getseqscommand.cpp

   1 /*
   2  *  getseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "getseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "counttable.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> GetSeqsCommand::setParameters(){
  17         try {
  18                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none","fasta",false,false,true); parameters.push_back(pfasta);
  19         CommandParameter pname("name", "InputTypes", "", "", "NameCount", "FNGLT", "none","name",false,false,true); parameters.push_back(pname);
  20         CommandParameter pcount("count", "InputTypes", "", "", "NameCount-CountGroup", "FNGLT", "none","count",false,false,true); parameters.push_back(pcount);
  21                 CommandParameter pgroup("group", "InputTypes", "", "", "CountGroup", "FNGLT", "none","group",false,false,true); parameters.push_back(pgroup);
  22                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none","list",false,false,true); parameters.push_back(plist);
  23                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none","taxonomy",false,false,true); parameters.push_back(ptaxonomy);
  24                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none","alignreport",false,false); parameters.push_back(palignreport);
  25                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none","qfile",false,false); parameters.push_back(pqfile);
  26                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(paccnos);
  27                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pdups);
  28                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  29                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  30                 CommandParameter paccnos2("accnos2", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(paccnos2);
  31
  32                 vector<string> myArray;
  33                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  34                 return myArray;
  35         }
  36         catch(exception& e) {
  37                 m->errorOut(e, "GetSeqsCommand", "setParameters");
  38                 exit(1);
  39         }
  40 }
  41 //**********************************************************************************************************************
  42 string GetSeqsCommand::getHelpString(){
  43         try {
  44                 string helpString = "";
  45                 helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
  46                 helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
  47                 helpString += "The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos unless you have a valid current accnos file, and at least one of the other parameters.\n";
  48                 helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n";
  49                 helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
  50                 helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "GetSeqsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59
  60 //**********************************************************************************************************************
  61 GetSeqsCommand::GetSeqsCommand(){
  62         try {
  63                 abort = true; calledHelp = true;
  64                 setParameters();
  65                 vector<string> tempOutNames;
  66                 outputTypes["fasta"] = tempOutNames;
  67                 outputTypes["taxonomy"] = tempOutNames;
  68                 outputTypes["name"] = tempOutNames;
  69                 outputTypes["group"] = tempOutNames;
  70                 outputTypes["alignreport"] = tempOutNames;
  71                 outputTypes["list"] = tempOutNames;
  72                 outputTypes["qfile"] = tempOutNames;
  73         outputTypes["count"] = tempOutNames;
  74                 outputTypes["accnosreport"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82 string GetSeqsCommand::getOutputPattern(string type) {
  83     try {
  84         string pattern = "";
  85
  86         if (type == "fasta")            {   pattern = "[filename],pick,[extension]";    }
  87         else if (type == "taxonomy")    {   pattern = "[filename],pick,[extension]";    }
  88         else if (type == "name")        {   pattern = "[filename],pick,[extension]";    }
  89         else if (type == "group")       {   pattern = "[filename],pick,[extension]";    }
  90         else if (type == "count")       {   pattern = "[filename],pick,[extension]";    }
  91         else if (type == "list")        {   pattern = "[filename],pick,[extension]";    }
  92         else if (type == "qfile")       {   pattern = "[filename],pick,[extension]";    }
  93         else if (type == "accnosreport")      {   pattern = "[filename],pick.accnos.report";    }
  94         else if (type == "alignreport")      {   pattern = "[filename],pick.align.report";    }
  95         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  96
  97         return pattern;
  98     }
  99     catch(exception& e) {
 100         m->errorOut(e, "GetSeqsCommand", "getOutputPattern");
 101         exit(1);
 102     }
 103 }
 104 //**********************************************************************************************************************
 105 GetSeqsCommand::GetSeqsCommand(string option)  {
 106         try {
 107                 abort = false; calledHelp = false;
 108
 109                 //allow user to run help
 110                 if(option == "help") { help(); abort = true; calledHelp = true; }
 111                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 112
 113                 else {
 114                         vector<string> myArray = setParameters();
 115
 116                         OptionParser parser(option);
