chimerauchimecommand.cpp

   1 /*
   2  *  chimerauchimecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 5/13/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "chimerauchimecommand.h"
  11 #include "deconvolutecommand.h"
  12 #include "uc.h"
  13 #include "sequence.hpp"
  14
  15
  16 //**********************************************************************************************************************
  17 vector<string> ChimeraUchimeCommand::setParameters(){
  18         try {
  19                 CommandParameter ptemplate("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(ptemplate);
  20                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
  23                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  24                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  25                 CommandParameter pabskew("abskew", "Number", "", "1.9", "", "", "",false,false); parameters.push_back(pabskew);
  26                 CommandParameter pchimealns("chimealns", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pchimealns);
  27                 CommandParameter pminh("minh", "Number", "", "0.3", "", "", "",false,false); parameters.push_back(pminh);
  28                 CommandParameter pmindiv("mindiv", "Number", "", "0.5", "", "", "",false,false); parameters.push_back(pmindiv);
  29                 CommandParameter pxn("xn", "Number", "", "8.0", "", "", "",false,false); parameters.push_back(pxn);
  30                 CommandParameter pdn("dn", "Number", "", "1.4", "", "", "",false,false); parameters.push_back(pdn);
  31                 CommandParameter pxa("xa", "Number", "", "1", "", "", "",false,false); parameters.push_back(pxa);
  32                 CommandParameter pchunks("chunks", "Number", "", "4", "", "", "",false,false); parameters.push_back(pchunks);
  33                 CommandParameter pminchunk("minchunk", "Number", "", "64", "", "", "",false,false); parameters.push_back(pminchunk);
  34                 CommandParameter pidsmoothwindow("idsmoothwindow", "Number", "", "32", "", "", "",false,false); parameters.push_back(pidsmoothwindow);
  35                 CommandParameter pminsmoothid("minsmoothid", "Number", "", "0.95", "", "", "",false,false); parameters.push_back(pminsmoothid);
  36                 CommandParameter pmaxp("maxp", "Number", "", "2", "", "", "",false,false); parameters.push_back(pmaxp);
  37                 CommandParameter pskipgaps("skipgaps", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pskipgaps);
  38                 CommandParameter pskipgaps2("skipgaps2", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pskipgaps2);
  39                 CommandParameter pminlen("minlen", "Number", "", "10", "", "", "",false,false); parameters.push_back(pminlen);
  40                 CommandParameter pmaxlen("maxlen", "Number", "", "10000", "", "", "",false,false); parameters.push_back(pmaxlen);
  41                 CommandParameter pucl("ucl", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pucl);
  42                 CommandParameter pqueryfract("queryfract", "Number", "", "0.5", "", "", "",false,false); parameters.push_back(pqueryfract);
  43
  44                 vector<string> myArray;
  45                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  46                 return myArray;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "ChimeraUchimeCommand", "setParameters");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 string ChimeraUchimeCommand::getHelpString(){
  55         try {
  56                 string helpString = "";
  57                 helpString += "The chimera.uchime command reads a fastafile and referencefile and outputs potentially chimeric sequences.\n";
  58                 helpString += "This command is a wrapper for uchime written by Robert C. Edgar.\n";
  59                 helpString += "The chimera.uchime command parameters are fasta, name, reference, processors, abskew, chimealns, minh, mindiv, xn, dn, xa, chunks, minchunk, idsmoothwindow, minsmoothid, maxp, skipgaps, skipgaps2, minlen, maxlen, ucl and queryfact.\n";
  60                 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";
  61                 helpString += "The name parameter allows you to provide a name file, if you are using template=self. \n";
  62                 helpString += "You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amazon.fasta \n";
  63                 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";
  64                 helpString += "The processors parameter allows you to specify how many processors you would like to use.  The default is 1. \n";
  65                 helpString += "The abskew parameter can only be used with template=self. Minimum abundance skew. Default 1.9. Abundance skew is: min [ abund(parent1), abund(parent2) ] / abund(query).\n";
  66                 helpString += "The chimealns parameter allows you to indicate you would like a file containing multiple alignments of query sequences to parents in human readable format. Alignments show columns with differences that support or contradict a chimeric model.\n";
  67                 helpString += "The minh parameter - mininum score to report chimera. Default 0.3. Values from 0.1 to 5 might be reasonable. Lower values increase sensitivity but may report more false positives. If you decrease xn you may need to increase minh, and vice versa.\n";
  68                 helpString += "The mindiv parameter - minimum divergence ratio, default 0.5. Div ratio is 100%% - %%identity between query sequence and the closest candidate for being a parent. If you don't care about very close chimeras, then you could increase mindiv to, say, 1.0 or 2.0, and also decrease --min h, say to 0.1, to increase sensitivity. How well this works will depend on your data. Best is to tune parameters on a good benchmark.\n";
  69                 helpString += "The xn parameter - weight of a no vote. Default 8.0. Decreasing this weight to around 3 or 4 may give better performance on denoised data.\n";
  70                 helpString += "The dn parameter - pseudo-count prior on number of no votes. Default 1.4. Probably no good reason to change this unless you can retune to a good benchmark for your data. Reasonable values are probably in the range from 0.2 to 2.\n";
  71                 helpString += "The xa parameter - weight of an abstain vote. Default 1. So far, results do not seem to be very sensitive to this parameter, but if you have a good training set might be worth trying. Reasonable values might range from 0.1 to 2.\n";
  72                 helpString += "The chunks parameter is the number of chunks to extract from the query sequence when searching for parents. Default 4.\n";
  73                 helpString += "The minchunk parameter is the minimum length of a chunk. Default 64.\n";
  74                 helpString += "The idsmoothwindow parameter is the length of id smoothing window. Default 32.\n";
  75                 helpString += "The minsmoothid parameter - minimum factional identity over smoothed window of candidate parent. Default 0.95.\n";
  76                 helpString += "The maxp parameter - maximum number of candidate parents to consider. Default 2. In tests so far, increasing maxp gives only a very small improvement in sensivity but tends to increase the error rate quite a bit.\n";
  77                 helpString += "The skipgaps parameter controls how gapped columns affect counting of diffs. If skipgaps is set to T, columns containing gaps do not found as diffs. Default = T.\n";
  78                 helpString += "The skipgaps2 parameter controls how gapped columns affect counting of diffs. If skipgaps2 is set to T, if column is immediately adjacent to a column containing a gap, it is not counted as a diff. Default = T.\n";
