         if (m->getDefaultPath() != "") { //default path is set
 341                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(groupFileNames[i]);
 342                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 343                                                                 ifstream in2;
 344                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 345                                                                 in2.close();
 346                                                                 groupFileNames[i] = tryPath;
 347                                                         }
 348                                                 }
 349
 350                                                 if (ableToOpen == 1) {
 351                                                         if (m->getOutputDir() != "") { //default path is set
 352                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(groupFileNames[i]);
 353                                                                 m->mothurOut("Unable to open " + groupFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 354                                                                 ifstream in2;
 355                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 356                                                                 in2.close();
 357                                                                 groupFileNames[i] = tryPath;
 358                                                         }
 359                                                 }
 360
 361                                                 in.close();
 362
 363                                                 if (ableToOpen == 1) {
 364                                                         m->mothurOut("Unable to open " + groupFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 365                                                         //erase from file list
 366                                                         groupFileNames.erase(groupFileNames.begin()+i);
 367                                                         i--;
 368                                                 }else {
 369                                                         m->setGroupFile(groupFileNames[i]);
 370                                                 }
 371                                         }
 372                                 }
 373
 374                                 //make sure there is at least one valid file left
 375                                 if (groupFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid group files."); m->mothurOutEndLine(); abort = true; }
 376                         }
 377
 378                         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; }
 379
 380
 381                         //if the user changes the output directory command factory will send this info to us in the output parameter
 382                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 383
 384                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 385                         m->setProcessors(temp);
 386                         convert(temp, processors);
 387
 388                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 389                         save = m->isTrue(temp);
 390                         rdb->save = save;
 391                         if (save) { //clear out old references
 392                                 rdb->clearMemory();
 393                         }
 394
 395                         string path;
 396                         it = parameters.find("reference");
 397                         //user has given a template file
 398                         if(it != parameters.end()){
 399                                 if (it->second == "self") {
 400                                         templatefile = "self";
 401                                         if (save) {
 402                                                 m->mothurOut("[WARNING]: You can't save reference=self, ignoring save.");
 403                                                 m->mothurOutEndLine();
 404                                                 save = false;
 405                                         }
 406                                 }
 407                                 else {
 408                                         path = m->hasPath(it->second);
 409                                         //if the user has not given a path then, add inputdir. else leave path alone.
 410                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 411
 412                                         templatefile = validParameter.validFile(parameters, "reference", true);
 413                                         if (templatefile == "not open") { abort = true; }
 414                                         else if (templatefile == "not found") { //check for saved reference sequences
 415                                                 if (rdb->referenceSeqs.size() != 0) {
 416                                                         templatefile = "saved";
 417                                                 }else {
 418                                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 419                                                         m->mothurOutEndLine();
 420                                                         abort = true;
 421                                                 }
 422                                         }else { if (save) {     rdb->setSavedReference(templatefile);   }       }
 423                                 }
 424                         }else if (hasName) {  templatefile = "self";
 425                                 if (save) {
 426                                         m->mothurOut("[WARNING]: You can't save reference=self, ignoring save.");
 427                                         m->mothurOutEndLine();
 428                                         save = false;
 429                                 }
 430                         }
 431                         else {
 432                                 if (rdb->referenceSeqs.size() != 0) {
 433                                         templatefile = "saved";
 434                                 }else {
 435                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 436                                         m->mothurOutEndLine();
 437                                         templatefile = ""; abort = true;
 438                                 }
 439                         }
 440
 441
 442
 443                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 444                         convert(temp, ksize);
 445
 446                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 447                         convert(temp, window);
 448
 449                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 450                         convert(temp, match);
 451
 452                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 453                         convert(temp, mismatch);
 454
 455                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 456                         convert(temp, divR);
 457
 458                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 459                         convert(temp, minSimilarity);
 460
 461                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 462                         convert(temp, minCoverage);
 463
 464                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 465                         convert(temp, minBS);
 466
 467                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "10"; }
 468                         convert(temp, minSNP);
 469
 470                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 471                         convert(temp, parents);
 472
 473                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "t"; }
 474                         realign = m->isTrue(temp);
 475
 476                         temp = validParameter.validFile(parameters, "trim", false);                             if (temp == "not found") { temp = "f"; }
 477                         trim = m->isTrue(temp);
 478
 479                         temp = validParameter.validFile(parameters, "split", false);                    if (temp == "not found") { temp = "f"; }
 480                         trimera = m->isTrue(temp);
 481
 482                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "blast"; }
 483
 484                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 485                         convert(temp, iters);
 486
 487                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 488                         convert(temp, increment);
 489
 490                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 491                         convert(temp, numwanted);
 492
 493                         blastlocation = validParameter.validFile(parameters, "blastlocation", false);
 494                         if (blastlocation == "not found") { blastlocation = ""; }
 495                         else {
 496                                 //add / to name if needed
 497                                 string lastChar = blastlocation.substr(blastlocation.length()-1);
 498 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 499                                 if (lastChar != "/") { blastlocation += "/"; }
 500 #else
 501                                 if (lastChar != "\\") { blastlocation += "\\"; }
 502 #endif
 503                                 blastlocation = m->getFullPathName(blastlocation);
 504                                 string formatdbCommand = "";
 505 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 506                                 formatdbCommand = blastlocation + "formatdb";
 507 #else
 508                                 formatdbCommand = blastlocation + "formatdb.exe";
 509 #endif
 510
 511                                 //test to make sure formatdb exists
 512                                 ifstream in;
 513                                 formatdbCommand = m->getFullPathName(formatdbCommand);
 514                                 int ableToOpen = m->openInputFile(formatdbCommand, in, "no error"); in.close();
 515                                 if(ableToOpen == 1) {   m->mothurOut("[ERROR]: " + formatdbCommand + " file does not exist. mothur requires formatdb.exe to run chimera.slayer."); m->mothurOutEndLine(); abort = true; }
 516
 517                                 string blastCommand = "";
 518 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 519                                 blastCommand = blastlocation + "megablast";
 520 #else
 521                                 blastCommand = blastlocation + "megablast.exe";
 522 #endif
 523                                 //test to make sure formatdb exists
 524                                 ifstream in2;
 525                                 blastCommand = m->getFullPathName(blastCommand);
 526                                 ableToOpen = m->openInputFile(blastCommand, in2, "no error"); in2.close();
 527                                 if(ableToOpen == 1) {   m->mothurOut("[ERROR]: " + blastCommand + " file does not exist. mothur requires blastall.exe to run chimera.slayer."); m->mothurOutEndLine(); abort = true; }
 528                         }
 529
 530                         if ((search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 531
 532                         if (hasName && (templatefile != "self")) { m->mothurOut("You have provided a namefile and the reference parameter is not set to self. I am not sure what reference you are trying to use, aborting."); m->mothurOutEndLine(); abort=true; }
 533                         if (hasGroup && (templatefile != "self")) { m->mothurOut("You have provided a group file and the reference parameter is not set to self. I am not sure what reference you are trying to use, aborting."); m->mothurOutEndLine(); abort=true; }
 534
 535                         //until we resolve the issue 10-18-11
 536 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 537 #else
 538                         //processors=1;
 539 #endif
 540                 }
 541         }
 542         catch(exception& e) {
 543                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 544                 exit(1);
 545         }
 546 }
 547 //***************************************************************************************************************
 548
 549 int ChimeraSlayerCommand::execute(){
 550         try{
 551                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 552
 553                 for (int s = 0; s < fastaFileNames.size(); s++) {
 554
 555                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 556
 557                         int start = time(NULL);
 558                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 559                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimera";
 560                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 561                         string trimFastaFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.fasta";
 562
 563                         //clears files
 564                         ofstream out, out1, out2;
 565                         m->openOutputFile(outputFileName, out); out.close();
 566                         m->openOutputFile(accnosFileName, out1); out1.close();
 567                         if (trim) { m->openOutputFile(trimFastaFileName, out2); out2.close(); }
 568                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 569                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 570                         if (trim) {  outputNames.push_back(trimFastaFileName); outputTypes["fasta"].push_back(trimFastaFileName); }
 571
 572                         //maps a filename to priority map.
