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
  14 //**********************************************************************************************************************
  15 vector<string> ChimeraSlayerCommand::setParameters(){
  16         try {
  17                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  19                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  20                 CommandParameter pwindow("window", "Number", "", "50", "", "", "",false,false); parameters.push_back(pwindow);
  21                 CommandParameter pksize("ksize", "Number", "", "7", "", "", "",false,false); parameters.push_back(pksize);
  22                 CommandParameter pmatch("match", "Number", "", "5.0", "", "", "",false,false); parameters.push_back(pmatch);
  23                 CommandParameter pmismatch("mismatch", "Number", "", "-4.0", "", "", "",false,false); parameters.push_back(pmismatch);
  24                 CommandParameter pminsim("minsim", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminsim);
  25                 CommandParameter pmincov("mincov", "Number", "", "70", "", "", "",false,false); parameters.push_back(pmincov);
  26                 CommandParameter pminsnp("minsnp", "Number", "", "10", "", "", "",false,false); parameters.push_back(pminsnp);
  27                 CommandParameter pminbs("minbs", "Number", "", "90", "", "", "",false,false); parameters.push_back(pminbs);
  28                 CommandParameter psearch("search", "Multiple", "kmer-blast", "blast", "", "", "",false,false); parameters.push_back(psearch);
  29                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  30                 CommandParameter prealign("realign", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(prealign);
  31                 CommandParameter ptrim("trim", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(ptrim);
  32                 CommandParameter psplit("split", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psplit);
  33                 CommandParameter pnumwanted("numwanted", "Number", "", "15", "", "", "",false,false); parameters.push_back(pnumwanted);
  34                 CommandParameter piters("iters", "Number", "", "1000", "", "", "",false,false); parameters.push_back(piters);
  35                 CommandParameter pdivergence("divergence", "Number", "", "1.007", "", "", "",false,false); parameters.push_back(pdivergence);
  36                 CommandParameter pparents("parents", "Number", "", "3", "", "", "",false,false); parameters.push_back(pparents);
  37                 CommandParameter pincrement("increment", "Number", "", "5", "", "", "",false,false); parameters.push_back(pincrement);
  38                 CommandParameter pblastlocation("blastlocation", "String", "", "", "", "", "",false,false); parameters.push_back(pblastlocation);
  39                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  40                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  41                 CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
  42
  43                 vector<string> myArray;
  44                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  45                 return myArray;
  46         }
  47         catch(exception& e) {
  48                 m->errorOut(e, "ChimeraSlayerCommand", "setParameters");
  49                 exit(1);
  50         }
  51 }
  52 //**********************************************************************************************************************
  53 string ChimeraSlayerCommand::getHelpString(){
  54         try {
  55                 string helpString = "";
  56                 helpString += "The chimera.slayer command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  57                 helpString += "This command was modeled after the chimeraSlayer written by the Broad Institute.\n";
  58                 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";
  59                 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";
  60                 helpString += "The name parameter allows you to provide a name file, if you are using reference=self. \n";
  61                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  62                 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";
  63                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  64 #ifdef USE_MPI
  65                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  66 #endif
  67                 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";
  68                 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";
  69                 helpString += "The window parameter allows you to specify the window size for searching for chimeras, default=50. \n";
  70                 helpString += "The increment parameter allows you to specify how far you move each window while finding chimeric sequences, default=5.\n";
  71                 helpString += "The numwanted parameter allows you to specify how many sequences you would each query sequence compared with, default=15.\n";
  72                 helpString += "The ksize parameter allows you to input kmersize, default is 7, used if search is kmer. \n";
  73                 helpString += "The match parameter allows you to reward matched bases in blast search, default is 5. \n";
  74                 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";
  75                 helpString += "The mismatch parameter allows you to penalize mismatched bases in blast search, default is -4. \n";
  76                 helpString += "The divergence parameter allows you to set a cutoff for chimera determination, default is 1.007. \n";
  77                 helpString += "The iters parameter allows you to specify the number of bootstrap iters to do with the chimeraslayer method, default=1000.\n";
  78                 helpString += "The minsim parameter allows you to specify a minimum similarity with the parent fragments, default=90. \n";
  79                 helpString += "The mincov parameter allows you to specify minimum coverage by closest matches found in template. Default is 70, meaning 70%. \n";
  80                 helpString += "The minbs parameter allows you to specify minimum bootstrap support for calling a sequence chimeric. Default is 90, meaning 90%. \n";
  81                 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";
  82                 helpString += "The search parameter allows you to specify search method for finding the closest parent. Choices are blast, and kmer, default blast. \n";
  83                 helpString += "The realign parameter allows you to realign the query to the potential parents. Choices are true or false, default true.  \n";
  84                 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";
  85                 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.";
  86                 helpString += "The chimera.slayer command should be in the following format: \n";
  87                 helpString += "chimera.slayer(fasta=yourFastaFile, reference=yourTemplate, search=yourSearch) \n";
  88                 helpString += "Example: chimera.slayer(fasta=AD.align, reference=core_set_aligned.imputed.fasta, search=kmer) \n";
  89                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile).\n";
  90                 return helpString;
  91         }
  92         catch(exception& e) {
  93                 m->errorOut(e, "ChimeraSlayerCommand", "getHelpString");
  94                 exit(1);
  95         }
  96 }
  97 //**********************************************************************************************************************
  98 ChimeraSlayerCommand::ChimeraSlayerCommand(){
  99         try {
 100                 abort = true; calledHelp = true;
 101                 setParameters();
 102                 vector<string> tempOutNames;
 103                 outputTypes["chimera"] = tempOutNames;
 104                 outputTypes["accnos"] = tempOutNames;
 105                 outputTypes["fasta"] = tempOutNames;
 106         }
 107         catch(exception& e) {
 108                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 109                 exit(1);
 110         }
 111 }
 112 //***************************************************************************************************************
 113 ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
 114         try {
 115                 abort = false; calledHelp = false;
 116                 ReferenceDB* rdb = ReferenceDB::getInstance();
 117
 118                 //allow user to run help
 119                 if(option == "help") { help(); abort = true; calledHelp = true; }
 120                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 121
 122                 else {
 123                         vector<string> myArray = setParameters();
 124
 125                         OptionParser parser(option);
 126                         map<string,string> parameters = parser.getParameters();
 127
 128                         ValidParameters validParameter("chimera.slayer");
 129                         map<string,string>::iterator it;
 130
 131                         //check to make sure all parameters are valid for command
 132                         for (it = parameters.begin(); it != parameters.end(); it++) {
 133                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 134                         }
 135
 136                         vector<string> tempOutNames;
 137                         outputTypes["chimera"] = tempOutNames;
 138                         outputTypes["accnos"] = tempOutNames;
 139                         outputTypes["fasta"] = tempOutNames;
 140
 141                         //if the user changes the input directory command factory will send this info to us in the output parameter
 142                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 143                         if (inputDir == "not found"){   inputDir = "";          }
 144
 145                         //check for required parameters
 146                         fastafile = validParameter.validFile(parameters, "fasta", false);
 147                         if (fastafile == "not found") {
 148                                 //if there is a current fasta file, use it
 149                                 string filename = m->getFastaFile();
 150                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 151                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 152                         }else {
 153                                 m->splitAtDash(fastafile, fastaFileNames);
 154
 155                                 //go through files and make sure they are good, if not, then disregard them