  79                 helpString += "The minlen parameter is the minimum unaligned sequence length. Defaults 10. Applies to both query and reference sequences.\n";
  80                 helpString += "The maxlen parameter is the maximum unaligned sequence length. Defaults 10000. Applies to both query and reference sequences.\n";
  81                 helpString += "The ucl parameter - use local-X alignments. Default is global-X or false. On tests so far, global-X is always better; this option is retained because it just might work well on some future type of data.\n";
  82                 helpString += "The queryfract parameter - minimum fraction of the query sequence that must be covered by a local-X alignment. Default 0.5. Applies only when ucl is true.\n";
  83 #ifdef USE_MPI
  84                 helpString += "When using MPI, the processors parameter is set to the number of MPI processes running. \n";
  85 #endif
  86                 helpString += "The chimera.uchime command should be in the following format: \n";
  87                 helpString += "chimera.uchime(fasta=yourFastaFile, reference=yourTemplate) \n";
  88                 helpString += "Example: chimera.uchime(fasta=AD.align, reference=silva.gold.align) \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, "ChimeraUchimeCommand", "getHelpString");
  94                 exit(1);
  95         }
  96 }
  97 //**********************************************************************************************************************
  98 ChimeraUchimeCommand::ChimeraUchimeCommand(){
  99         try {
 100                 abort = true; calledHelp = true;
 101                 setParameters();
 102                 vector<string> tempOutNames;
 103                 outputTypes["chimera"] = tempOutNames;
 104                 outputTypes["accnos"] = tempOutNames;
 105                 outputTypes["alns"] = tempOutNames;
 106         }
 107         catch(exception& e) {
 108                 m->errorOut(e, "ChimeraUchimeCommand", "ChimeraUchimeCommand");
 109                 exit(1);
 110         }
 111 }
 112 //***************************************************************************************************************
 113 ChimeraUchimeCommand::ChimeraUchimeCommand(string option)  {
 114         try {
 115                 abort = false; calledHelp = false;
 116
 117                 //allow user to run help
 118                 if(option == "help") { help(); abort = true; calledHelp = true; }
 119                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 120
 121                 else {
 122                         vector<string> myArray = setParameters();
 123
 124                         OptionParser parser(option);
 125                         map<string,string> parameters = parser.getParameters();
 126
 127                         ValidParameters validParameter("chimera.uchime");
 128                         map<string,string>::iterator it;
 129
 130                         //check to make sure all parameters are valid for command
 131                         for (it = parameters.begin(); it != parameters.end(); it++) {
 132                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 133                         }
 134
 135                         vector<string> tempOutNames;
 136                         outputTypes["chimera"] = tempOutNames;
 137                         outputTypes["accnos"] = tempOutNames;
 138                         outputTypes["alns"] = tempOutNames;
 139
 140                         //if the user changes the input directory command factory will send this info to us in the output parameter
 141                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 142                         if (inputDir == "not found"){   inputDir = "";          }
 143
 144                         //check for required parameters
 145                         fastafile = validParameter.validFile(parameters, "fasta", false);
 146                         if (fastafile == "not found") {
 147                                 //if there is a current fasta file, use it
 148                                 string filename = m->getFastaFile();
 149                                 if (filename != "") { fastaFileNames.push_back(filename); m->mothurOut("Using " + filename + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 150                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 151                         }else {
 152                                 m->splitAtDash(fastafile, fastaFileNames);
 153
 154                                 //go through files and make sure they are good, if not, then disregard them
 155                                 for (int i = 0; i < fastaFileNames.size(); i++) {
 156
 157                                         bool ignore = false;
 158                                         if (fastaFileNames[i] == "current") {
 159                                                 fastaFileNames[i] = m->getFastaFile();
 160                                                 if (fastaFileNames[i] != "") {  m->mothurOut("Using " + fastaFileNames[i] + " as input file for the fasta parameter where you had given current."); m->mothurOutEndLine(); }
 161                                                 else {
 162                                                         m->mothurOut("You have no current fastafile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 163                                                         //erase from file list
 164                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 165                                                         i--;
 166                                                 }
 167                                         }
 168
 169                                         if (!ignore) {
 170
 171                                                 if (inputDir != "") {
 172                                                         string path = m->hasPath(fastaFileNames[i]);
 173                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 174                                                         if (path == "") {       fastaFileNames[i] = inputDir + fastaFileNames[i];               }
 175                                                 }
 176
 177                                                 int ableToOpen;
 178                                                 ifstream in;
 179
 180                                                 ableToOpen = m->openInputFile(fastaFileNames[i], in, "noerror");
 181
 182                                                 //if you can't open it, try default location
 183                                                 if (ableToOpen == 1) {
 184                                                         if (m->getDefaultPath() != "") { //default path is set
 185                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(fastaFileNames[i]);
 186                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 187                                                                 ifstream in2;
 188                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 189                                                                 in2.close();
 190                                                                 fastaFileNames[i] = tryPath;
 191                                                         }
 192                                                 }
 193
 194                                                 if (ableToOpen == 1) {
 195                                                         if (m->getOutputDir() != "") { //default path is set
 196                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(fastaFileNames[i]);