 117                         map<string,string> parameters = parser.getParameters();
 118
 119                         ValidParameters validParameter;
 120                         map<string,string>::iterator it;
 121
 122                         //check to make sure all parameters are valid for command
 123                         for (it = parameters.begin(); it != parameters.end(); it++) {
 124                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 125                         }
 126
 127                         //initialize outputTypes
 128                         vector<string> tempOutNames;
 129                         outputTypes["fasta"] = tempOutNames;
 130                         outputTypes["taxonomy"] = tempOutNames;
 131                         outputTypes["name"] = tempOutNames;
 132                         outputTypes["group"] = tempOutNames;
 133                         outputTypes["alignreport"] = tempOutNames;
 134                         outputTypes["list"] = tempOutNames;
 135                         outputTypes["qfile"] = tempOutNames;
 136                         outputTypes["accnosreport"] = tempOutNames;
 137             outputTypes["count"] = tempOutNames;
 138
 139                         //if the user changes the output directory command factory will send this info to us in the output parameter
 140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 141
 142                         //if the user changes the input directory command factory will send this info to us in the output parameter
 143                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 144                         if (inputDir == "not found"){   inputDir = "";          }
 145                         else {
 146                                 string path;
 147                                 it = parameters.find("alignreport");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 153                                 }
 154
 155                                 it = parameters.find("fasta");
 156                                 //user has given a template file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("accnos");
 164                                 //user has given a template file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 169                                 }
 170
 171                                 it = parameters.find("accnos2");
 172                                 //user has given a template file
 173                                 if(it != parameters.end()){
 174                                         path = m->hasPath(it->second);
 175                                         //if the user has not given a path then, add inputdir. else leave path alone.
 176                                         if (path == "") {       parameters["accnos2"] = inputDir + it->second;          }
 177                                 }
 178
 179                                 it = parameters.find("list");
 180                                 //user has given a template file
 181                                 if(it != parameters.end()){
 182                                         path = m->hasPath(it->second);
 183                                         //if the user has not given a path then, add inputdir. else leave path alone.
 184                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 185                                 }
 186
 187                                 it = parameters.find("name");
 188                                 //user has given a template file
 189                                 if(it != parameters.end()){
 190                                         path = m->hasPath(it->second);
 191                                         //if the user has not given a path then, add inputdir. else leave path alone.
 192                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 193                                 }
 194
 195                                 it = parameters.find("group");
 196                                 //user has given a template file
 197                                 if(it != parameters.end()){
 198                                         path = m->hasPath(it->second);
 199                                         //if the user has not given a path then, add inputdir. else leave path alone.
 200                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 201                                 }
 202
 203                                 it = parameters.find("taxonomy");
 204                                 //user has given a template file
 205                                 if(it != parameters.end()){
 206                                         path = m->hasPath(it->second);
 207                                         //if the user has not given a path then, add inputdir. else leave path alone.
 208                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 209                                 }
 210
 211                                 it = parameters.find("qfile");
 212                                 //user has given a template file
 213                                 if(it != parameters.end()){
 214                                         path = m->hasPath(it->second);
 215                                         //if the user has not given a path then, add inputdir. else leave path alone.
 216                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 217                                 }
 218
 219                 it = parameters.find("count");
 220                                 //user has given a template file
 221                                 if(it != parameters.end()){
 222                                         path = m->hasPath(it->second);
 223                                         //if the user has not given a path then, add inputdir. else leave path alone.