 156                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 157
 158                                         bool ignore = false;
 159                                         if (fastaFileNames[i] == "current") {
 160                                                 fastaFileNames[i] = m->getFastaFile();
 161                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 162                                                 else {
 163                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 164                                                         //erase from file list
 165                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 166                                                         i--;
 167                                                 }
 168                                         }
 169
 170                                         if (!ignore) {
 171
 172                                                 if (inputDir != "") {
 173                                                         string path = m->hasPath(fastaFileNames[i]);
 174                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 175                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 176                                                 }
 177
 178                                                 int ableToOpen;
 179                                                 ifstream in;
 180
 181                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 182
 183                                                 //if you can't open it, try default location
 184                                                 if (ableToOpen == 1) {
 185                                                         if (m->getDefaultPath() != "") { //default path is set
 186                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 187                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 188                                                                 ifstream in2;
 189                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 190                                                                 in2.close();
 191                                                                 fastaFileNames[i] = tryPath;
 192                                                         }
 193                                                 }
 194
 195                                                 if (ableToOpen == 1) {
 196                                                         if (m->getOutputDir() != "") { //default path is set
 197                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 198                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 199                                                                 ifstream in2;
 200                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 201                                                                 in2.close();
 202                                                                 fastaFileNames[i] = tryPath;
 203                                                         }
 204                                                 }
 205
 206                                                 in.close();
 207
 208                                                 if (ableToOpen == 1) {
 209                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 210                                                         //erase from file list
 211                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 212                                                         i--;
 213                                                 }else {
 214                                                         m->setFastaFile(fastaFileNames[i]);
 215                                                 }
 216                                         }
 217                                 }
 218
 219                                 //make sure there is at least one valid file left
 220                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 221                         }
 222
 223
 224                         //check for required parameters
 225                         bool hasName = true;
 226                         namefile = validParameter.validFile(parameters, "name", false);
 227                         if (namefile == "not found") { namefile = "";  hasName = false; }
 228                         else {
 229                                 m->splitAtDash(namefile, nameFileNames);
 230
 231                                 //go through files and make sure they are good, if not, then disregard them
 232                                 for (int i = 0; i < nameFileNames.size(); i++) {
 233
 234                                         bool ignore = false;
 235                                         if (nameFileNames[i] == "current") {
 236                                                 nameFileNames[i] = m->getNameFile();
 237                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 238                                                 else {
 239                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 240                                                         //erase from file list
 241                                                         nameFileNames.erase(nameFileNames.begin()+i);
 242                                                         i--;
 243                                                 }
 244                                         }
 245
 246                                         if (!ignore) {
 247
 248                                                 if (inputDir != "") {
 249                                                         string path = m->hasPath(nameFileNames[i]);
 250                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 251                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 252                                                 }
 253
 254                                                 int ableToOpen;
 255                                                 ifstream in;
 256
 257                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 258
 259                                                 //if you can't open it, try default location
 260                                                 if (ableToOpen == 1) {
 261                                                         if (m->getDefaultPath() != "") { //default path is set
 262                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 263                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 264                                                                 ifstream in2;
 265                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 266                                                                 in2.close();
 267                                                                 nameFileNames[i] = tryPath;
 268                                                         }
 269                                                 }
 270
 271                                                 if (ableToOpen == 1) {
 272                                                         if (m->getOutputDir() != "") { //default path is set
 273                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 274                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 275                                                                 ifstream in2;
 276                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 277                                                                 in2.close();
 278                                                                 nameFileNames[i] = tryPath;
 279                                                         }
 280                                                 }
 281
 282                                                 in.close();
 283
 284                                                 if (ableToOpen == 1) {
 285                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 286                                                         //erase from file list
 287                                                         nameFileNames.erase(nameFileNames.begin()+i);
 288                                                         i--;
 289                                                 }else {
 290                                                         m->setNameFile(nameFileNames[i]);
 291                                                 }
 292                                         }
 293                                 }
 294
 295                                 //make sure there is at least one valid file left
 296                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 297                         }
 298
 299                         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; }
 300
 301                         //if the user changes the output directory command factory will send this info to us in the output parameter
 302                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 303
 304                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 305                         m->setProcessors(temp);
 306                         convert(temp, processors);
 307
 308                         temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
 309                         save = m->isTrue(temp);
 310                         rdb->save = save;
 311                         if (save) { //clear out old references
 312                                 rdb->clearMemory();
 313                         }
 314
 315                         string path;
 316                         it = parameters.find("reference");
 317                         //user has given a template file
 318                         if(it != parameters.end()){
 319                                 if (it->second == "self") {
 320                                         templatefile = "self";
 321                                         if (save) {
 322                                                 m->mothurOut("[WARNING]: You can't save reference=self, ignoring save.");
 323                                                 m->mothurOutEndLine();
 324                                                 save = false;
 325                                         }
 326                                 }
 327                                 else {
 328                                         path = m->hasPath(it->second);
 329                                         //if the user has not given a path then, add inputdir. else leave path alone.