 573                         //if no groupfile this is fastafileNames[s] -> prioirity
 574                         //if groupfile then this is each groups seqs -> priority
 575                         map<string, map<string, int> > fileToPriority;
 576                         map<string, map<string, int> >::iterator itFile;
 577                         map<string, string> fileGroup;
 578                         fileToPriority[fastaFileNames[s]] = priority; //default
 579                         fileGroup[fastaFileNames[s]] = "noGroup";
 580                         SequenceParser* parser = NULL;
 581                         int totalChimeras = 0;
 582                         lines.clear();
 583
 584                         if (templatefile == "self") { setUpForSelfReference(parser, fileGroup, fileToPriority, s); }
 585
 586                         if (m->control_pressed) {  if (parser != NULL) { delete parser; } for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;    }
 587
 588                         if (fileToPriority.size() == 1) { //you running without a groupfile
 589                                 itFile = fileToPriority.begin();
 590                                 string thisFastaName = itFile->first;
 591                                 map<string, int> thisPriority = itFile->second;
 592 #ifdef USE_MPI
 593                                 MPIExecute(thisFastaName, outputFileName, accnosFileName, trimFastaFileName, thisPriority);
 594 #else
 595                                 //break up file
 596                                 vector<unsigned long long> positions;
 597 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 598                                 positions = m->divideFile(thisFastaName, processors);
 599                                 for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(linePair(positions[i], positions[(i+1)]));      }
 600 #else
 601                                 if (processors == 1) {  lines.push_back(linePair(0, 1000)); }
 602                                 else {
 603                                         positions = m->setFilePosFasta(thisFastaName, numSeqs);
 604
 605                                         //figure out how many sequences you have to process
 606                                         int numSeqsPerProcessor = numSeqs / processors;
 607                                         for (int i = 0; i < processors; i++) {
 608                                                 int startIndex =  i * numSeqsPerProcessor;
 609                                                 if(i == (processors - 1)){      numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;        }
 610                                                 lines.push_back(linePair(positions[startIndex], numSeqsPerProcessor));
 611                                         }
 612                                 }
 613 #endif
 614                                 if(processors == 1){ numSeqs = driver(lines[0], outputFileName, thisFastaName, accnosFileName, trimFastaFileName, thisPriority);  }
 615                                 else{ numSeqs = createProcesses(outputFileName, thisFastaName, accnosFileName, trimFastaFileName, thisPriority); }
 616
 617                                 if (m->control_pressed) { if (parser != NULL) { delete parser; }  outputTypes.clear(); if (trim) { m->mothurRemove(trimFastaFileName); } m->mothurRemove(outputFileName); m->mothurRemove(accnosFileName); for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        }  return 0; }
 618 #endif
 619                         }else { //you have provided a groupfile
 620 #ifdef USE_MPI
 621                                 MPIExecuteGroups(outputFileName, accnosFileName, trimFastaFileName, fileToPriority, fileGroup);
 622 #else
 623                                 if (processors == 1) { numSeqs = driverGroups(outputFileName, accnosFileName, trimFastaFileName, fileToPriority, fileGroup);    }
 624                                 else {  numSeqs = createProcessesGroups(outputFileName, accnosFileName, trimFastaFileName, fileToPriority, fileGroup);          } //destroys fileToPriority
 625 #endif
 626
 627 #ifdef USE_MPI
 628                                 int pid;
 629                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 630
 631                                 if (pid == 0) {
 632 #endif
 633
 634                                 totalChimeras = deconvoluteResults(parser, outputFileName, accnosFileName, trimFastaFileName);
 635 #ifdef USE_MPI
 636                                 }
 637                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait
 638 #endif
 639                         }
 640
 641                         if (parser != NULL) { delete parser; }
 642
 643                         m->mothurOutEndLine(); m->mothurOut(toString(totalChimeras) + " chimera found."); m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");       m->mothurOutEndLine();
 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                 if (trim) {
 654                         itTypes = outputTypes.find("fasta");
 655                         if (itTypes != outputTypes.end()) {
 656                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 657                         }
 658                 }
 659
 660                 m->mothurOutEndLine();
 661                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 662                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 663                 m->mothurOutEndLine();
 664
 665                 return 0;
 666
 667         }
 668         catch(exception& e) {
 669                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 670                 exit(1);
 671         }
 672 }
 673 //**********************************************************************************************************************
 674 int ChimeraSlayerCommand::MPIExecuteGroups(string outputFileName, string accnosFileName, string trimFastaFileName, map<string, map<string, int> >& fileToPriority, map<string, string>& fileGroup){
 675         try {
 676 #ifdef USE_MPI
 677                 int pid;
 678                 int tag = 2001;
 679
 680                 MPI_Status status;
 681                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 682                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 683
 684                 //put filenames in a vector, then pass each process a starting and ending point in the vector
 685                 //all processes already have the fileToPriority and fileGroup, they just need to know which files to process
 686                 map<string, map<string, int> >::iterator itFile;
 687                 vector<string> filenames;
 688                 for(itFile = fileToPriority.begin(); itFile != fileToPriority.end(); itFile++) { filenames.push_back(itFile->first); }
 689
 690                 int numGroupsPerProcessor = filenames.size() / processors;
 691                 int startIndex =  pid * numGroupsPerProcessor;
 692                 int endIndex = (pid+1) * numGroupsPerProcessor;
 693                 if(pid == (processors - 1)){    endIndex = filenames.size();    }
 694
 695                 vector<unsigned long long> MPIPos;
 696
 697                 MPI_File outMPI;
 698                 MPI_File outMPIAccnos;
 699                 MPI_File outMPIFasta;
 700
 701                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 702                 int inMode=MPI_MODE_RDONLY;
 703
 704                 char outFilename[1024];
 705                 strcpy(outFilename, outputFileName.c_str());
 706
 707                 char outAccnosFilename[1024];
 708                 strcpy(outAccnosFilename, accnosFileName.c_str());
 709
 710                 char outFastaFilename[1024];
 711                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 712
 713                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 714                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 715                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 716
 717                 if (m->control_pressed) {   MPI_File_close(&outMPI); if (trim) {  MPI_File_close(&outMPIFasta);  } MPI_File_close(&outMPIAccnos);  return 0;  }
 718
 719                 //print headers
 720                 if (pid == 0) { //you are the root process
 721                         m->mothurOutEndLine();
 722                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 723                         m->mothurOutEndLine();
 724
 725                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 726
 727                         //print header
 728                         int length = outTemp.length();
 729                         char* buf2 = new char[length];
 730                         memcpy(buf2, outTemp.c_str(), length);
 731
 732                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 733                         delete buf2;
 734                 }
 735                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait
 736
 737                 for (int i = startIndex; i < endIndex; i++) {
 738
 739                         int start = time(NULL);
 740                         int num = 0;
 741                         string thisFastaName = filenames[i];
 742                         map<string, int> thisPriority = fileToPriority[thisFastaName];
 743
 744                         char inFileName[1024];
 745                         strcpy(inFileName, thisFastaName.c_str());
 746                         MPI_File inMPI;
 747                         MPI_File_open(MPI_COMM_SELF, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 748
 749                         MPIPos = m->setFilePosFasta(thisFastaName, num); //fills MPIPos, returns numSeqs
 750
 751                         cout << endl << "Checking sequences from group: " << fileGroup[thisFastaName] << "." << endl;