 224                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 225                                 }
 226                         }
 227
 228
 229                         //check for required parameters
 230                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 231                         if (accnosfile == "not open") { abort = true; }
 232                         else if (accnosfile == "not found") {
 233                                 accnosfile = m->getAccnosFile();
 234                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 235                                 else {
 236                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 237                                         abort = true;
 238                                 }
 239                         }else { m->setAccnosFile(accnosfile); }
 240
 241                         if (accnosfile2 == "not found") { accnosfile2 = ""; }
 242
 243                         fastafile = validParameter.validFile(parameters, "fasta", true);
 244                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 245                         else if (fastafile == "not found") {  fastafile = "";  }
 246                         else { m->setFastaFile(fastafile); }
 247
 248                         namefile = validParameter.validFile(parameters, "name", true);
 249                         if (namefile == "not open") { namefile = ""; abort = true; }
 250                         else if (namefile == "not found") {  namefile = "";  }
 251                         else { m->setNameFile(namefile); }
 252
 253                         groupfile = validParameter.validFile(parameters, "group", true);
 254                         if (groupfile == "not open") { abort = true; }
 255                         else if (groupfile == "not found") {  groupfile = "";  }
 256                         else { m->setGroupFile(groupfile); }
 257
 258                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 259                         if (alignfile == "not open") { abort = true; }
 260                         else if (alignfile == "not found") {  alignfile = "";  }
 261
 262                         listfile = validParameter.validFile(parameters, "list", true);
 263                         if (listfile == "not open") { abort = true; }
 264                         else if (listfile == "not found") {  listfile = "";  }
 265                         else { m->setListFile(listfile); }
 266
 267                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 268                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 269                         else if (taxfile == "not found") {  taxfile = "";  }
 270                         else { m->setTaxonomyFile(taxfile); }
 271
 272                         qualfile = validParameter.validFile(parameters, "qfile", true);
 273                         if (qualfile == "not open") { abort = true; }
 274                         else if (qualfile == "not found") {  qualfile = "";  }
 275                         else { m->setQualFile(qualfile); }
 276
 277                         accnosfile2 = validParameter.validFile(parameters, "accnos2", true);
 278                         if (accnosfile2 == "not open") { abort = true; }
 279                         else if (accnosfile2 == "not found") {  accnosfile2 = "";  }
 280
 281             countfile = validParameter.validFile(parameters, "count", true);
 282             if (countfile == "not open") { countfile = ""; abort = true; }
 283             else if (countfile == "not found") { countfile = "";  }
 284             else { m->setCountTableFile(countfile); }
 285
 286             if ((namefile != "") && (countfile != "")) {
 287                 m->mothurOut("[ERROR]: you may only use one of the following: name or count."); m->mothurOutEndLine(); abort = true;
 288             }
 289
 290             if ((groupfile != "") && (countfile != "")) {
 291                 m->mothurOut("[ERROR]: you may only use one of the following: group or count."); m->mothurOutEndLine(); abort=true;
 292             }
 293
 294
 295                         string usedDups = "true";
 296                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "true"; usedDups = ""; }
 297                         dups = m->isTrue(temp);
 298
 299                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == "") && (accnosfile2 == "") && (countfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, count, alignreport, taxonomy, quality or listfile."); m->mothurOutEndLine(); abort = true; }
 300
 301             if (countfile == "") {
 302                 if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 303                     vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 304                     parser.getNameFile(files);
 305                 }
 306             }
 307                 }
 308
 309         }
 310         catch(exception& e) {
 311                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 312                 exit(1);
 313         }
 314 }
 315 //**********************************************************************************************************************
 316
 317 int GetSeqsCommand::execute(){
 318         try {
 319
 320                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 321
 322                 //get names you want to keep
 323                 names = m->readAccnos(accnosfile);
 324
 325                 if (m->control_pressed) { return 0; }
 326
 327         if (countfile != "") {
 328             if ((fastafile != "") || (listfile != "") || (taxfile != "")) {
 329                 m->mothurOut("\n[NOTE]: The count file should contain only unique names, so mothur assumes your fasta, list and taxonomy files also contain only uniques.\n\n");
 330             }
 331         }
 332
 333                 //read through the correct file and output lines you want to keep