 197                                                                 m->mothurOut("Unable to open " + fastaFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 198                                                                 ifstream in2;
 199                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 200                                                                 in2.close();
 201                                                                 fastaFileNames[i] = tryPath;
 202                                                         }
 203                                                 }
 204
 205                                                 in.close();
 206
 207                                                 if (ableToOpen == 1) {
 208                                                         m->mothurOut("Unable to open " + fastaFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 209                                                         //erase from file list
 210                                                         fastaFileNames.erase(fastaFileNames.begin()+i);
 211                                                         i--;
 212                                                 }
 213                                         }
 214                                 }
 215
 216                                 //make sure there is at least one valid file left
 217                                 if (fastaFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid files."); m->mothurOutEndLine(); abort = true; }
 218                         }
 219
 220
 221                         //check for required parameters
 222                         bool hasName = true;
 223                         namefile = validParameter.validFile(parameters, "name", false);
 224                         if (namefile == "not found") { namefile = "";  hasName = false; }
 225                         else {
 226                                 m->splitAtDash(namefile, nameFileNames);
 227
 228                                 //go through files and make sure they are good, if not, then disregard them
 229                                 for (int i = 0; i < nameFileNames.size(); i++) {
 230
 231                                         bool ignore = false;
 232                                         if (nameFileNames[i] == "current") {
 233                                                 nameFileNames[i] = m->getNameFile();
 234                                                 if (nameFileNames[i] != "") {  m->mothurOut("Using " + nameFileNames[i] + " as input file for the name parameter where you had given current."); m->mothurOutEndLine(); }
 235                                                 else {
 236                                                         m->mothurOut("You have no current namefile, ignoring current."); m->mothurOutEndLine(); ignore=true;
 237                                                         //erase from file list
 238                                                         nameFileNames.erase(nameFileNames.begin()+i);
 239                                                         i--;
 240                                                 }
 241                                         }
 242
 243                                         if (!ignore) {
 244
 245                                                 if (inputDir != "") {
 246                                                         string path = m->hasPath(nameFileNames[i]);
 247                                                         //if the user has not given a path then, add inputdir. else leave path alone.
 248                                                         if (path == "") {       nameFileNames[i] = inputDir + nameFileNames[i];         }
 249                                                 }
 250
 251                                                 int ableToOpen;
 252                                                 ifstream in;
 253
 254                                                 ableToOpen = m->openInputFile(nameFileNames[i], in, "noerror");
 255
 256                                                 //if you can't open it, try default location
 257                                                 if (ableToOpen == 1) {
 258                                                         if (m->getDefaultPath() != "") { //default path is set
 259                                                                 string tryPath = m->getDefaultPath() + m->getSimpleName(nameFileNames[i]);
 260                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
 261                                                                 ifstream in2;
 262                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 263                                                                 in2.close();
 264                                                                 nameFileNames[i] = tryPath;
 265                                                         }
 266                                                 }
 267
 268                                                 if (ableToOpen == 1) {
 269                                                         if (m->getOutputDir() != "") { //default path is set
 270                                                                 string tryPath = m->getOutputDir() + m->getSimpleName(nameFileNames[i]);
 271                                                                 m->mothurOut("Unable to open " + nameFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
 272                                                                 ifstream in2;
 273                                                                 ableToOpen = m->openInputFile(tryPath, in2, "noerror");
 274                                                                 in2.close();
 275                                                                 nameFileNames[i] = tryPath;
 276                                                         }
 277                                                 }
 278
 279                                                 in.close();
 280
 281                                                 if (ableToOpen == 1) {
 282                                                         m->mothurOut("Unable to open " + nameFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
 283                                                         //erase from file list
 284                                                         nameFileNames.erase(nameFileNames.begin()+i);
 285                                                         i--;
 286                                                 }
 287                                         }
 288                                 }
 289
 290                                 //make sure there is at least one valid file left
 291                                 if (nameFileNames.size() == 0) { m->mothurOut("[ERROR]: no valid name files."); m->mothurOutEndLine(); abort = true; }
 292                         }
 293
 294                         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; }
 295
 296                         //if the user changes the output directory command factory will send this info to us in the output parameter
 297                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = ""; }
 298
 299
 300                         string path;
 301                         it = parameters.find("reference");
 302                         //user has given a template file
 303                         if(it != parameters.end()){
 304                                 if (it->second == "self") { templatefile = "self"; }
 305                                 else {
 306                                         path = m->hasPath(it->second);
 307                                         //if the user has not given a path then, add inputdir. else leave path alone.