 330                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 331
 332                                         templatefile = validParameter.validFile(parameters, "reference", true);
 333                                         if (templatefile == "not open") { abort = true; }
 334                                         else if (templatefile == "not found") { //check for saved reference sequences
 335                                                 if (rdb->referenceSeqs.size() != 0) {
 336                                                         templatefile = "saved";
 337                                                 }else {
 338                                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 339                                                         m->mothurOutEndLine();
 340                                                         abort = true;
 341                                                 }
 342                                         }else { if (save) {     rdb->setSavedReference(templatefile);   }       }
 343                                 }
 344                         }else if (hasName) {  templatefile = "self";
 345                                 if (save) {
 346                                         m->mothurOut("[WARNING]: You can't save reference=self, ignoring save.");
 347                                         m->mothurOutEndLine();
 348                                         save = false;
 349                                 }
 350                         }
 351                         else {
 352                                 if (rdb->referenceSeqs.size() != 0) {
 353                                         templatefile = "saved";
 354                                 }else {
 355                                         m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
 356                                         m->mothurOutEndLine();
 357                                         templatefile = ""; abort = true;
 358                                 }
 359                         }
 360
 361
 362
 363                         temp = validParameter.validFile(parameters, "ksize", false);                    if (temp == "not found") { temp = "7"; }
 364                         convert(temp, ksize);
 365
 366                         temp = validParameter.validFile(parameters, "window", false);                   if (temp == "not found") { temp = "50"; }
 367                         convert(temp, window);
 368
 369                         temp = validParameter.validFile(parameters, "match", false);                    if (temp == "not found") { temp = "5"; }
 370                         convert(temp, match);
 371
 372                         temp = validParameter.validFile(parameters, "mismatch", false);                 if (temp == "not found") { temp = "-4"; }
 373                         convert(temp, mismatch);
 374
 375                         temp = validParameter.validFile(parameters, "divergence", false);               if (temp == "not found") { temp = "1.007"; }
 376                         convert(temp, divR);
 377
 378                         temp = validParameter.validFile(parameters, "minsim", false);                   if (temp == "not found") { temp = "90"; }
 379                         convert(temp, minSimilarity);
 380
 381                         temp = validParameter.validFile(parameters, "mincov", false);                   if (temp == "not found") { temp = "70"; }
 382                         convert(temp, minCoverage);
 383
 384                         temp = validParameter.validFile(parameters, "minbs", false);                    if (temp == "not found") { temp = "90"; }
 385                         convert(temp, minBS);
 386
 387                         temp = validParameter.validFile(parameters, "minsnp", false);                   if (temp == "not found") { temp = "10"; }
 388                         convert(temp, minSNP);
 389
 390                         temp = validParameter.validFile(parameters, "parents", false);                  if (temp == "not found") { temp = "3"; }
 391                         convert(temp, parents);
 392
 393                         temp = validParameter.validFile(parameters, "realign", false);                  if (temp == "not found") { temp = "t"; }
 394                         realign = m->isTrue(temp);
 395
 396                         temp = validParameter.validFile(parameters, "trim", false);                             if (temp == "not found") { temp = "f"; }
 397                         trim = m->isTrue(temp);
 398
 399                         temp = validParameter.validFile(parameters, "split", false);                    if (temp == "not found") { temp = "f"; }
 400                         trimera = m->isTrue(temp);
 401
 402                         search = validParameter.validFile(parameters, "search", false);                 if (search == "not found") { search = "blast"; }
 403
 404                         temp = validParameter.validFile(parameters, "iters", false);                    if (temp == "not found") { temp = "1000"; }
 405                         convert(temp, iters);
 406
 407                         temp = validParameter.validFile(parameters, "increment", false);                if (temp == "not found") { temp = "5"; }
 408                         convert(temp, increment);
 409
 410                         temp = validParameter.validFile(parameters, "numwanted", false);                if (temp == "not found") { temp = "15"; }
 411                         convert(temp, numwanted);
 412
 413                         blastlocation = validParameter.validFile(parameters, "blastlocation", false);
 414                         if (blastlocation == "not found") { blastlocation = ""; }
 415                         else {
 416                                 //add / to name if needed
 417                                 string lastChar = blastlocation.substr(blastlocation.length()-1);
 418 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 419                                 if (lastChar != "/") { blastlocation += "/"; }
 420 #else
 421                                 if (lastChar != "\\") { blastlocation += "\\"; }
 422 #endif
 423                                 blastlocation = m->getFullPathName(blastlocation);
 424                                 string formatdbCommand = "";
 425 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 426                                 formatdbCommand = blastlocation + "formatdb";
 427 #else
 428                                 formatdbCommand = blastlocation + "formatdb.exe";
 429 #endif
 430
 431                                 //test to make sure formatdb exists
 432                                 ifstream in;
 433                                 formatdbCommand = m->getFullPathName(formatdbCommand);
 434                                 int ableToOpen = m->openInputFile(formatdbCommand, in, "no error"); in.close();
 435                                 if(ableToOpen == 1) {   m->mothurOut("[ERROR]: " + formatdbCommand + " file does not exist. mothur requires formatdb.exe to run chimera.slayer."); m->mothurOutEndLine(); abort = true; }
 436
 437                                 string blastCommand = "";
 438 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 439                                 blastCommand = blastlocation + "megablast";
 440 #else
 441                                 blastCommand = blastlocation + "megablast.exe";
 442 #endif
 443                                 //test to make sure formatdb exists
 444                                 ifstream in2;
 445                                 blastCommand = m->getFullPathName(blastCommand);
 446                                 ableToOpen = m->openInputFile(blastCommand, in2, "no error"); in2.close();
 447                                 if(ableToOpen == 1) {   m->mothurOut("[ERROR]: " + blastCommand + " file does not exist. mothur requires blastall.exe to run chimera.slayer."); m->mothurOutEndLine(); abort = true; }
 448                         }
 449
 450                         if ((search != "blast") && (search != "kmer")) { m->mothurOut(search + " is not a valid search."); m->mothurOutEndLine(); abort = true;  }
 451
 452                         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; }
 453                 }
 454         }
 455         catch(exception& e) {
 456                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 457                 exit(1);
 458         }
 459 }
 460 //***************************************************************************************************************
 461
 462 int ChimeraSlayerCommand::execute(){
 463         try{