 752
 753                         driverMPI(0, num, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos, thisFastaName, thisPriority, true);
 754                         numSeqs += num;
 755
 756                         MPI_File_close(&inMPI);
 757                         m->mothurRemove(thisFastaName);
 758
 759                         cout << endl << "It took " << toString(time(NULL) - start) << " secs to check " + toString(num) + " sequences from group " << fileGroup[thisFastaName] << "." << endl;
 760                 }
 761
 762                 if (pid == 0) {
 763                         for(int i = 1; i < processors; i++) {
 764                                 int temp = 0;
 765                                 MPI_Recv(&temp, 1, MPI_INT, i, 2001, MPI_COMM_WORLD, &status);
 766                                 numSeqs += temp;
 767                         }
 768                 }else{ MPI_Send(&numSeqs, 1, MPI_INT, 0, 2001, MPI_COMM_WORLD); }
 769
 770                 MPI_File_close(&outMPI);
 771                 MPI_File_close(&outMPIAccnos);
 772                 if (trim) { MPI_File_close(&outMPIFasta); }
 773
 774                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait
 775 #endif
 776                 return 0;
 777
 778         }catch(exception& e) {
 779                 m->errorOut(e, "ChimeraSlayerCommand", "MPIExecuteGroups");
 780                 exit(1);
 781         }
 782 }
 783 //**********************************************************************************************************************
 784 int ChimeraSlayerCommand::MPIExecute(string inputFile, string outputFileName, string accnosFileName, string trimFastaFileName, map<string, int>& priority){
 785         try {
 786
 787 #ifdef USE_MPI
 788                 int pid, numSeqsPerProcessor;
 789                 int tag = 2001;
 790                 vector<unsigned long long> MPIPos;
 791
 792                 MPI_Status status;
 793                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 794                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 795
 796                 MPI_File inMPI;
 797                 MPI_File outMPI;
 798                 MPI_File outMPIAccnos;
 799                 MPI_File outMPIFasta;
 800
 801                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 802                 int inMode=MPI_MODE_RDONLY;
 803
 804                 char outFilename[1024];
 805                 strcpy(outFilename, outputFileName.c_str());
 806
 807                 char outAccnosFilename[1024];
 808                 strcpy(outAccnosFilename, accnosFileName.c_str());
 809
 810                 char outFastaFilename[1024];
 811                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 812
 813                 char inFileName[1024];
 814                 strcpy(inFileName, inputFile.c_str());
 815
 816                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 817                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 818                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 819                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 820
 821                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) {  MPI_File_close(&outMPIFasta);  } MPI_File_close(&outMPIAccnos);  return 0;  }
 822
 823                 if (pid == 0) { //you are the root process
 824                         m->mothurOutEndLine();
 825                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 826                         m->mothurOutEndLine();
 827
 828                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 829
 830                         //print header
 831                         int length = outTemp.length();
 832                         char* buf2 = new char[length];
 833                         memcpy(buf2, outTemp.c_str(), length);
 834
 835                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 836                         delete buf2;
 837
 838                         MPIPos = m->setFilePosFasta(inputFile, numSeqs); //fills MPIPos, returns numSeqs
 839
 840                         if (templatefile != "self") { //if template=self we can only use 1 processor
 841                                 //send file positions to all processes
 842                                 for(int i = 1; i < processors; i++) {
 843                                         MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 844                                         MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 845                                 }
 846                         }
 847                         //figure out how many sequences you have to align
 848                         numSeqsPerProcessor = numSeqs / processors;
 849                         int startIndex =  pid * numSeqsPerProcessor;
 850                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 851
 852                         if (templatefile == "self") { //if template=self we can only use 1 processor
 853                                 startIndex = 0;
 854                                 numSeqsPerProcessor = numSeqs;
 855                         }
 856
 857                         //do your part
 858                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos, inputFile, priority, false);
 859
 860                         if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);   return 0;  }
 861
 862                 }else{ //you are a child process
 863                         if (templatefile != "self") { //if template=self we can only use 1 processor
 864                                 MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 865                                 MPIPos.resize(numSeqs+1);
 866                                 MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 867
 868                                 //figure out how many sequences you have to align
 869                                 numSeqsPerProcessor = numSeqs / processors;
 870                                 int startIndex =  pid * numSeqsPerProcessor;
 871                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 872
 873                                 //do your part
 874                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos, inputFile, priority, false);
 875
 876                                 if (m->control_pressed) {  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  return 0;  }
 877
 878                         }
 879                 }
 880
 881                 //close files
 882                 MPI_File_close(&inMPI);
 883                 MPI_File_close(&outMPI);
 884                 MPI_File_close(&outMPIAccnos);
 885                 if (trim) { MPI_File_close(&outMPIFasta); }
 886                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 887
 888
 889 #endif
 890                 return numSeqs;
 891         }
 892         catch(exception& e) {
 893                 m->errorOut(e, "ChimeraSlayerCommand", "MPIExecute");
 894                 exit(1);
 895         }
 896 }
 897 //**********************************************************************************************************************
 898 int ChimeraSlayerCommand::deconvoluteResults(SequenceParser* parser, string outputFileName, string accnosFileName, string trimFileName){
 899         try {
 900                 map<string, string> uniqueNames = parser->getAllSeqsMap();
 901                 map<string, string>::iterator itUnique;
 902                 int total = 0;
 903
 904                 //edit accnos file
 905                 ifstream in2;
 906                 m->openInputFile(accnosFileName, in2, "no error");
 907
 908                 ofstream out2;
 909                 m->openOutputFile(accnosFileName+".temp", out2);
 910
 911                 string name; name = "";
 912                 set<string> chimerasInFile;
 913                 set<string>::iterator itChimeras;
 914
 915                 while (!in2.eof()) {
 916                         if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(outputFileName); m->mothurRemove((accnosFileName+".temp")); return 0; }
 917
 918                         in2 >> name; m->gobble(in2);
 919
 920                         //find unique name
 921                         itUnique = uniqueNames.find(name);
 922
 923                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing accnos results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
 924                         else {
 925                                 itChimeras = chimerasInFile.find((itUnique->second));
 926
 927                                 if (itChimeras == chimerasInFile.end()) {
 928                                         out2 << itUnique->second << endl;
 929                                         chimerasInFile.insert((itUnique->second));
 930                                         total++;
 931                                 }
 932                         }
 933                 }
 934                 in2.close();
 935                 out2.close();
 936
 937                 m->mothurRemove(accnosFileName);
 938                 rename((accnosFileName+".temp").c_str(), accnosFileName.c_str());
 939
 940
 941                 //edit chimera file
 942                 ifstream in;
 943                 m->openInputFile(outputFileName, in);
 944
 945                 ofstream out;
 946                 m->openOutputFile(outputFileName+".temp", out); out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
 947
 948                 string rest, parent1, parent2, line;
 949                 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
 950                 set<string>::iterator itNames;
 951
 952                 //assumptions - in file each read will always look like...