 308                                         if (path == "") {       parameters["reference"] = inputDir + it->second;                }
 309
 310                                         templatefile = validParameter.validFile(parameters, "reference", true);
 311                                         if (templatefile == "not open") { abort = true; }
 312                                         else if (templatefile == "not found") { templatefile = "";  m->mothurOut("reference is a required parameter for the chimera.uchime command."); m->mothurOutEndLine(); abort = true;  }
 313                                 }
 314                         }else if (hasName) {  templatefile = "self"; }
 315                         else { templatefile = "";  m->mothurOut("reference is a required parameter for the chimera.uchime command, unless you have a namefile."); m->mothurOutEndLine(); abort = true;  }
 316
 317
 318                         string temp = validParameter.validFile(parameters, "processors", false);        if (temp == "not found"){       temp = m->getProcessors();      }
 319                         m->setProcessors(temp);
 320                         convert(temp, processors);
 321
 322                         abskew = validParameter.validFile(parameters, "abskew", false); if (abskew == "not found"){     useAbskew = false;  abskew = "1.9";     }else{  useAbskew = true;  }
 323                         if (useAbskew && templatefile != "self") { m->mothurOut("The abskew parameter is only valid with template=self, ignoring."); m->mothurOutEndLine(); useAbskew = false; }
 324
 325                         temp = validParameter.validFile(parameters, "chimealns", false);                        if (temp == "not found") { temp = "f"; }
 326                         chimealns = m->isTrue(temp);
 327
 328                         minh = validParameter.validFile(parameters, "minh", false);                                             if (minh == "not found")                        { useMinH = false; minh = "0.3";                                        }       else{ useMinH = true;                   }
 329                         mindiv = validParameter.validFile(parameters, "mindiv", false);                                 if (mindiv == "not found")                      { useMindiv = false; mindiv = "0.5";                            }       else{ useMindiv = true;                 }
 330                         xn = validParameter.validFile(parameters, "xn", false);                                                 if (xn == "not found")                          { useXn = false; xn = "8.0";                                            }       else{ useXn = true;                             }
 331                         dn = validParameter.validFile(parameters, "dn", false);                                                 if (dn == "not found")                          { useDn = false; dn = "1.4";                                            }       else{ useDn = true;                             }
 332                         xa = validParameter.validFile(parameters, "xa", false);                                                 if (xa == "not found")                          { useXa = false; xa = "1";                                                      }       else{ useXa = true;                             }
 333                         chunks = validParameter.validFile(parameters, "chunks", false);                                 if (chunks == "not found")                      { useChunks = false; chunks = "4";                                      }       else{ useChunks = true;                 }
 334                         minchunk = validParameter.validFile(parameters, "minchunk", false);                             if (minchunk == "not found")            { useMinchunk = false; minchunk = "64";                         }       else{ useMinchunk = true;               }
 335                         idsmoothwindow = validParameter.validFile(parameters, "idsmoothwindow", false); if (idsmoothwindow == "not found")      { useIdsmoothwindow = false; idsmoothwindow = "32";     }       else{ useIdsmoothwindow = true; }
 336                         minsmoothid = validParameter.validFile(parameters, "minsmoothid", false);               if (minsmoothid == "not found")         { useMinsmoothid = false; minsmoothid = "0.95";         }       else{ useMinsmoothid = true;    }
 337                         maxp = validParameter.validFile(parameters, "maxp", false);                                             if (maxp == "not found")                        { useMaxp = false; maxp = "2";                                          }       else{ useMaxp = true;                   }
 338                         minlen = validParameter.validFile(parameters, "minlen", false);                                 if (minlen == "not found")                      { useMinlen = false; minlen = "10";                                     }       else{ useMinlen = true;                 }
 339                         maxlen = validParameter.validFile(parameters, "maxlen", false);                                 if (maxlen == "not found")                      { useMaxlen = false; maxlen = "10000";                          }       else{ useMaxlen = true;                 }
 340
 341                         temp = validParameter.validFile(parameters, "ucl", false);                                              if (temp == "not found") { temp = "f"; }
 342                         ucl = m->isTrue(temp);
 343
 344                         queryfract = validParameter.validFile(parameters, "queryfract", false);                 if (queryfract == "not found")          { useQueryfract = false; queryfract = "0.5";            }       else{ useQueryfract = true;             }
 345                         if (!ucl && useQueryfract) { m->mothurOut("queryfact may only be used when ucl=t, ignoring."); m->mothurOutEndLine(); useQueryfract = false; }
 346
 347                         temp = validParameter.validFile(parameters, "skipgaps", false);                                 if (temp == "not found") { temp = "t"; }
 348                         skipgaps = m->isTrue(temp);
 349
 350                         temp = validParameter.validFile(parameters, "skipgaps2", false);                                if (temp == "not found") { temp = "t"; }
 351                         skipgaps2 = m->isTrue(temp);
 352
 353                 }
 354         }
 355         catch(exception& e) {