 464                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 465
 466                 for (int s = 0; s < fastaFileNames.size(); s++) {
 467
 468                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 469
 470                         int start = time(NULL);
 471
 472                         if (templatefile == "self") {
 473                                 if (processors != 1) { m->mothurOut("When using template=self, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 474                                 string nameFile = "";
 475                                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 476                                         nameFile = nameFileNames[s];
 477                                 }else {
 478                                         m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 479
 480                                         //use unique.seqs to create new name and fastafile
 481                                         string inputString = "fasta=" + fastaFileNames[s];
 482                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 483                                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 484
 485                                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 486                                         uniqueCommand->execute();
 487
 488                                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 489
 490                                         delete uniqueCommand;
 491
 492                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 493
 494                                         nameFile = filenames["name"][0];
 495                                         fastaFileNames[s] = filenames["fasta"][0];
 496                                 }
 497
 498                                 //sort fastafile by abundance, returns new sorted fastafile name
 499                                 m->mothurOut("Sorting fastafile according to abundance..."); cout.flush();
 500                                 priority = sortFastaFile(fastaFileNames[s], nameFile);
 501                                 m->mothurOut("Done."); m->mothurOutEndLine();
 502
 503                                 if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       m->mothurRemove(outputNames[j]);        }  return 0;    }
 504                         }
 505
 506                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 507                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "slayer.chimera";
 508                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.accnos";
 509                         string trimFastaFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "slayer.fasta";
 510
 511                         //create chimera here if you are mac or linux because fork will copy for you. Create in create processes instead if you are windows.
 512                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 513                         if (templatefile != "self") { //you want to run slayer with a reference template
 514                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
 515                         }else {
 516                                 chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, priority, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
 517                         }
 518
 519                         if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;    }
 520
 521                         if (chimera->getUnaligned()) {
 522                                 m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 523                                 delete chimera;
 524                                 return 0;
 525                         }
 526                         templateSeqsLength = chimera->getLength();
 527                         #else
 528                         if (processors == 1) {
 529                                 if (templatefile != "self") { //you want to run slayer with a reference template
 530                                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
 531                                 }else {
 532                                         chimera = new ChimeraSlayer(fastaFileNames[s], templatefile, trim, priority, search, ksize, match, mismatch, window, divR, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, realign, blastlocation, rand());
 533                                 }
 534
 535                                 if (m->control_pressed) { delete chimera; for (int j = 0; j < outputNames.size(); j++) {        m->mothurRemove(outputNames[j]);        }  return 0;    }
 536
 537                                 if (chimera->getUnaligned()) {
 538                                         m->mothurOut("Your template sequences are different lengths, please correct."); m->mothurOutEndLine();
 539                                         delete chimera;
 540                                         return 0;
 541                                 }
 542                                 templateSeqsLength = chimera->getLength();
 543                         }
 544                         #endif
 545
 546                 #ifdef USE_MPI
 547                         int pid, numSeqsPerProcessor;
 548                                 int tag = 2001;
 549                                 vector<unsigned long int> MPIPos;
 550
 551                                 MPI_Status status;
 552                                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 553                                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 554
 555                                 MPI_File inMPI;
 556                                 MPI_File outMPI;
 557                                 MPI_File outMPIAccnos;
 558                                 MPI_File outMPIFasta;
 559
 560                                 int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
 561                                 int inMode=MPI_MODE_RDONLY;
 562
 563                                 char outFilename[1024];
 564                                 strcpy(outFilename, outputFileName.c_str());
 565
 566                                 char outAccnosFilename[1024];
 567                                 strcpy(outAccnosFilename, accnosFileName.c_str());
 568
 569                                 char outFastaFilename[1024];
 570                                 strcpy(outFastaFilename, trimFastaFileName.c_str());
 571
 572                                 char inFileName[1024];
 573                                 strcpy(inFileName, fastaFileNames[s].c_str());
 574
 575                                 MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI);  //comm, filename, mode, info, filepointer
 576                                 MPI_File_open(MPI_COMM_WORLD, outFilename, outMode, MPI_INFO_NULL, &outMPI);
 577                                 MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
 578                                 if (trim) { MPI_File_open(MPI_COMM_WORLD, outFastaFilename, outMode, MPI_INFO_NULL, &outMPIFasta); }
 579
 580                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) {  MPI_File_close(&outMPIFasta);  } MPI_File_close(&outMPIAccnos); for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }   delete chimera; return 0;  }
 581
 582                                 if (pid == 0) { //you are the root process
 583                                         m->mothurOutEndLine();
 584                                         m->mothurOut("Only reporting sequence supported by " + toString(minBS) + "% of bootstrapped results.");
 585                                         m->mothurOutEndLine();
 586
 587                                         string outTemp = "Name\tLeftParent\tRightParent\tDivQLAQRB\tPerIDQLAQRB\tBootStrapA\tDivQLBQRA\tPerIDQLBQRA\tBootStrapB\tFlag\tLeftWindow\tRightWindow\n";
 588
 589                                         //print header
 590                                         int length = outTemp.length();
 591                                         char* buf2 = new char[length];
 592                                         memcpy(buf2, outTemp.c_str(), length);
 593
 594                                         MPI_File_write_shared(outMPI, buf2, length, MPI_CHAR, &status);
 595                                         delete buf2;
 596
 597                                         MPIPos = m->setFilePosFasta(fastaFileNames[s], numSeqs); //fills MPIPos, returns numSeqs
 598
 599                                         if (templatefile != "self") { //if template=self we can only use 1 processor