 953                 /*
 954                  F11Fcsw_92754  no
 955                  F11Fcsw_63104  F11Fcsw_33372   F11Fcsw_37007   0.89441 80.4469 0.2     1.03727 93.2961 52.2    no      0-241   243-369
 956                  */
 957
 958                 //get header line
 959                 if (!in.eof()) {
 960                         line = m->getline(in); m->gobble(in);
 961                         out << line << endl;
 962                 }
 963
 964                 //for the chimera file, we want to make sure if any group finds a sequence to be chimeric then all groups do,
 965                 //so if this is a report that did not find it to be chimeric, but it appears in the accnos file,
 966                 //then ignore this report and continue until we find the report that found it to be chimeric
 967
 968                 while (!in.eof()) {
 969
 970                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove((outputFileName+".temp")); return 0; }
 971
 972                         in >> name;             m->gobble(in);
 973                         in >> parent1;  m->gobble(in);
 974
 975                         if (name == "Name") { //name = "Name" because we append the header line each time we add results from the groups
 976                                 line = m->getline(in); m->gobble(in);
 977                         }else {
 978                                 if (parent1 == "no") {
 979                                         //find unique name
 980                                         itUnique = uniqueNames.find(name);
 981
 982                                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
 983                                         else {
 984                                                 //is this sequence really not chimeric??
 985                                                 itChimeras = chimerasInFile.find(itUnique->second);
 986
 987                                                 if (itChimeras == chimerasInFile.end()) {
 988                                                         //is this sequence not already in the file
 989                                                         itNames = namesInFile.find((itUnique->second));
 990
 991                                                         if (itNames == namesInFile.end()) { out << itUnique->second << '\t' << "no" << endl; namesInFile.insert(itUnique->second); }
 992                                                 }
 993                                         }
 994                                 }else { //read the rest of the line
 995                                         double DivQLAQRB,PerIDQLAQRB,BootStrapA,DivQLBQRA,PerIDQLBQRA,BootStrapB;
 996                                         string flag, range1, range2;
 997                                         bool print = false;
 998                                         in >> parent2 >> DivQLAQRB >> PerIDQLAQRB >> BootStrapA >> DivQLBQRA >> PerIDQLBQRA >> BootStrapB >> flag >> range1 >> range2;  m->gobble(in);
 999
1000                                         //find unique name
1001                                         itUnique = uniqueNames.find(name);
1002
1003                                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find "+ name + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1004                                         else {
1005                                                 name = itUnique->second;
1006                                                 //is this name already in the file
1007                                                 itNames = namesInFile.find((name));
1008
1009                                                 if (itNames == namesInFile.end()) { //no not in file
1010                                                         if (flag == "no") { //are you really a no??
1011                                                                 //is this sequence really not chimeric??
1012                                                                 itChimeras = chimerasInFile.find(name);
1013
1014                                                                 //then you really are a no so print, otherwise skip
1015                                                                 if (itChimeras == chimerasInFile.end()) { print = true; }
1016
1017                                                         }else{ print = true; }
1018                                                 }
1019                                         }
1020
1021                                         if (print) {
1022                                                 out << name << '\t';
1023
1024                                                 namesInFile.insert(name);
1025
1026                                                 //output parent1's name
1027                                                 itUnique = uniqueNames.find(parent1);
1028                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parentA "+ parent1 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1029                                                 else { out << itUnique->second << '\t'; }
1030
1031                                                 //output parent2's name
1032                                                 itUnique = uniqueNames.find(parent2);
1033                                                 if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing chimera results. Cannot find parentA "+ parent2 + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1034                                                 else { out << itUnique->second << '\t'; }
1035
1036                                                 out << DivQLAQRB << '\t' << PerIDQLAQRB << '\t' << BootStrapA << '\t' << DivQLBQRA << '\t' << PerIDQLBQRA << '\t' << BootStrapB << '\t' << flag << '\t' << range1 << '\t' << range2 << endl;
1037                                         }
1038                                 }
1039                         }
1040                 }
1041                 in.close();
1042                 out.close();
1043
1044                 m->mothurRemove(outputFileName);
1045                 rename((outputFileName+".temp").c_str(), outputFileName.c_str());
1046
1047                 //edit fasta file
1048                 if (trim) {
1049                         ifstream in3;
1050                         m->openInputFile(trimFileName, in3);
1051
1052                         ofstream out3;
1053                         m->openOutputFile(trimFileName+".temp", out3);
1054
1055                         namesInFile.clear();
1056
1057                         while (!in3.eof()) {
1058                                 if (m->control_pressed) { in3.close(); out3.close(); m->mothurRemove(outputFileName); m->mothurRemove(accnosFileName); m->mothurRemove((trimFileName+".temp")); return 0; }
1059
1060                                 Sequence seq(in3); m->gobble(in3);
1061
1062                                 if (seq.getName() != "") {
1063                                         //find unique name
1064                                         itUnique = uniqueNames.find(seq.getName());
1065
1066                                         if (itUnique == uniqueNames.end()) { m->mothurOut("[ERROR]: trouble parsing accnos results. Cannot find "+ seq.getName() + "."); m->mothurOutEndLine(); m->control_pressed = true; }
1067                                         else {
1068                                                 itNames = namesInFile.find((itUnique->second));
1069
1070                                                 if (itNames == namesInFile.end()) {
1071                                                         seq.printSequence(out3);
1072                                                 }
1073                                         }
1074                                 }
1075                         }
1076                         in3.close();
1077                         out3.close();
1078
1079                         m->mothurRemove(trimFileName);
1080                         rename((trimFileName+".temp").c_str(), trimFileName.c_str());
1081                 }
1082
1083                 return total;
1084         }
1085         catch(exception& e) {
1086                 m->errorOut(e, "ChimeraSlayerCommand", "deconvoluteResults");
1087                 exit(1);
1088         }
1089 }
1090 //**********************************************************************************************************************
1091 int ChimeraSlayerCommand::setUpForSelfReference(SequenceParser*& parser, map<string, string>& fileGroup, map<string, map<string, int> >& fileToPriority, int s){
1092         try {
1093                 fileGroup.clear();
1094                 fileToPriority.clear();
1095
1096                 string nameFile = "";
1097                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
1098                         nameFile = nameFileNames[s];
1099                 }else {  nameFile = getNamesFile(fastaFileNames[s]); }
1100
1101                 //you provided a groupfile
1102                 string groupFile = "";
1103                 if (groupFileNames.size() != 0) { groupFile = groupFileNames[s]; }
1104
1105                 if (groupFile == "") {
1106                         if (processors != 1) { m->mothurOut("When using template=self, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
1107
1108                         //sort fastafile by abundance, returns new sorted fastafile name
1109                         m->mothurOut("Sorting fastafile according to abundance..."); cout.flush();
1110                         priority = sortFastaFile(fastaFileNames[s], nameFile);
1111                         m->mothurOut("Done."); m->mothurOutEndLine();