 356                 m->errorOut(e, "ChimeraSlayerCommand", "ChimeraSlayerCommand");
 357                 exit(1);
 358         }
 359 }
 360 //***************************************************************************************************************
 361
 362 int ChimeraUchimeCommand::execute(){
 363         try{
 364                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 365
 366                 m->mothurOut("\nuchime by Robert C. Edgar\nhttp://drive5.com/uchime\nThis code is donated to the public domain.\n\n");
 367
 368                 for (int s = 0; s < fastaFileNames.size(); s++) {
 369
 370                         m->mothurOut("Checking sequences from " + fastaFileNames[s] + " ..." ); m->mothurOutEndLine();
 371
 372                         int start = time(NULL);
 373                         string nameFile = "";
 374
 375                         if (templatefile == "self") { //you want to run uchime with a refernce template
 376
 377                                 #ifdef USE_MPI
 378                                         int pid;
 379                                         MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 380                                         if (pid == 0) { //you are the root process
 381                                 #endif
 382
 383                                 if (processors != 1) { m->mothurOut("When using template=self, mothur can only use 1 processor, continuing."); m->mothurOutEndLine(); processors = 1; }
 384                                 if (nameFileNames.size() != 0) { //you provided a namefile and we don't need to create one
 385                                         nameFile = nameFileNames[s];
 386                                 }else {
 387                                         m->mothurOutEndLine(); m->mothurOut("No namesfile given, running unique.seqs command to generate one."); m->mothurOutEndLine(); m->mothurOutEndLine();
 388
 389                                         //use unique.seqs to create new name and fastafile
 390                                         string inputString = "fasta=" + fastaFileNames[s];
 391                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 392                                         m->mothurOut("Running command: unique.seqs(" + inputString + ")"); m->mothurOutEndLine();
 393
 394                                         Command* uniqueCommand = new DeconvoluteCommand(inputString);
 395                                         uniqueCommand->execute();
 396
 397                                         map<string, vector<string> > filenames = uniqueCommand->getOutputFiles();
 398
 399                                         delete uniqueCommand;
 400
 401                                         m->mothurOut("/******************************************/"); m->mothurOutEndLine();
 402
 403                                         nameFile = filenames["name"][0];
 404                                         fastaFileNames[s] = filenames["fasta"][0];
 405                                 }
 406
 407                                 //create input file for uchime
 408                                 //read through fastafile and store info
 409                                 map<string, string> seqs;
 410                                 ifstream in;
 411                                 m->openInputFile(fastaFileNames[s], in);
 412
 413                                 while (!in.eof()) {
 414
 415                                         if (m->control_pressed) { in.close(); for (int j = 0; j < outputNames.size(); j++) {    remove(outputNames[j].c_str()); }  return 0; }
 416
 417                                         Sequence seq(in); m->gobble(in);
 418                                         seqs[seq.getName()] = seq.getAligned();
 419                                 }
 420                                 in.close();
 421
 422                                 //read namefile
 423                                 vector<seqPriorityNode> nameMapCount;
 424                                 int error = m->readNames(nameFile, nameMapCount, seqs);
 425
 426                                 if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0; }
 427
 428                                 if (error == 1) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); }  return 0; }
 429                                 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(); for (int j = 0; j < outputNames.size(); j++) {  remove(outputNames[j].c_str()); }  return 0; }
 430
 431                                 sort(nameMapCount.begin(), nameMapCount.end(), compareSeqPriorityNodes);
 432
 433                                 string newFasta = fastaFileNames[s] + ".temp";
 434                                 ofstream out;
 435                                 m->openOutputFile(newFasta, out);
 436
 437                                 //print new file in order of
 438                                 for (int i = 0; i < nameMapCount.size(); i++) {
 439                                         out << ">" << nameMapCount[i].name  << "/ab=" << nameMapCount[i].numIdentical << "/" << endl << nameMapCount[i].seq << endl;
 440                                 }
 441                                 out.close();
 442
 443                                 fastaFileNames[s] = newFasta;
 444
 445                                 #ifdef USE_MPI
 446                                         }
 447                                 #endif
 448                                 if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); }  return 0;    }
 449                         }
 450
 451                         if (outputDir == "") { outputDir = m->hasPath(fastaFileNames[s]);  }//if user entered a file with a path then preserve it
 452                         string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s])) + "uchime.chimera";
 453                         string accnosFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "uchime.accnos";
 454                         string alnsFileName = outputDir + m->getRootName(m->getSimpleName(fastaFileNames[s]))  + "uchime.alns";
 455
 456                         if (m->control_pressed) {  for (int j = 0; j < outputNames.size(); j++) {       remove(outputNames[j].c_str()); }  return 0;    }
 457
 458                         int numSeqs = 0;
 459 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 460                         if(processors == 1){ numSeqs = driver(outputFileName, fastaFileNames[s], accnosFileName, alnsFileName); }