 600                                                 //send file positions to all processes
 601                                                 for(int i = 1; i < processors; i++) {
 602                                                         MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
 603                                                         MPI_Send(&MPIPos[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
 604                                                 }
 605                                         }
 606                                         //figure out how many sequences you have to align
 607                                         numSeqsPerProcessor = numSeqs / processors;
 608                                         int startIndex =  pid * numSeqsPerProcessor;
 609                                         if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 610
 611                                         if (templatefile == "self") { //if template=self we can only use 1 processor
 612                                                 startIndex = 0;
 613                                                 numSeqsPerProcessor = numSeqs;
 614                                         }
 615
 616                                         //do your part
 617                                         driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 618
 619                                         int numNoParents = chimera->getNumNoParents();
 620                                         int temp;
 621                                         for(int i = 1; i < processors; i++) {
 622                                                 MPI_Recv(&temp, 1, MPI_INT, 1, tag, MPI_COMM_WORLD, &status);
 623                                                 numNoParents += temp;
 624                                         }
 625
 626
 627                                         if (numSeqs == numNoParents) {  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(); }
 628
 629                                         if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  m->mothurRemove(outputFileName);  m->mothurRemove(accnosFileName);  delete chimera; return 0;  }
 630
 631                                 }else{ //you are a child process
 632                                         if (templatefile != "self") { //if template=self we can only use 1 processor
 633                                                 MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
 634                                                 MPIPos.resize(numSeqs+1);
 635                                                 MPI_Recv(&MPIPos[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
 636
 637                                                 //figure out how many sequences you have to align
 638                                                 numSeqsPerProcessor = numSeqs / processors;
 639                                                 int startIndex =  pid * numSeqsPerProcessor;
 640                                                 if(pid == (processors - 1)){    numSeqsPerProcessor = numSeqs - pid * numSeqsPerProcessor;      }
 641
 642                                                 //do your part
 643                                                 driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPI, outMPIAccnos, outMPIFasta, MPIPos);
 644
 645                                                 int numNoParents = chimera->getNumNoParents();
 646                                                 MPI_Send(&numNoParents, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 647
 648                                                 if (m->control_pressed) { outputTypes.clear();  MPI_File_close(&inMPI);  MPI_File_close(&outMPI); if (trim) { MPI_File_close(&outMPIFasta); }  MPI_File_close(&outMPIAccnos);  for (int j = 0; j < outputNames.size(); j++) {   m->mothurRemove(outputNames[j]);        }  delete chimera; return 0;  }
 649
 650                                         }
 651                                 }
 652
 653                                 //close files
 654                                 MPI_File_close(&inMPI);
 655                                 MPI_File_close(&outMPI);
 656                                 MPI_File_close(&outMPIAccnos);
 657                                 if (trim) { MPI_File_close(&outMPIFasta); }
 658                                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 659
 660                 #else
 661                         //break up file
 662                         vector<unsigned long int> positions;
 663 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 664                         positions = m->divideFile(fastaFileNames[s], processors);
 665                         for (int i = 0; i < (positions.size()-1); i++) {        lines.push_back(new linePair(positions[i], positions[(i+1)]));  }
 666 #else
 667                         if (processors == 1) {
 668                                 lines.push_back(new linePair(0, 1000));
 669                         }else {
 670                                 positions = m->setFilePosFasta(fastaFileNames[s], numSeqs);
 671
 672                                 //figure out how many sequences you have to process
 673                                 int numSeqsPerProcessor = numSeqs / processors;
 674                                 for (int i = 0; i < processors; i++) {
 675                                         int startIndex =  i * numSeqsPerProcessor;
 676                                         if(i == (processors - 1)){      numSeqsPerProcessor = numSeqs - i * numSeqsPerProcessor;        }
 677                                         lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
 678                                 }
 679                         }
 680 #endif
 681
 682                         if(processors == 1){
 683                                 numSeqs = driver(lines[0], outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 684
 685                                 int numNoParents = chimera->getNumNoParents();
 686                                 if (numNoParents == numSeqs) { 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(); }
 687
 688                                 if (m->control_pressed) { 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]);        } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 689
 690                         }else{
 691                                 processIDS.resize(0);
 692
 693                                 numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName, trimFastaFileName);
 694
 695                                 rename((outputFileName + toString(processIDS[0]) + ".temp").c_str(), outputFileName.c_str());
 696                                 rename((accnosFileName + toString(processIDS[0]) + ".temp").c_str(), accnosFileName.c_str());
 697                                 if (trim) {  rename((trimFastaFileName + toString(processIDS[0]) + ".temp").c_str(), trimFastaFileName.c_str()); }
 698
 699                                 //append output files
 700                                 for(int i=1;i<processIDS.size();i++){
 701                                         m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 702                                         m->mothurRemove((outputFileName + toString(processIDS[i]) + ".temp"));
 703
 704                                         m->appendFiles((accnosFileName + toString(processIDS[i]) + ".temp"), accnosFileName);
 705                                         m->mothurRemove((accnosFileName + toString(processIDS[i]) + ".temp"));
 706
 707                                         if (trim) {
 708                                                 m->appendFiles((trimFastaFileName + toString(processIDS[i]) + ".temp"), trimFastaFileName);
 709                                                 m->mothurRemove((trimFastaFileName + toString(processIDS[i]) + ".temp"));
 710                                         }
 711                                 }
 712
 713                                 if (m->control_pressed) { 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]);        } for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear(); delete chimera; return 0; }
 714                         }
 715
 716
 717                 #endif
 718
 719                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 720                         delete chimera;
 721                         #endif
 722
 723                         for (int i = 0; i < lines.size(); i++) {  delete lines[i];  }  lines.clear();
 724