1112
1113                         fileToPriority[fastaFileNames[s]] = priority;
1114                         fileGroup[fastaFileNames[s]] = "noGroup";
1115                 }else {
1116                         //Parse sequences by group
1117                         parser = new SequenceParser(groupFile, fastaFileNames[s], nameFile);
1118                         vector<string> groups = parser->getNamesOfGroups();
1119
1120                         for (int i = 0; i < groups.size(); i++) {
1121                                 vector<Sequence> thisGroupsSeqs = parser->getSeqs(groups[i]);
1122                                 map<string, string> thisGroupsMap = parser->getNameMap(groups[i]);
1123                                 string newFastaFile = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + groups[i] + "-sortedTemp.fasta";
1124                                 priority = sortFastaFile(thisGroupsSeqs, thisGroupsMap, newFastaFile);
1125                                 fileToPriority[newFastaFile] = priority;
1126                                 fileGroup[newFastaFile] = groups[i];
1127                         }
1128                 }
1129
1130
1131                 return 0;
1132         }
1133         catch(exception& e) {
1134                 m->errorOut(e, "ChimeraSlayerCommand", "setUpForSelfReference");
1135                 exit(1);
1136         }
1137 }
1138
1139 //**********************************************************************************************************************
1140 string ChimeraSlayerCommand::getNamesFile(string& inputFile){
1141         try {
1142                 string nameFile = "";
1143
1144                 m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
1145
1146                 //use unique.seqs to create new name and fastafile
1147                 string inputString = "fasta=" + inputFile;
1148                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
1149                 m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
1150
1151                 Command* uniqueCommand = new DeconvoluteCommand(inputString);
1152                 uniqueCommand->execute();
1153
1154                 map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
1155
1156                 delete uniqueCommand;
1157
1158                 m->mothurOut("/******************************************/"); m->mothurOutEndLine();
1159
1160                 nameFile = filenames["name"][0];
1161                 inputFile = filenames["fasta"][0];
1162
1163                 return nameFile;
1164         }
1165         catch(exception& e) {
1166                 m->errorOut(e, "ChimeraSlayerCommand", "getNamesFile");
1167                 exit(1);
1168         }
1169 }
1170 //**********************************************************************************************************************
1171
1172 int ChimeraSlayerCommand::driverGroups(string outputFName, string accnos, string fasta, map<string, map<string, int> >& fileToPriority, map<string, string>& fileGroup){
1173         try {
1174                 int totalSeqs = 0;
1175
1176                 for (map<string, map<string, int> >::iterator itFile = fileToPriority.begin(); itFile != fileToPriority.end(); itFile++) {
1177
1178                         if (m->control_pressed) {  return 0;  }
1179
1180                         int start = time(NULL);
1181                         string thisFastaName = itFile->first;
1182                         map<string, int> thisPriority = itFile->second;
1183                         string thisoutputFileName = outputDir + m->getRootName(m->getSimpleName(thisFastaName)) + fileGroup[thisFastaName] + "slayer.chimera";
1184                         string thisaccnosFileName = outputDir + m->getRootName(m->getSimpleName(thisFastaName)) + fileGroup[thisFastaName] + "slayer.accnos";
1185                         string thistrimFastaFileName = outputDir + m->getRootName(m->getSimpleName(thisFastaName)) + fileGroup[thisFastaName] + "slayer.fasta";
1186
1187                         m->mothurOutEndLine(); m->mothurOut("Checking sequences from group: " + fileGroup[thisFastaName] + "."); m->mothurOutEndLine();
1188
1189                         lines.clear();
1190 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1191                         int proc = 1;
1192                         vector<unsigned long long> positions = m->divideFile(thisFastaName, proc);
1193                         lines.push_back(linePair(positions[0], positions[1]));
1194 #else
1195                         lines.push_back(linePair(0, 1000));
1196 #endif
1197                         int numSeqs = driver(lines[0], thisoutputFileName, thisFastaName, thisaccnosFileName, thistrimFastaFileName, thisPriority);
1198
1199                         //append files
1200                         m->appendFiles(thisoutputFileName, outputFName); m->mothurRemove(thisoutputFileName);
1201                         m->appendFiles(thisaccnosFileName, accnos); m->mothurRemove(thisaccnosFileName);
1202                         if (trim) { m->appendFiles(thistrimFastaFileName, fasta); m->mothurRemove(thistrimFastaFileName); }
1203                         m->mothurRemove(thisFastaName);
1204
1205                         totalSeqs += numSeqs;
1206
1207                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences from group " + fileGroup[thisFastaName] + ".");     m->mothurOutEndLine();
1208                 }
1209
1210                 return totalSeqs;
1211         }
1212         catch(exception& e) {
1213                 m->errorOut(e, "ChimeraSlayerCommand", "driverGroups");
1214                 exit(1);
1215         }
1216 }
1217 /**************************************************************************************************/
1218 int ChimeraSlayerCommand::createProcessesGroups(string outputFName, string accnos, string fasta, map<string, map<string, int> >& fileToPriority, map<string, string>& fileGroup) {
1219         try {
1220                 int process = 1;
1221                 int num = 0;
1222                 processIDS.clear();
1223
1224                 if (fileToPriority.size() < processors) { processors = fileToPriority.size(); }
1225
1226                 int groupsPerProcessor = fileToPriority.size() / processors;
1227                 int remainder = fileToPriority.size() % processors;
1228
1229                 vector< map<string, map<string, int> > > breakUp;
1230
1231                 for (int i = 0; i < processors; i++) {
1232                         map<string, map<string, int> > thisFileToPriority;
1233                         map<string, map<string, int> >::iterator itFile;
1234                         int count = 0;
1235                         int enough = groupsPerProcessor;
1236                         if (i == 0) { enough = groupsPerProcessor + remainder; }
1237
1238                         for (itFile = fileToPriority.begin(); itFile != fileToPriority.end();) {
1239                                 thisFileToPriority[itFile->first] = itFile->second;
1240                                 fileToPriority.erase(itFile++);
1241                                 count++;
1242                                 if (count == enough) { break; }
1243                         }
1244                         breakUp.push_back(thisFileToPriority);
1245                 }
1246
1247 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1248                 //loop through and create all the processes you want
1249                 while (process != processors) {
1250                         int pid = fork();
1251
1252                         if (pid > 0) {
1253                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1254                                 process++;
1255                         }else if (pid == 0){
1256                                 num = driverGroups(outputFName + toString(getpid()) + ".temp", accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp", breakUp[process], fileGroup);
1257
1258                                 //pass numSeqs to parent
1259                                 ofstream out;
1260                                 string tempFile = outputFName + toString(getpid()) + ".num.temp";
1261                                 m->openOutputFile(tempFile, out);
1262                                 out << num << endl;
1263                                 out.close();
1264                                 exit(0);
1265                         }else {
1266                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1267                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1268                                 exit(0);
1269                         }
1270                 }
1271
1272                 num = driverGroups(outputFName, accnos, fasta, breakUp[0], fileGroup);
1273
1274                 //force parent to wait until all the processes are done
1275                 for (int i=0;i<processors;i++) {
1276                         int temp = processIDS[i];
1277                         wait(&temp);
1278                 }
1279
1280                 for (int i = 0; i < processIDS.size(); i++) {
1281                         ifstream in;
1282                         string tempFile =  outputFName + toString(processIDS[i]) + ".num.temp";
1283                         m->openInputFile(tempFile, in);
1284                         if (!in.eof()) { int tempNum = 0;  in >> tempNum; num += tempNum; }
1285                         in.close(); m->mothurRemove(tempFile);
1286                 }
1287 #else
1288
1289                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1290                 //Windows version shared memory, so be careful when passing variables through the slayerData struct.