 461                         else{   numSeqs = createProcesses(outputFileName, fastaFileNames[s], accnosFileName, alnsFileName); }
 462 #else
 463                         numSeqs = driver(outputFileName, fastaFileNames[s], accnosFileName, alnsFileName);
 464 #endif
 465                         if (m->control_pressed) { for (int j = 0; j < outputNames.size(); j++) {        remove(outputNames[j].c_str()); } return 0; }
 466
 467                         //remove file made for uchime
 468                         if (templatefile == "self") {  remove(fastaFileNames[s].c_str()); }
 469
 470                         outputNames.push_back(outputFileName); outputTypes["chimera"].push_back(outputFileName);
 471                         outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName);
 472                         if (chimealns) { outputNames.push_back(alnsFileName); outputTypes["alns"].push_back(alnsFileName); }
 473
 474                         m->mothurOutEndLine(); m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences."); m->mothurOutEndLine();
 475                 }
 476
 477                 //set accnos file as new current accnosfile
 478                 string current = "";
 479                 itTypes = outputTypes.find("accnos");
 480                 if (itTypes != outputTypes.end()) {
 481                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 482                 }
 483
 484                 m->mothurOutEndLine();
 485                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 486                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 487                 m->mothurOutEndLine();
 488
 489                 return 0;
 490
 491         }
 492         catch(exception& e) {
 493                 m->errorOut(e, "ChimeraUchimeCommand", "execute");
 494                 exit(1);
 495         }
 496 }
 497 //**********************************************************************************************************************
 498
 499 int ChimeraUchimeCommand::driver(string outputFName, string filename, string accnos, string alns){
 500         try {
 501
 502                 vector<char*> cPara;
 503
 504                 char* tempUchime = new char[8];
 505                 strcpy(tempUchime, "./uchime ");
 506                 cPara.push_back(tempUchime);
 507
 508                 char* tempIn = new char[7];
 509                 strcpy(tempIn, "--input");
 510                 cPara.push_back(tempIn);
 511                 char* temp = new char[filename.length()];
 512                 strcpy(temp, filename.c_str());
 513                 cPara.push_back(temp);
 514
 515                 //are you using a reference file
 516                 if (templatefile != "self") {
 517
 518                         //add reference file
 519                         char* tempRef = new char[4];
 520                         strcpy(tempRef, "--db");
 521                         cPara.push_back(tempRef);
 522                         char* tempR = new char[templatefile.length()];
 523                         strcpy(tempR, templatefile.c_str());
 524                         cPara.push_back(tempR);
 525                 }
 526
 527                 char* tempO = new char[11];
 528                 strcpy(tempO, "--uchimeout");
 529                 cPara.push_back(tempO);
 530                 char* tempout = new char[outputFName.length()];
 531                 strcpy(tempout, outputFName.c_str());
 532                 cPara.push_back(tempout);
 533
 534                 if (chimealns) {
 535                         char* tempA = new char[12];
 536                         strcpy(tempA, "--uchimealns");
 537                         cPara.push_back(tempA);
 538                         char* tempa = new char[alns.length()];
 539                         strcpy(tempa, alns.c_str());
 540                         cPara.push_back(tempa);
 541                 }
 542
 543                 if (useAbskew) {
 544                         char* tempskew = new char[8];
 545                         strcpy(tempskew, "--abskew");
 546                         cPara.push_back(tempskew);
 547                         char* tempSkew = new char[abskew.length()];
 548                         strcpy(tempSkew, abskew.c_str());
 549                         cPara.push_back(tempSkew);
 550                 }
 551
 552                 if (useMinH) {
 553                         char* tempminh = new char[6];
 554                         strcpy(tempminh, "--minh");
 555                         cPara.push_back(tempminh);
 556                         char* tempMinH = new char[minh.length()];
 557                         strcpy(tempMinH, minh.c_str());
 558                         cPara.push_back(tempMinH);
 559                 }
 560
 561                 if (useMindiv) {
 562                         char* tempmindiv = new char[8];
 563                         strcpy(tempmindiv, "--mindiv");
 564                         cPara.push_back(tempmindiv);
 565                         char* tempMindiv = new char[mindiv.length()];
 566                         strcpy(tempMindiv, mindiv.c_str());
 567                         cPara.push_back(tempMindiv);
 568                 }
 569
 570                 if (useXn) {
 571                         char* tempxn = new char[4];
 572                         strcpy(tempxn, "--xn");
 573                         cPara.push_back(tempxn);
 574                         char* tempXn = new char[xn.length()];
 575                         strcpy(tempXn, xn.c_str());
 576                         cPara.push_back(tempXn);
 577                 }
 578
 579                 if (useDn) {
 580                         char* tempdn = new char[4];
 581                         strcpy(tempdn, "--dn");
 582                         cPara.push_back(tempdn);
 583                         char* tempDn = new char[dn.length()];
 584                         strcpy(tempDn, dn.c_str());
 585                         cPara.push_back(tempDn);
 586                 }
 587
 588                 if (useXa) {
 589                         char* tempxa = new char[4];
 590                         strcpy(tempxa, "--xa");
 591                         cPara.push_back(tempxa);
 592                         char* tempXa = new char[xa.length()];
 593                         strcpy(tempXa, xa.c_str());
 594                         cPara.push_back(tempXa);