 334                 if (namefile != "")                     {               readName();                     }
 335                 if (fastafile != "")            {               readFasta();            }
 336                 if (groupfile != "")            {               readGroup();            }
 337         if (countfile != "")            {               readCount();            }
 338                 if (alignfile != "")            {               readAlign();            }
 339                 if (listfile != "")                     {               readList();                     }
 340                 if (taxfile != "")                      {               readTax();                      }
 341                 if (qualfile != "")                     {               readQual();                     }
 342                 if (accnosfile2 != "")          {               compareAccnos();        }
 343
 344         if (m->debug) { runSanityCheck(); }
 345
 346                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);  } return 0; }
 347
 348
 349                 if (outputNames.size() != 0) {
 350                         m->mothurOutEndLine();
 351                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 352                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 353                         m->mothurOutEndLine();
 354
 355                         //set fasta file as new current fastafile
 356                         string current = "";
 357                         itTypes = outputTypes.find("fasta");
 358                         if (itTypes != outputTypes.end()) {
 359                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 360                         }
 361
 362                         itTypes = outputTypes.find("name");
 363                         if (itTypes != outputTypes.end()) {
 364                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 365                         }
 366
 367                         itTypes = outputTypes.find("group");
 368                         if (itTypes != outputTypes.end()) {
 369                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 370                         }
 371
 372                         itTypes = outputTypes.find("list");
 373                         if (itTypes != outputTypes.end()) {
 374                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 375                         }
 376
 377                         itTypes = outputTypes.find("taxonomy");
 378                         if (itTypes != outputTypes.end()) {
 379                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 380                         }
 381
 382                         itTypes = outputTypes.find("qfile");
 383                         if (itTypes != outputTypes.end()) {
 384                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 385                         }
 386
 387             itTypes = outputTypes.find("count");
 388                         if (itTypes != outputTypes.end()) {
 389                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
 390                         }
 391                 }
 392
 393                 return 0;
 394         }
 395
 396         catch(exception& e) {
 397                 m->errorOut(e, "GetSeqsCommand", "execute");
 398                 exit(1);
 399         }
 400 }
 401
 402 //**********************************************************************************************************************
 403 int GetSeqsCommand::readFasta(){
 404         try {
 405                 string thisOutputDir = outputDir;
 406                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 407                 map<string, string> variables;
 408         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(fastafile));
 409         variables["[extension]"] = m->getExtension(fastafile);
 410                 string outputFileName = getOutputFileName("fasta", variables);
 411                 ofstream out;
 412                 m->openOutputFile(outputFileName, out);
 413
 414
 415                 ifstream in;
 416                 m->openInputFile(fastafile, in);
 417                 string name;
 418
 419                 bool wroteSomething = false;
 420                 int selectedCount = 0;
 421
 422         if (m->debug) { set<string> temp; sanity["fasta"] = temp; }
 423
 424                 while(!in.eof()){
 425
 426                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 427
 428                         Sequence currSeq(in);
 429                         name = currSeq.getName();
 430
 431                         if (name != "") {
 432                                 //if this name is in the accnos file
 433                                 if (names.count(name) != 0) {
 434                                         wroteSomething = true;
 435
 436                                         currSeq.printSequence(out);
 437                                         selectedCount++;
 438
 439                     if (m->debug) { sanity["fasta"].insert(name); }
 440                                 }
 441                         }
 442                         m->gobble(in);
 443                 }
 444                 in.close();
 445                 out.close();
 446
 447
 448                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 449                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 450
 451                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 452
 453                 return 0;
 454
 455         }
 456         catch(exception& e) {
 457                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 458                 exit(1);
 459         }
 460 }
 461 //**********************************************************************************************************************
 462 int GetSeqsCommand::readQual(){
 463         try {
 464                 string thisOutputDir = outputDir;
 465                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 466                 map<string, string> variables;