1291                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1292                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1293
1294                 vector<slayerData*> pDataArray;
1295                 DWORD   dwThreadIdArray[processors-1];
1296                 HANDLE  hThreadArray[processors-1];
1297
1298                 //Create processor worker threads.
1299                 for(int i=1; i<processors; i++ ){
1300                         string extension = toString(i) + ".temp";
1301                         slayerData* tempslayer = new slayerData((outputFName + extension), (fasta + extension), (accnos + extension), templatefile, search, blastlocation, trimera, trim, realign, m, breakUp[i], fileGroup, ksize, match, mismatch, window, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, divR, priority, i);
1302                         pDataArray.push_back(tempslayer);
1303                         processIDS.push_back(i);
1304
1305                         //MySlayerThreadFunction is in header. It must be global or static to work with the threads.
1306                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1307                         hThreadArray[i-1] = CreateThread(NULL, 0, MySlayerGroupThreadFunction, pDataArray[i-1], 0, &dwThreadIdArray[i-1]);
1308                 }
1309
1310                 num = driverGroups(outputFName, accnos, fasta, breakUp[0], fileGroup);
1311
1312                 //Wait until all threads have terminated.
1313                 WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
1314
1315                 //Close all thread handles and free memory allocations.
1316                 for(int i=0; i < pDataArray.size(); i++){
1317                         num += pDataArray[i]->count;
1318                         CloseHandle(hThreadArray[i]);
1319                         delete pDataArray[i];
1320                 }
1321 #endif
1322
1323                 //append output files
1324                 for(int i=0;i<processIDS.size();i++){
1325                         m->appendFiles((outputFName + toString(processIDS[i]) + ".temp"), outputFName);
1326                         m->mothurRemove((outputFName + toString(processIDS[i]) + ".temp"));
1327
1328                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
1329                         m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));
1330
1331                         if (trim) {
1332                                 m->appendFiles((fasta + toString(processIDS[i]) + ".temp"), fasta);
1333                                 m->mothurRemove((fasta + toString(processIDS[i]) + ".temp"));
1334                         }
1335                 }
1336
1337
1338                 return num;
1339         }
1340         catch(exception& e) {
1341                 m->errorOut(e, "ChimeraSlayerCommand", "createProcessesGroups");
1342                 exit(1);
1343         }
1344 }
1345 //**********************************************************************************************************************
1346
1347 int ChimeraSlayerCommand::driver(linePair filePos, string outputFName, string filename, string accnos, string fasta, map<string, int>& priority){
1348         try {
1349
1350                 Chimera* chimera;
1351                 if (templatefile != "self") { //you want to run slayer with a reference template
1352                         chimera = new ChimeraSlayer(filename, templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
1353                 }else {
1354                         chimera = new ChimeraSlayer(filename, templatefile, trim, priority, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
1355                 }
1356
1357                 if (m->control_pressed) { delete chimera; return 0; }
1358
1359                 if (chimera->getUnaligned()) { delete chimera; m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); m->control_pressed = true; return 0; }
1360                 templateSeqsLength = chimera->getLength();
1361
1362                 ofstream out;
1363                 m->openOutputFile(outputFName, out);
1364
1365                 ofstream out2;
1366                 m->openOutputFile(accnos, out2);
1367
1368                 ofstream out3;
1369                 if (trim) {  m->openOutputFile(fasta, out3); }
1370
1371                 ifstream inFASTA;
1372                 m->openInputFile(filename, inFASTA);
1373
1374                 inFASTA.seekg(filePos.start);
1375
1376                 if (filePos.start == 0) { chimera->printHeader(out); }
1377
1378                 bool done = false;
1379                 int count = 0;
1380
1381                 while (!done) {
1382
1383                         if (m->control_pressed) {       delete chimera; out.close(); out2.close(); if (trim) { out3.close(); } inFASTA.close(); return 1;       }
1384
1385                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
1386                         string candidateAligned = candidateSeq->getAligned();
1387
1388                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
1389                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
1390                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
1391                                 }else{
1392                                         //find chimeras
1393                                         chimera->getChimeras(candidateSeq);
1394
1395                                         if (m->control_pressed) {       delete chimera; delete candidateSeq; return 1;  }
1396
1397                                         //if you are not chimeric, then check each half
1398                                         data_results wholeResults = chimera->getResults();
1399
1400                                         //determine if we need to split
1401                                         bool isChimeric = false;
1402
1403                                         if (wholeResults.flag == "yes") {
1404                                                 string chimeraFlag = "no";
1405                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
1406                                                    ||
1407                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
1408
1409
1410                                                 if (chimeraFlag == "yes") {
1411                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
1412                                                 }
1413                                         }
1414
1415                                         if ((!isChimeric) && trimera) {
1416
1417                                                 //split sequence in half by bases
1418                                                 string leftQuery, rightQuery;
1419                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
1420                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
1421
1422                                                 //run chimeraSlayer on each piece
1423                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
1424                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
1425
1426                                                 //find chimeras
1427                                                 chimera->getChimeras(left);
1428                                                 data_results leftResults = chimera->getResults();
1429
1430                                                 chimera->getChimeras(right);
1431                                                 data_results rightResults = chimera->getResults();
1432
1433                                                 //if either piece is chimeric then report
1434                                                 Sequence trimmed = chimera->print(out, out2, leftResults, rightResults);
1435                                                 if (trim) { trimmed.printSequence(out3);  }
1436
1437                                                 delete left; delete right;
1438
1439                                         }else { //already chimeric
1440                                                 //print results
1441                                                 Sequence trimmed = chimera->print(out, out2);
1442                                                 if (trim) { trimmed.printSequence(out3);  }
1443                                         }
1444
1445
1446                                 }
1447                                 count++;
1448                         }
1449
1450                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1451                                 unsigned long long pos = inFASTA.tellg();
1452                                 if ((pos == -1) || (pos >= filePos.end)) { break; }
1453                         #else
1454                                 if (inFASTA.eof()) { break; }
1455                         #endif
1456
1457                         delete candidateSeq;
1458                         //report progress
1459                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1460                 }
1461                 //report progress
1462                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
1463
1464                 int numNoParents = chimera->getNumNoParents();
1465                 if (numNoParents == count) { m->mothurOut("[WARNING]: megablast returned 0 potential parents for all your sequences. This could be due to formatdb.exe not being setup properly, please check formatdb.log for errors."); m->mothurOutEndLine(); }
1466
1467                 out.close();
1468                 out2.close();
1469                 if (trim) { out3.close(); }
1470                 inFASTA.close();
1471                 delete chimera;
1472
1473                 return count;
1474
1475
1476         }
1477         catch(exception& e) {
1478                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
1479                 exit(1);
1480         }
1481 }
1482 //**********************************************************************************************************************
1483 #ifdef USE_MPI
1484 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, MPI_File& outFastaMPI, vector<unsigned long long>& MPIPos, string filename, map<string, int>& priority, bool byGroup){
1485         try {
1486                 MPI_Status status;
1487                 int pid;
1488                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
1489
1490                 Chimera* chimera;
1491                 if (templatefile != "self") { //you want to run slayer with a reference template
1492                         chimera = new ChimeraSlayer(filename, templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