 725                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 726                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 727                         if (trim) {  outputNames.push_back(trimFastaFileName); outputTypes["fasta"].push_back(trimFastaFileName); }
 728
 729                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 730                 }
 731
 732                 //set accnos file as new current accnosfile
 733                 string current = "";
 734                 itTypes = outputTypes.find("accnos");
 735                 if (itTypes != outputTypes.end()) {
 736                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 737                 }
 738
 739                 if (trim) {
 740                         itTypes = outputTypes.find("fasta");
 741                         if (itTypes != outputTypes.end()) {
 742                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 743                         }
 744                 }
 745
 746                 m->mothurOutEndLine();
 747                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 748                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 749                 m->mothurOutEndLine();
 750
 751                 return 0;
 752
 753         }
 754         catch(exception& e) {
 755                 m->errorOut(e, "ChimeraSlayerCommand", "execute");
 756                 exit(1);
 757         }
 758 }
 759 //**********************************************************************************************************************
 760
 761 int ChimeraSlayerCommand::driver(linePair* filePos, string outputFName, string filename, string accnos, string fasta){
 762         try {
 763                 ofstream out;
 764                 m->openOutputFile(outputFName, out);
 765
 766                 ofstream out2;
 767                 m->openOutputFile(accnos, out2);
 768
 769                 ofstream out3;
 770                 if (trim) {  m->openOutputFile(fasta, out3); }
 771
 772                 ifstream inFASTA;
 773                 m->openInputFile(filename, inFASTA);
 774
 775                 inFASTA.seekg(filePos->start);
 776
 777                 if (filePos->start == 0) { chimera->printHeader(out); }
 778
 779                 bool done = false;
 780                 int count = 0;
 781
 782                 while (!done) {
 783
 784                         if (m->control_pressed) {       out.close(); out2.close(); if (trim) { out3.close(); } inFASTA.close(); return 1;       }
 785
 786                         Sequence* candidateSeq = new Sequence(inFASTA);  m->gobble(inFASTA);
 787                         string candidateAligned = candidateSeq->getAligned();
 788
 789                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 790                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 791                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 792                                 }else{
 793                                         //find chimeras
 794                                         chimera->getChimeras(candidateSeq);
 795
 796                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 797
 798                                         //if you are not chimeric, then check each half
 799                                         data_results wholeResults = chimera->getResults();
 800
 801                                         //determine if we need to split
 802                                         bool isChimeric = false;
 803
 804                                         if (wholeResults.flag == "yes") {
 805                                                 string chimeraFlag = "no";
 806                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
 807                                                    ||
 808                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
 809
 810
 811                                                 if (chimeraFlag == "yes") {
 812                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
 813                                                 }
 814                                         }
 815
 816                                         if ((!isChimeric) && trimera) {
 817
 818                                                 //split sequence in half by bases
 819                                                 string leftQuery, rightQuery;
 820                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
 821                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
 822
 823                                                 //run chimeraSlayer on each piece
 824                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
 825                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
 826
 827                                                 //find chimeras
 828                                                 chimera->getChimeras(left);
 829                                                 data_results leftResults = chimera->getResults();
 830
 831                                                 chimera->getChimeras(right);
 832                                                 data_results rightResults = chimera->getResults();
 833
 834                                                 //if either piece is chimeric then report
 835                                                 Sequence trimmed = chimera->print(out, out2, leftResults, rightResults);
 836                                                 if (trim) { trimmed.printSequence(out3);  }
 837
 838                                                 delete left; delete right;
 839
 840                                         }else { //already chimeric
 841                                                 //print results
 842                                                 Sequence trimmed = chimera->print(out, out2);
 843                                                 if (trim) { trimmed.printSequence(out3);  }
 844                                         }
 845
 846
 847                                 }
 848                                 count++;
 849                         }
 850
 851                         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 852                                 unsigned long int pos = inFASTA.tellg();
 853                                 if ((pos == -1) || (pos >= filePos->end)) { break; }
 854                         #else
 855                                 if (inFASTA.eof()) { break; }
 856                         #endif
 857
 858                         delete candidateSeq;
 859                         //report progress
 860                         if((count) % 100 == 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 861                 }
 862                 //report progress
 863                 if((count) % 100 != 0){ m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();         }
 864
 865                 out.close();
 866                 out2.close();
 867                 if (trim) { out3.close(); }
 868                 inFASTA.close();
 869
 870                 return count;
 871         }
 872         catch(exception& e) {
 873                 m->errorOut(e, "ChimeraSlayerCommand", "driver");
 874                 exit(1);
 875         }
 876 }
 877 //**********************************************************************************************************************
 878 #ifdef USE_MPI
 879 int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& outMPI, MPI_File& outAccMPI, MPI_File& outFastaMPI, vector<unsigned long int>& MPIPos){
 880         try {
 881                 MPI_Status status;
 882                 int pid;
 883                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 884
 885                 for(int i=0;i<num;i++){
 886
 887                         if (m->control_pressed) {       return 1;       }
 888
 889                         //read next sequence
 890                         int length = MPIPos[start+i+1] - MPIPos[start+i];
 891
 892                         char* buf4 = new char[length];
 893                         MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
 894
 895                         string tempBuf = buf4;
 896                         if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length);  }
 897                         istringstream iss (tempBuf,istringstream::in);
 898
 899                         delete buf4;
 900
 901                         Sequence* candidateSeq = new Sequence(iss);  m->gobble(iss);
 902                         string candidateAligned = candidateSeq->getAligned();
 903
 904                         if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
 905
 906                                 if (candidateSeq->getAligned().length() != templateSeqsLength) {