 595                 }
 596
 597                 if (useChunks) {
 598                         char* tempchunks = new char[8];
 599                         strcpy(tempchunks, "--chunks");
 600                         cPara.push_back(tempchunks);
 601                         char* tempChunks = new char[chunks.length()];
 602                         strcpy(tempChunks, chunks.c_str());
 603                         cPara.push_back(tempChunks);
 604                 }
 605
 606                 if (useMinchunk) {
 607                         char* tempminchunk = new char[10];
 608                         strcpy(tempminchunk, "--minchunk");
 609                         cPara.push_back(tempminchunk);
 610                         char* tempMinchunk = new char[minchunk.length()];
 611                         strcpy(tempMinchunk, minchunk.c_str());
 612                         cPara.push_back(tempMinchunk);
 613                 }
 614
 615                 if (useIdsmoothwindow) {
 616                         char* tempidsmoothwindow = new char[16];
 617                         strcpy(tempidsmoothwindow, "--idsmoothwindow");
 618                         cPara.push_back(tempidsmoothwindow);
 619                         char* tempIdsmoothwindow = new char[idsmoothwindow.length()];
 620                         strcpy(tempIdsmoothwindow, idsmoothwindow.c_str());
 621                         cPara.push_back(tempIdsmoothwindow);
 622                 }
 623
 624                 if (useMinsmoothid) {
 625                         char* tempminsmoothid = new char[13];
 626                         strcpy(tempminsmoothid, "--minsmoothid");
 627                         cPara.push_back(tempminsmoothid);
 628                         char* tempMinsmoothid = new char[minsmoothid.length()];
 629                         strcpy(tempMinsmoothid, minsmoothid.c_str());
 630                         cPara.push_back(tempMinsmoothid);
 631                 }
 632
 633                 if (useMaxp) {
 634                         char* tempmaxp = new char[6];
 635                         strcpy(tempmaxp, "--maxp");
 636                         cPara.push_back(tempmaxp);
 637                         char* tempMaxp = new char[maxp.length()];
 638                         strcpy(tempMaxp, maxp.c_str());
 639                         cPara.push_back(tempMaxp);
 640                 }
 641
 642                 if (!skipgaps) {
 643                         char* tempskipgaps = new char[14];
 644                         strcpy(tempskipgaps, "--[no]skipgaps");
 645                         cPara.push_back(tempskipgaps);
 646                 }
 647
 648                 if (!skipgaps2) {
 649                         char* tempskipgaps2 = new char[15];
 650                         strcpy(tempskipgaps2, "--[no]skipgaps2");
 651                         cPara.push_back(tempskipgaps2);
 652                 }
 653
 654                 if (useMinlen) {
 655                         char* tempminlen = new char[8];
 656                         strcpy(tempminlen, "--minlen");
 657                         cPara.push_back(tempminlen);
 658                         char* tempMinlen = new char[minlen.length()];
 659                         strcpy(tempMinlen, minlen.c_str());
 660                         cPara.push_back(tempMinlen);
 661                 }
 662
 663                 if (useMaxlen) {
 664                         char* tempmaxlen = new char[8];
 665                         strcpy(tempmaxlen, "--maxlen");
 666                         cPara.push_back(tempmaxlen);
 667                         char* tempMaxlen = new char[maxlen.length()];
 668                         strcpy(tempMaxlen, maxlen.c_str());
 669                         cPara.push_back(tempMaxlen);
 670                 }
 671
 672                 if (ucl) {
 673                         char* tempucl = new char[5];
 674                         strcpy(tempucl, "--ucl");
 675                         cPara.push_back(tempucl);
 676                 }
 677
 678                 if (useQueryfract) {
 679                         char* tempqueryfract = new char[12];
 680                         strcpy(tempqueryfract, "--queryfract");
 681                         cPara.push_back(tempqueryfract);
 682                         char* tempQueryfract = new char[queryfract.length()];
 683                         strcpy(tempQueryfract, queryfract.c_str());
 684                         cPara.push_back(tempQueryfract);
 685                 }
 686
 687
 688                 char** uchimeParameters;
 689                 uchimeParameters = new char*[cPara.size()];
 690                 for (int i = 0; i < cPara.size(); i++) {  uchimeParameters[i] = cPara[i];  }
 691                 int numArgs = cPara.size();
 692
 693                 uchime_main(numArgs, uchimeParameters);
 694
 695                 //free memory
 696                 for(int i = 0; i < cPara.size(); i++)  {  delete[] cPara[i];  }
 697                 delete[] uchimeParameters;
 698
 699                 if (m->control_pressed) { return 0; }
 700
 701                 //create accnos file from uchime results
 702                 ifstream in;
 703                 m->openInputFile(outputFName, in);
 704
 705                 ofstream out;
 706                 m->openOutputFile(accnos, out);
 707
 708                 int num = 0;
 709                 while(!in.eof()) {
 710
 711                         if (m->control_pressed) { break; }
 712
 713                         string name = "";
 714                         string chimeraFlag = "";
 715                         in >> chimeraFlag >> name;
 716
 717                         //fix name if needed
 718                         if (templatefile == "self") {
 719                                 name = name.substr(0, name.length()-1); //rip off last /
 720                                 name = name.substr(0, name.find_last_of('/'));
 721                         }
 722
 723                         for (int i = 0; i < 15; i++) {  in >> chimeraFlag; }
 724                         m->gobble(in);
 725
 726                         if (chimeraFlag == "Y") {  out << name << endl; }
 727                         num++;
 728                 }
 729                 in.close();
 730                 out.close();
 731
 732                 return num;
 733         }
 734         catch(exception& e) {
 735                 m->errorOut(e, "ChimeraUchimeCommand", "driver");
 736                 exit(1);
 737         }
 738 }