 467         variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(qualfile));
 468         variables["[extension]"] = m->getExtension(qualfile);
 469                 string outputFileName = getOutputFileName("qfile", variables);
 470                 ofstream out;
 471                 m->openOutputFile(outputFileName, out);
 472
 473
 474                 ifstream in;
 475                 m->openInputFile(qualfile, in);
 476                 string name;
 477
 478                 bool wroteSomething = false;
 479                 int selectedCount = 0;
 480
 481         if (m->debug) { set<string> temp; sanity["qual"] = temp; }
 482
 483                 while(!in.eof()){
 484                         string saveName = "";
 485                         string name = "";
 486                         string scores = "";
 487
 488                         in >> name;
 489
 490                         if (name.length() != 0) {
 491                                 saveName = name.substr(1);
 492                                 while (!in.eof())       {
 493                                         char c = in.get();
 494                                         if (c == 10 || c == 13){        break;  }
 495                                         else { name += c; }
 496                                 }
 497                                 m->gobble(in);
 498                         }
 499
 500                         while(in){
 501                                 char letter= in.get();
 502                                 if(letter == '>'){      in.putback(letter);     break;  }
 503                                 else{ scores += letter; }
 504                         }
 505
 506                         m->gobble(in);
 507
 508                         if (names.count(saveName) != 0) {
 509                                 wroteSomething = true;
 510
 511                                 out << name << endl << scores;
 512                                 selectedCount++;
 513                 if (m->debug) { sanity["qual"].insert(name); }
 514                         }
 515
 516                         m->gobble(in);
 517                 }
 518                 in.close();
 519                 out.close();
 520
 521
 522                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 523                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 524
 525                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 526
 527
 528                 return 0;
 529
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "GetSeqsCommand", "readQual");
 533                 exit(1);
 534         }
 535 }
 536 //**********************************************************************************************************************
 537 int GetSeqsCommand::readCount(){
 538         try {
 539                 string thisOutputDir = outputDir;
 540                 if (outputDir == "") {  thisOutputDir += m->hasPath(countfile);  }
 541                 map<string, string> variables;
 542                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(countfile));
 543         variables["[extension]"] = m->getExtension(countfile);
 544                 string outputFileName = getOutputFileName("count", variables);
 545
 546                 ofstream out;
 547                 m->openOutputFile(outputFileName, out);
 548
 549                 ifstream in;
 550                 m->openInputFile(countfile, in);
 551
 552                 bool wroteSomething = false;
 553                 int selectedCount = 0;
 554
 555         string headers = m->getline(in); m->gobble(in);
 556         out << headers << endl;
 557
 558         string name, rest; int thisTotal;
 559         while (!in.eof()) {
 560
 561             if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 562
 563             in >> name; m->gobble(in);
 564             in >> thisTotal; m->gobble(in);
 565             rest = m->getline(in); m->gobble(in);
 566             if (m->debug) { m->mothurOut("[DEBUG]: " + name + '\t' + rest + "\n"); }
 567
 568             if (names.count(name) != 0) {
 569                 out << name << '\t' << thisTotal << '\t' << rest << endl;
 570                 wroteSomething = true;
 571                 selectedCount+= thisTotal;
 572             }
 573         }
 574         in.close();
 575                 out.close();
 576
 577         //check for groups that have been eliminated
 578         CountTable ct;
 579         if (ct.testGroups(outputFileName)) {
 580             ct.readTable(outputFileName);
 581             ct.printTable(outputFileName);
 582         }
 583
 584                 if (wroteSomething == false) {  m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 585                 outputTypes["count"].push_back(outputFileName); outputNames.push_back(outputFileName);
 586
 587                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your count file."); m->mothurOutEndLine();
 588
 589                 return 0;
 590         }
 591         catch(exception& e) {
 592                 m->errorOut(e, "GetSeqsCommand", "readCount");
 593                 exit(1);
 594         }
 595 }
 596
 597 //**********************************************************************************************************************
 598 int GetSeqsCommand::readList(){
 599         try {
 600                 string thisOutputDir = outputDir;
 601                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 602         map<string, string> variables;
 603                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
 604         variables["[extension]"] = m->getExtension(listfile);
 605                 string outputFileName = getOutputFileName("list", variables);
 606                 ofstream out;
 607                 m->openOutputFile(outputFileName, out);
 608
 609                 ifstream in;
 610                 m->openInputFile(listfile, in);
 611