 907                                         m->mothurOut(candidateSeq->getName() + " is not the same length as the template sequences. Skipping."); m->mothurOutEndLine();
 908                                 }else{
 909
 910                                         //find chimeras
 911                                         chimera->getChimeras(candidateSeq);
 912
 913                                         if (m->control_pressed) {       delete candidateSeq; return 1;  }
 914
 915                                         //if you are not chimeric, then check each half
 916                                         data_results wholeResults = chimera->getResults();
 917
 918                                         //determine if we need to split
 919                                         bool isChimeric = false;
 920
 921                                         if (wholeResults.flag == "yes") {
 922                                                 string chimeraFlag = "no";
 923                                                 if(  (wholeResults.results[0].bsa >= minBS && wholeResults.results[0].divr_qla_qrb >= divR)
 924                                                    ||
 925                                                    (wholeResults.results[0].bsb >= minBS && wholeResults.results[0].divr_qlb_qra >= divR) ) { chimeraFlag = "yes"; }
 926
 927
 928                                                 if (chimeraFlag == "yes") {
 929                                                         if ((wholeResults.results[0].bsa >= minBS) || (wholeResults.results[0].bsb >= minBS)) { isChimeric = true; }
 930                                                 }
 931                                         }
 932
 933                                         if ((!isChimeric) && trimera) {
 934                                                 //split sequence in half by bases
 935                                                 string leftQuery, rightQuery;
 936                                                 Sequence tempSeq(candidateSeq->getName(), candidateAligned);
 937                                                 divideInHalf(tempSeq, leftQuery, rightQuery);
 938
 939                                                 //run chimeraSlayer on each piece
 940                                                 Sequence* left = new Sequence(candidateSeq->getName(), leftQuery);
 941                                                 Sequence* right = new Sequence(candidateSeq->getName(), rightQuery);
 942
 943                                                 //find chimeras
 944                                                 chimera->getChimeras(left);
 945                                                 data_results leftResults = chimera->getResults();
 946
 947                                                 chimera->getChimeras(right);
 948                                                 data_results rightResults = chimera->getResults();
 949
 950                                                 //if either piece is chimeric then report
 951                                                 Sequence trimmed = chimera->print(outMPI, outAccMPI, leftResults, rightResults);
 952                                                 if (trim) {
 953                                                         string outputString = ">" + trimmed.getName() + "\n" + trimmed.getAligned() + "\n";
 954
 955                                                         //write to accnos file
 956                                                         int length = outputString.length();
 957                                                         char* buf2 = new char[length];
 958                                                         memcpy(buf2, outputString.c_str(), length);
 959
 960                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 961                                                         delete buf2;
 962                                                 }
 963
 964                                                 delete left; delete right;
 965
 966                                         }else {
 967                                                 //print results
 968                                                 Sequence trimmed = chimera->print(outMPI, outAccMPI);
 969
 970                                                 if (trim) {
 971                                                         string outputString = ">" + trimmed.getName() + "\n" + trimmed.getAligned() + "\n";
 972
 973                                                         //write to accnos file
 974                                                         int length = outputString.length();
 975                                                         char* buf2 = new char[length];
 976                                                         memcpy(buf2, outputString.c_str(), length);
 977
 978                                                         MPI_File_write_shared(outFastaMPI, buf2, length, MPI_CHAR, &status);
 979                                                         delete buf2;
 980                                                 }
 981                                         }
 982
 983                                 }
 984                         }
 985                         delete candidateSeq;
 986
 987                         //report progress
 988                         if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;        m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");          }
 989                 }
 990                 //report progress
 991                 if(num % 100 != 0){             cout << "Processing sequence: " << num << endl; m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n");  }
 992
 993
 994                 return 0;
 995         }
 996         catch(exception& e) {
 997                 m->errorOut(e, "ChimeraSlayerCommand", "driverMPI");
 998                 exit(1);
 999         }
1000 }
1001 #endif
1002
1003 /**************************************************************************************************/
1004
1005 int ChimeraSlayerCommand::createProcesses(string outputFileName, string filename, string accnos, string fasta) {
1006         try {
1007                 int process = 0;
1008                 int num = 0;
1009                 int numNoParents = 0;
1010
1011 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
1012                 //loop through and create all the processes you want
1013                 while (process != processors) {
1014                         int pid = fork();
1015
1016                         if (pid > 0) {
1017                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
1018                                 process++;
1019                         }else if (pid == 0){
1020                                 num = driver(lines[process], outputFileName + toString(getpid()) + ".temp", filename, accnos + toString(getpid()) + ".temp", fasta + toString(getpid()) + ".temp");
1021
1022                                 //pass numSeqs to parent
1023                                 ofstream out;
1024                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
1025                                 m->openOutputFile(tempFile, out);
1026                                 out << num << '\t' << chimera->getNumNoParents() << endl;
1027                                 out.close();
1028                                 exit(0);
1029                         }else {
1030                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
1031                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
1032                                 exit(0);
1033                         }
1034                 }
1035
1036                 //force parent to wait until all the processes are done
1037                 for (int i=0;i<processors;i++) {
1038                         int temp = processIDS[i];
1039                         wait(&temp);
1040                 }
1041
1042                 for (int i = 0; i < processIDS.size(); i++) {
1043                         ifstream in;
1044                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
1045                         m->openInputFile(tempFile, in);
1046                         if (!in.eof()) { int tempNum = 0; int tempNumParents = 0; in >> tempNum >> tempNumParents; num += tempNum; numNoParents += tempNumParents; }
1047                         in.close(); m->mothurRemove(tempFile);
1048                 }
1049 #else
1050
1051                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1052                 //Windows version shared memory, so be careful when passing variables through the slayerData struct.