1493                 }else {
1494                         chimera = new ChimeraSlayer(filename, templatefile, trim, priority, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand(), byGroup);
1495                 }
1496
1497                 if (m->control_pressed) { delete chimera; return 0; }
1498
1499                 if (chimera->getUnaligned()) { delete chimera; m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine(); m->control_pressed = true; return 0; }
1500                 templateSeqsLength = chimera->getLength();
1501
1502                 for(int i=0;i<num;i++){
1503
1504                         if (m->control_pressed) {       delete chimera; return 1;       }
1505
1506                         //read next sequence
1507                         int length = MPIPos[start+i+1] - MPIPos[start+i];
1508
1509                         char* buf4 = new char[length];
1510                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
1511
1512                         string tempBuf = buf4;
1513                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
1514                         istringstream iss (tempBuf,istringstream::in);
1515
1516                         delete buf4;
1517
1518                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
1519                         string candidateAligned = candidateSeq->getAligned();
1520
1521                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
1522
1523                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
1524                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
1525                                 }else{
1526
1527                                         //find chimeras
1528                                         chimera->getChimeras(candidateSeq);
1529
1530                                         if (m->control_pressed) {       delete chimera; delete candidateSeq; return 1;  }
1531
1532                                         //if you are not chimeric, then check each half
1533                                         data_results wholeResults = chimera->getResults();
1534
1535                                         //determine if we need to split
1536                                         bool isChimeric = false;
1537
1538                                         if (wholeResults.flag == "yes") {
1539                                                 string chimeraFlag = "no";
1540                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
1541                                                    ||
1542                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
1543
1544
1545                                                 if (chimeraFlag == "yes") {
1546                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
1547                                                 }
1548                                         }
1549
1550                                         if ((!isChimeric) && trimera) {
1551                                                 //split sequence in half by bases
1552                                                 string leftQuery, rightQuery;
1553                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
1554                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
1555
1556                                                 //run chimeraSlayer on each piece
1557                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
1558                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
1559
1560                                                 //find chimeras
1561                                                 chimera->getChimeras(left);
1562                                                 data_results leftResults = chimera->getResults();
1563
1564                                                 chimera->getChimeras(right);
1565                                                 data_results rightResults = chimera->getResults();
1566
1567                                                 //if either piece is chimeric then report
1568                                                 Sequence trimmed = chimera->print(outMPI, outAccMPI, leftResults, rightResults);
1569                                                 if (trim) {
1570                                                         string outputString = ">" + trimmed.getName() + "\n" + trimmed.getAligned() + "\n";
1571
1572                                                         //write to accnos file
1573                                                         int length = outputString.length();
1574                                                         char* buf2 = new char[length];
1575                                                         memcpy(buf2, outputString.c_str(), length);
1576
1577                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
1578                                                         delete buf2;
1579                                                 }
1580
1581                                                 delete left; delete right;
1582
1583                                         }else {
1584                                                 //print results
1585                                                 Sequence trimmed = chimera->print(outMPI, outAccMPI);
1586
1587                                                 if (trim) {
1588                                                         string outputString = ">" + trimmed.getName() + "\n" + trimmed.getAligned() + "\n";
1589
1590                                                         //write to accnos file
1591                                                         int length = outputString.length();
1592                                                         char* buf2 = new char[length];
1593                                                         memcpy(buf2, outputString.c_str(), length);
1594
1595                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
1596                                                         delete buf2;
1597                                                 }
1598                                         }
1599
1600                                 }
1601                         }
1602                         delete candidateSeq;
1603
1604                         //report progress
1605                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
1606                 }
1607                 //report progress
1608                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
1609
1610                 int numNoParents = chimera->getNumNoParents();
1611                 if (numNoParents == num) { cout << "[WARNING]: megablast returned 0 potential parents for all your sequences. This could be due to formatdb.exe not being setup properly, please check formatdb.log for errors." << endl; }
1612
1613                 delete chimera;
1614                 return 0;
1615         }
1616         catch(exception& e) {
1617                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
1618                 exit(1);
1619         }
1620 }
1621 #endif
1622
1623 /**************************************************************************************************/
1624
1625 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos, string fasta, map<string, int>& thisPriority) {
1626         try {
1627                 int process = 0;
1628                 int num = 0;
1629                 processIDS.clear();
1630
1631 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1632                 //loop through and create all the processes you want
1633                 while (process != processors) {
1634                         int pid = fork();
1635
1636                         if (pid > 0) {
1637                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1638                                 process++;
1639                         }else if (pid == 0){
1640                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp", thisPriority);
1641
1642                                 //pass numSeqs to parent
1643                                 ofstream out;
1644                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
1645                                 m->openOutputFile(tempFile, out);
1646                                 out << num << endl;
1647                                 out.close();
1648                                 exit(0);
1649                         }else {
1650                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1651                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1652                                 exit(0);
1653                         }
1654                 }
1655
1656                 //force parent to wait until all the processes are done
1657                 for (int i=0;i<processors;i++) {
1658                         int temp = processIDS[i];
1659                         wait(&temp);
1660                 }
1661
1662                 for (int i = 0; i < processIDS.size(); i++) {
1663                         ifstream in;
1664                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
1665                         m->openInputFile(tempFile, in);
1666                         if (!in.eof()) { int tempNum = 0;  in >> tempNum; num += tempNum; }
1667                         in.close(); m->mothurRemove(tempFile);
1668                 }
1669 #else
1670
1671                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1672                 //Windows version shared memory, so be careful when passing variables through the slayerData struct.
1673                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1674                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1675
1676                 vector<slayerData*> pDataArray;
1677                 DWORD   dwThreadIdArray[processors];
1678                 HANDLE  hThreadArray[processors];
1679
1680                 //Create processor worker threads.
1681                 for( int i=0; i<processors; i++ ){
1682                         string extension = toString(i) + ".temp";
1683                         slayerData* tempslayer = new slayerData((outputFileName + extension), (fasta + extension), (accnos + extension), filename, templatefile, search, blastlocation, trimera, trim, realign, m, lines[i].start, lines[i].end, ksize, match, mismatch, window, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, divR, priority, i);
1684                         pDataArray.push_back(tempslayer);
1685                         processIDS.push_back(i);
1686
1687                         //MySlayerThreadFunction is in header. It must be global or static to work with the threads.