 612                 bool wroteSomething = false;
 613                 int selectedCount = 0;
 614
 615         if (m->debug) { set<string> temp; sanity["list"] = temp; }
 616
 617                 while(!in.eof()){
 618
 619                         selectedCount = 0;
 620
 621                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 622
 623                         //read in list vector
 624                         ListVector list(in);
 625
 626                         //make a new list vector
 627                         ListVector newList;
 628                         newList.setLabel(list.getLabel());
 629
 630                         //for each bin
 631                         for (int i = 0; i < list.getNumBins(); i++) {
 632
 633                                 //parse out names that are in accnos file
 634                                 string binnames = list.get(i);
 635                 vector<string> bnames;
 636                 m->splitAtComma(binnames, bnames);
 637
 638                                 string newNames = "";
 639                 for (int i = 0; i < bnames.size(); i++) {
 640                                         string name = bnames[i];
 641                                         //if that name is in the .accnos file, add it
 642                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++; if (m->debug) { sanity["list"].insert(name); } }
 643                                 }
 644
 645                                 //if there are names in this bin add to new list
 646                                 if (newNames != "") {
 647                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 648                                         newList.push_back(newNames);
 649                                 }
 650                         }
 651
 652                         //print new listvector
 653                         if (newList.getNumBins() != 0) {
 654                                 wroteSomething = true;
 655                                 newList.print(out);
 656                         }
 657
 658                         m->gobble(in);
 659                 }
 660                 in.close();
 661                 out.close();
 662
 663                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 664                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 665
 666                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 667
 668                 return 0;
 669
 670         }
 671         catch(exception& e) {
 672                 m->errorOut(e, "GetSeqsCommand", "readList");
 673                 exit(1);
 674         }
 675 }
 676 //**********************************************************************************************************************
 677 int GetSeqsCommand::readName(){
 678         try {
 679                 string thisOutputDir = outputDir;
 680                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 681         map<string, string> variables;
 682                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(namefile));
 683         variables["[extension]"] = m->getExtension(namefile);
 684                 string outputFileName = getOutputFileName("name", variables);
 685                 ofstream out;
 686                 m->openOutputFile(outputFileName, out);
 687
 688
 689                 ifstream in;
 690                 m->openInputFile(namefile, in);
 691                 string name, firstCol, secondCol;
 692
 693                 bool wroteSomething = false;
 694                 int selectedCount = 0;
 695
 696         if (m->debug) { set<string> temp; sanity["name"] = temp; }
 697         if (m->debug) { set<string> temp; sanity["dupname"] = temp; }
 698
 699                 while(!in.eof()){
 700
 701                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 702
 703                         in >> firstCol;
 704                         in >> secondCol;
 705
 706                         string hold = "";
 707                         if (dups) { hold = secondCol; }
 708
 709                         vector<string> parsedNames;
 710                         m->splitAtComma(secondCol, parsedNames);
 711
 712                         vector<string> validSecond;
 713                         for (int i = 0; i < parsedNames.size(); i++) {
 714                                 if (names.count(parsedNames[i]) != 0) {
 715                                         validSecond.push_back(parsedNames[i]);
 716                     if (m->debug) { sanity["dupname"].insert(parsedNames[i]); }
 717                                 }
 718                         }
 719
 720                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 721                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]); if (m->debug) { sanity["dupname"].insert(parsedNames[i]); } }
 722                                 out << firstCol << '\t' << hold << endl;
 723                                 wroteSomething = true;
 724                                 selectedCount += parsedNames.size();
 725                 if (m->debug) { sanity["name"].insert(firstCol); }
 726                         }else {
 727                                 selectedCount += validSecond.size();
 728
 729                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 730                                 if (names.count(firstCol) != 0) {
 731
 732                                         wroteSomething = true;
 733
 734                                         out << firstCol << '\t';
 735
 736                                         //you know you have at least one valid second since first column is valid
 737                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 738                                         out << validSecond[validSecond.size()-1] << endl;
 739
 740                     if (m->debug) { sanity["name"].insert(firstCol); }
 741
 742