 739 /**************************************************************************************************/
 740
 741 int ChimeraUchimeCommand::createProcesses(string outputFileName, string filename, string accnos, string alns) {
 742         try {
 743
 744                 processIDS.clear();
 745                 int process = 1;
 746                 int num = 0;
 747 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 748                 //break up file into multiple files
 749                 vector<string> files;
 750                 m->divideFile(filename, processors, files);
 751
 752                 if (m->control_pressed) {  return 0;  }
 753
 754 #ifdef USE_MPI
 755                 int pid, numSeqsPerProcessor;
 756                 int tag = 2001;
 757
 758                 MPI_Status status;
 759                 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
 760                 MPI_Comm_size(MPI_COMM_WORLD, &processors);
 761
 762                 if (pid == 0) { //you are the root process
 763                         num = driver(outputFileName, files[0], accnos, alns);
 764
 765                         if (templatefile != "self") {
 766                                 //wait on chidren
 767                                 for(int j = 1; j < processors; j++) {
 768                                         int temp;
 769                                         MPI_Recv(&temp, 1, MPI_INT, j, tag, MPI_COMM_WORLD, &status);
 770                                         num += temp;
 771
 772                                         m->appendFiles((outputFileName + toString(j) + ".temp"), outputFileName);
 773                                         remove((outputFileName + toString(j) + ".temp").c_str());
 774
 775                                         m->appendFiles((accnos + toString(j) + ".temp"), accnos);
 776                                         remove((accnos + toString(j) + ".temp").c_str());
 777
 778                                         if (chimealns) {
 779                                                 m->appendFiles((alns + toString(j) + ".temp"), alns);
 780                                                 remove((alns + toString(j) + ".temp").c_str());
 781                                         }
 782                                 }
 783                         }
 784                 }else{ //you are a child process
 785                         if (templatefile != "self") { //if template=self we can only use 1 processor
 786                                 num = driver(outputFileName+toString(pid) + ".temp", files[pid], accnos+toString(pid) + ".temp", alns+toString(pid) + ".temp");
 787
 788                                 //send numSeqs to parent
 789                                 MPI_Send(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
 790                         }
 791                 }
 792
 793                 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 794 #else
 795
 796                 //loop through and create all the processes you want
 797                 while (process != processors) {
 798                         int pid = fork();
 799
 800                         if (pid > 0) {
 801                                 processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
 802                                 process++;
 803                         }else if (pid == 0){
 804                                 num = driver(outputFileName + toString(getpid()) + ".temp", files[process], accnos + toString(getpid()) + ".temp", alns + toString(getpid()) + ".temp");
 805
 806                                 //pass numSeqs to parent
 807                                 ofstream out;
 808                                 string tempFile = outputFileName + toString(getpid()) + ".num.temp";
 809                                 m->openOutputFile(tempFile, out);
 810                                 out << num << endl;
 811                                 out.close();
 812
 813                                 exit(0);
 814                         }else {
 815                                 m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
 816                                 for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
 817                                 exit(0);
 818                         }
 819                 }
 820
 821                 //do my part
 822                 num = driver(outputFileName, files[0], accnos, alns);
 823
 824                 //force parent to wait until all the processes are done
 825                 for (int i=0;i<processIDS.size();i++) {
 826                         int temp = processIDS[i];
 827                         wait(&temp);
 828                 }
 829
 830                 for (int i = 0; i < processIDS.size(); i++) {
 831                         ifstream in;
 832                         string tempFile =  outputFileName + toString(processIDS[i]) + ".num.temp";
 833                         m->openInputFile(tempFile, in);
 834                         if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
 835                         in.close(); remove(tempFile.c_str());
 836                 }
 837
 838
 839                 //append output files
 840                 for(int i=0;i<processIDS[i];i++){
 841                         m->appendFiles((outputFileName + toString(processIDS[i]) + ".temp"), outputFileName);
 842                         remove((outputFileName + toString(processIDS[i]) + ".temp").c_str());
 843
 844                         m->appendFiles((accnos + toString(processIDS[i]) + ".temp"), accnos);
 845                         remove((accnos + toString(processIDS[i]) + ".temp").c_str());
 846
 847                         if (chimealns) {
 848                                 m->appendFiles((alns + toString(processIDS[i]) + ".temp"), alns);
 849                                 remove((alns + toString(processIDS[i]) + ".temp").c_str());
 850                         }
 851                 }
 852 #endif
 853                 //get rid of the file pieces.
 854                 for (int i = 0; i < files.size(); i++) { remove(files[i].c_str()); }
 855 #endif
 856                 return num;
 857         }
 858         catch(exception& e) {
 859                 m->errorOut(e, "ChimeraUchimeCommand", "createProcesses");
 860                 exit(1);
 861         }
 862 }
 863
 864 /**************************************************************************************************/
 865