1053                 //Above fork() will clone, so memory is separate, but that's not the case with windows,
1054                 //////////////////////////////////////////////////////////////////////////////////////////////////////
1055
1056                 vector<slayerData*> pDataArray;
1057                 DWORD   dwThreadIdArray[processors];
1058                 HANDLE  hThreadArray[processors];
1059
1060                 //Create processor worker threads.
1061                 for( int i=0; i<processors; i++ ){
1062                         string extension = toString(i) + ".temp";
1063                         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);
1064                         pDataArray.push_back(tempslayer);
1065                         processIDS.push_back(i);
1066
1067                         //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
1068                         //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
1069                         hThreadArray[i] = CreateThread(NULL, 0, MySlayerThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
1070                 }
1071
1072                 //Wait until all threads have terminated.
1073                 WaitForMultipleObjects(processors, hThreadArray, TRUE, INFINITE);
1074
1075                 //Close all thread handles and free memory allocations.
1076                 for(int i=0; i < pDataArray.size(); i++){
1077                         num += pDataArray[i]->count;
1078                         numNoParents += pDataArray[i]->numNoParents;
1079                         CloseHandle(hThreadArray[i]);
1080                         delete pDataArray[i];
1081                 }
1082 #endif
1083                 if (num == numNoParents) {  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(); }
1084                 return num;
1085         }
1086         catch(exception& e) {
1087                 m->errorOut(e, "ChimeraSlayerCommand", "createProcesses");
1088                 exit(1);
1089         }
1090 }
1091
1092 /**************************************************************************************************/
1093
1094 int ChimeraSlayerCommand::divideInHalf(Sequence querySeq, string& leftQuery, string& rightQuery) {
1095         try {
1096
1097                 string queryUnAligned = querySeq.getUnaligned();
1098                 int numBases = int(queryUnAligned.length() * 0.5);
1099
1100                 string queryAligned = querySeq.getAligned();
1101                 leftQuery = querySeq.getAligned();
1102                 rightQuery = querySeq.getAligned();
1103
1104                 int baseCount = 0;
1105                 int leftSpot = 0;
1106                 for (int i = 0; i < queryAligned.length(); i++) {
1107                         //if you are a base
1108                         if (isalpha(queryAligned[i])) {
1109                                 baseCount++;
1110                         }
1111
1112                         //if you have half
1113                         if (baseCount >= numBases) {  leftSpot = i; break; } //first half
1114                 }
1115
1116                 //blank out right side
1117                 for (int i = leftSpot; i < leftQuery.length(); i++) { leftQuery[i] = '.'; }
1118
1119                 //blank out left side
1120                 for (int i = 0; i < leftSpot; i++) { rightQuery[i] = '.'; }
1121
1122                 return 0;
1123
1124         }
1125         catch(exception& e) {
1126                 m->errorOut(e, "ChimeraSlayerCommand", "divideInHalf");
1127                 exit(1);
1128         }
1129 }
1130 /**************************************************************************************************/
1131 map<string, int> ChimeraSlayerCommand::sortFastaFile(string fastaFile, string nameFile) {
1132         try {
1133                 map<string, int> nameAbund;
1134
1135                 //read through fastafile and store info
1136                 map<string, string> seqs;
1137                 ifstream in;
1138                 m->openInputFile(fastaFile, in);
1139
1140                 while (!in.eof()) {
1141
1142                         if (m->control_pressed) { in.close(); return nameAbund; }
1143
1144                         Sequence seq(in); m->gobble(in);
1145                         seqs[seq.getName()] = seq.getAligned();
1146                 }
1147
1148                 in.close();
1149
1150                 //read namefile
1151                 vector<seqPriorityNode> nameMapCount;
1152                 int error = m->readNames(nameFile, nameMapCount, seqs);
1153
1154                 if (m->control_pressed) { return nameAbund; }
1155
1156                 if (error == 1) { m->control_pressed = true; return nameAbund; }
1157                 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; }
1158
1159                 sort(nameMapCount.begin(), nameMapCount.end(), compareSeqPriorityNodes);
1160
1161                 string newFasta = fastaFile + ".temp";
1162                 ofstream out;
1163                 m->openOutputFile(newFasta, out);
1164
1165                 //print new file in order of
1166                 for (int i = 0; i < nameMapCount.size(); i++) {
1167                         out << ">" << nameMapCount[i].name << endl << nameMapCount[i].seq << endl;
1168                         nameAbund[nameMapCount[i].name] = nameMapCount[i].numIdentical;
1169                 }
1170                 out.close();
1171
1172                 rename(newFasta.c_str(), fastaFile.c_str());
1173
1174                 return nameAbund;
1175
1176         }
1177         catch(exception& e) {
1178                 m->errorOut(e, "ChimeraSlayerCommand", "sortFastaFile");
1179                 exit(1);
1180         }
1181 }
1182
1183 /**************************************************************************************************/
1184