 743                                 //make first name in set you come to first column and then add the remaining names to second column
 744                                 }else {
 745                                         //you want part of this row
 746                                         if (validSecond.size() != 0) {
 747
 748                                                 wroteSomething = true;
 749
 750                                                 out << validSecond[0] << '\t';
 751
 752                                                 //you know you have at least one valid second since first column is valid
 753                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 754                                                 out << validSecond[validSecond.size()-1] << endl;
 755
 756                         if (m->debug) { sanity["name"].insert(validSecond[0]); }
 757                                         }
 758                                 }
 759                         }
 760                         m->gobble(in);
 761                 }
 762                 in.close();
 763                 out.close();
 764
 765                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 766                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 767
 768                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 769
 770                 return 0;
 771
 772         }
 773         catch(exception& e) {
 774                 m->errorOut(e, "GetSeqsCommand", "readName");
 775                 exit(1);
 776         }
 777 }
 778
 779 //**********************************************************************************************************************
 780 int GetSeqsCommand::readGroup(){
 781         try {
 782                 string thisOutputDir = outputDir;
 783                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 784                 map<string, string> variables;
 785                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(groupfile));
 786         variables["[extension]"] = m->getExtension(groupfile);
 787                 string outputFileName = getOutputFileName("group", variables);
 788                 ofstream out;
 789                 m->openOutputFile(outputFileName, out);
 790
 791
 792                 ifstream in;
 793                 m->openInputFile(groupfile, in);
 794                 string name, group;
 795
 796                 bool wroteSomething = false;
 797                 int selectedCount = 0;
 798
 799         if (m->debug) { set<string> temp; sanity["group"] = temp; }
 800
 801                 while(!in.eof()){
 802
 803                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 804
 805
 806                         in >> name;                             //read from first column
 807                         in >> group;                    //read from second column
 808
 809                         //if this name is in the accnos file
 810                         if (names.count(name) != 0) {
 811                                 wroteSomething = true;
 812
 813                                 out << name << '\t' << group << endl;
 814                                 selectedCount++;
 815
 816                 if (m->debug) {  sanity["group"].insert(name); }
 817                         }
 818
 819                         m->gobble(in);
 820                 }
 821                 in.close();
 822                 out.close();
 823
 824                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 825                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 826
 827                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 828
 829
 830                 return 0;
 831
 832         }
 833         catch(exception& e) {
 834                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 835                 exit(1);
 836         }
 837 }
 838 //**********************************************************************************************************************
 839 int GetSeqsCommand::readTax(){
 840         try {
 841                 string thisOutputDir = outputDir;
 842                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 843                 map<string, string> variables;
 844                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(taxfile));
 845         variables["[extension]"] = m->getExtension(taxfile);
 846                 string outputFileName = getOutputFileName("taxonomy", variables);
 847                 ofstream out;
 848                 m->openOutputFile(outputFileName, out);
 849
 850                 ifstream in;
 851                 m->openInputFile(taxfile, in);
 852                 string name, tax;
 853
 854                 bool wroteSomething = false;
 855                 int selectedCount = 0;
 856
 857         if (m->debug) { set<string> temp; sanity["tax"] = temp; }
 858
 859                 while(!in.eof()){
 860
 861                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 862
 863                         in >> name;                             //read from first column
 864                         in >> tax;                      //read from second column
 865
 866                         //if this name is in the accnos file
 867                         if (names.count(name) != 0) {
 868                                 wroteSomething = true;
 869
 870                                 out << name << '\t' << tax << endl;
 871                                 selectedCount++;
 872
 873                 if (m->debug) { sanity["tax"].insert(name); }
 874                         }
 875
 876                         m->gobble(in);
 877                 }
 878                 in.close();
 879                 out.close();
 880
 881                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 882                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 883
 884                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 885
 886                 return 0;
 887