1688                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1689                         hThreadArray[i] = CreateThread(NULL, 0, MySlayerThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
1690                 }
1691
1692                 //Wait until all threads have terminated.
1693                 WaitForMultipleObjects(processors, hThreadArray, TRUE, INFINITE);
1694
1695                 //Close all thread handles and free memory allocations.
1696                 for(int i=0; i < pDataArray.size(); i++){
1697                         num += pDataArray[i]->count;
1698                         CloseHandle(hThreadArray[i]);
1699                         delete pDataArray[i];
1700                 }
1701 #endif
1702
1703                 rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
1704                 rename((accnos + toString(processIDS[0]) + ".temp").c_str(), accnos.c_str());
1705                 if (trim) {  rename((fasta + toString(processIDS[0]) + ".temp").c_str(), fasta.c_str()); }
1706
1707                 //append output files
1708                 for(int i=1;i<processIDS.size();i++){
1709                         m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
1710                         m->mothurRemove((outputFileName + toString(processIDS[i]) + ".temp"));
1711
1712                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
1713                         m->mothurRemove((accnos + toString(processIDS[i]) + ".temp"));
1714
1715                         if (trim) {
1716                                 m->appendFiles((fasta + toString(processIDS[i]) + ".temp"), fasta);
1717                                 m->mothurRemove((fasta + toString(processIDS[i]) + ".temp"));
1718                         }
1719                 }
1720
1721
1722                 return num;
1723         }
1724         catch(exception& e) {
1725                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
1726                 exit(1);
1727         }
1728 }
1729
1730 /**************************************************************************************************/
1731
1732 int ChimeraSlayerCommand::divideInHalf(Sequence querySeq, string& leftQuery, string& rightQuery) {
1733         try {
1734
1735                 string queryUnAligned = querySeq.getUnaligned();
1736                 int numBases = int(queryUnAligned.length() * 0.5);
1737
1738                 string queryAligned = querySeq.getAligned();
1739                 leftQuery = querySeq.getAligned();
1740                 rightQuery = querySeq.getAligned();
1741
1742                 int baseCount = 0;
1743                 int leftSpot = 0;
1744                 for (int i = 0; i < queryAligned.length(); i++) {
1745                         //if you are a base
1746                         if (isalpha(queryAligned[i])) {
1747                                 baseCount++;
1748                         }
1749
1750                         //if you have half
1751                         if (baseCount >= numBases) {  leftSpot = i; break; } //first half
1752                 }
1753
1754                 //blank out right side
1755                 for (int i = leftSpot; i < leftQuery.length(); i++) { leftQuery[i] = '.'; }
1756
1757                 //blank out left side
1758                 for (int i = 0; i < leftSpot; i++) { rightQuery[i] = '.'; }
1759
1760                 return 0;
1761
1762         }
1763         catch(exception& e) {
1764                 m->errorOut(e, "ChimeraSlayerCommand", "divideInHalf");
1765                 exit(1);
1766         }
1767 }
1768 /**************************************************************************************************/
1769 map<string, int> ChimeraSlayerCommand::sortFastaFile(string fastaFile, string nameFile) {
1770         try {
1771                 map<string, int> nameAbund;
1772
1773                 //read through fastafile and store info
1774                 map<string, string> seqs;
1775                 ifstream in;
1776                 m->openInputFile(fastaFile, in);
1777
1778                 while (!in.eof()) {
1779
1780                         if (m->control_pressed) { in.close(); return nameAbund; }
1781
1782                         Sequence seq(in); m->gobble(in);
1783                         seqs[seq.getName()] = seq.getAligned();
1784                 }
1785
1786                 in.close();
1787
1788                 //read namefile
1789                 vector<seqPriorityNode> nameMapCount;
1790                 int error = m->readNames(nameFile, nameMapCount, seqs);
1791
1792                 if (m->control_pressed) { return nameAbund; }
1793
1794                 if (error == 1) { m->control_pressed = true; return nameAbund; }
1795                 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; }
1796
1797                 sort(nameMapCount.begin(), nameMapCount.end(), compareSeqPriorityNodes);
1798
1799                 string newFasta = fastaFile + ".temp";
1800                 ofstream out;
1801                 m->openOutputFile(newFasta, out);
1802
1803                 //print new file in order of
1804                 for (int i = 0; i < nameMapCount.size(); i++) {
1805                         out << ">" << nameMapCount[i].name << endl << nameMapCount[i].seq << endl;
1806                         nameAbund[nameMapCount[i].name] = nameMapCount[i].numIdentical;
1807                 }
1808                 out.close();
1809
1810                 rename(newFasta.c_str(), fastaFile.c_str());
1811
1812                 return nameAbund;
1813
1814         }
1815         catch(exception& e) {
1816                 m->errorOut(e, "ChimeraSlayerCommand", "sortFastaFile");
1817                 exit(1);
1818         }
1819 }
1820 /**************************************************************************************************/
1821 map<string, int> ChimeraSlayerCommand::sortFastaFile(vector<Sequence>& thisseqs, map<string, string>& nameMap, string newFile) {
1822         try {
1823                 map<string, int> nameAbund;
1824                 vector<seqPriorityNode> nameVector;
1825
1826                 //read through fastafile and store info
1827                 map<string, string> seqs;
1828
1829                 for (int i = 0; i < thisseqs.size(); i++) {
1830
1831                         if (m->control_pressed) { return nameAbund; }
1832
1833                         map<string, string>::iterator itNameMap = nameMap.find(thisseqs[i].getName());
1834
1835                         if (itNameMap == nameMap.end()){
1836                                 m->control_pressed = true;
1837                                 m->mothurOut("[ERROR]: " + thisseqs[i].getName() + " is in your fastafile, but is not in your namesfile, please correct."); m->mothurOutEndLine();
1838                         }else {
1839                                 int num = m->getNumNames(itNameMap->second);
1840
1841                                 seqPriorityNode temp(num, thisseqs[i].getAligned(), thisseqs[i].getName());
1842                                 nameVector.push_back(temp);
1843                         }
1844                 }
1845
1846                 //sort by num represented
1847                 sort(nameVector.begin(), nameVector.end(), compareSeqPriorityNodes);
1848
1849                 if (m->control_pressed) { return nameAbund; }
1850
1851                 if (thisseqs.size() != nameVector.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; }
1852
1853                 ofstream out;
1854                 m->openOutputFile(newFile, out);
1855
1856                 //print new file in order of
1857                 for (int i = 0; i < nameVector.size(); i++) {
1858                         out << ">" << nameVector[i].name << endl << nameVector[i].seq << endl;
1859                         nameAbund[nameVector[i].name] = nameVector[i].numIdentical;
1860                 }
1861                 out.close();
1862
1863                 return nameAbund;
1864
1865         }
1866         catch(exception& e) {
1867                 m->errorOut(e, "ChimeraSlayerCommand", "sortFastaFile");
1868                 exit(1);
1869         }
1870 }
1871 /**************************************************************************************************/
1872