 888         }
 889         catch(exception& e) {
 890                 m->errorOut(e, "GetSeqsCommand", "readTax");
 891                 exit(1);
 892         }
 893 }
 894 //**********************************************************************************************************************
 895 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
 896 int GetSeqsCommand::readAlign(){
 897         try {
 898                 string thisOutputDir = outputDir;
 899                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 900         map<string, string> variables;
 901                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(alignfile));
 902                 string outputFileName = getOutputFileName("alignreport", variables);
 903                 ofstream out;
 904                 m->openOutputFile(outputFileName, out);
 905
 906
 907                 ifstream in;
 908                 m->openInputFile(alignfile, in);
 909                 string name, junk;
 910
 911                 bool wroteSomething = false;
 912                 int selectedCount = 0;
 913
 914                 //read column headers
 915                 for (int i = 0; i < 16; i++) {
 916                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 917                         else                    {       break;                  }
 918                 }
 919                 out << endl;
 920
 921                 while(!in.eof()){
 922
 923                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 924
 925
 926                         in >> name;                             //read from first column
 927
 928                         //if this name is in the accnos file
 929                         if (names.count(name) != 0) {
 930                                 wroteSomething = true;
 931                                 selectedCount++;
 932
 933                                 out << name << '\t';
 934
 935                                 //read rest
 936                                 for (int i = 0; i < 15; i++) {
 937                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 938                                         else                    {       break;                  }
 939                                 }
 940                                 out << endl;
 941
 942                         }else {//still read just don't do anything with it
 943                                 //read rest
 944                                 for (int i = 0; i < 15; i++) {
 945                                         if (!in.eof())  {       in >> junk;             }
 946                                         else                    {       break;                  }
 947                                 }
 948                         }
 949
 950                         m->gobble(in);
 951                 }
 952                 in.close();
 953                 out.close();
 954
 955                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 956                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 957
 958                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 959
 960                 return 0;
 961
 962         }
 963         catch(exception& e) {
 964                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 965                 exit(1);
 966         }
 967 }
 968 //**********************************************************************************************************************
 969 //just looking at common mistakes.
 970 int GetSeqsCommand::runSanityCheck(){
 971         try {
 972         string thisOutputDir = outputDir;
 973         if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 974         string filename = outputDir + "get.seqs.debug.report";
 975
 976         ofstream out;
 977                 m->openOutputFile(filename, out);
 978
 979
 980         //compare fasta, name, qual and taxonomy if given to make sure they contain the same seqs
 981         if (fastafile != "") {
 982             if (namefile != "") { //compare with fasta
 983                 if (sanity["fasta"] != sanity["name"]) { //create mismatch file
 984                     createMisMatchFile(out, fastafile, namefile, sanity["fasta"], sanity["name"]);
 985                 }
 986             }
 987             if (qualfile != "") {
 988                 if (sanity["fasta"] != sanity["qual"]) { //create mismatch file
 989                     createMisMatchFile(out, fastafile, qualfile, sanity["fasta"], sanity["qual"]);
 990                 }
 991             }
 992             if (taxfile != "") {
 993                 if (sanity["fasta"] != sanity["tax"]) { //create mismatch file
 994                     createMisMatchFile(out, fastafile, taxfile, sanity["fasta"], sanity["tax"]);
 995                 }
 996             }
 997         }
 998
 999         //compare dupnames, groups and list if given to make sure they match
1000         if (namefile != "") {
1001             if (groupfile != "") {
1002                 if (sanity["dupname"] != sanity["group"]) { //create mismatch file
1003                     createMisMatchFile(out, namefile, groupfile, sanity["dupname"], sanity["group"]);
1004                 }
1005             }
1006             if (listfile != "") {
1007                 if (sanity["dupname"] != sanity["list"]) { //create mismatch file
1008                     createMisMatchFile(out, namefile, listfile, sanity["dupname"], sanity["list"]);
1009                 }
1010             }
1011         }else{
1012
1013             if ((groupfile != "") && (fastafile != "")) {
1014                 if (sanity["fasta"] != sanity["group"]) { //create mismatch file
1015                     createMisMatchFile(out, fastafile, groupfile, sanity["fasta"], sanity["group"]);
1016                 }
1017             }
1018         }
1019
1020         out.close();
1021
1022         if (m->isBlank(filename)) { m->mothurRemove(filename); }
1023         else { m->mothurOut("\n[DEBUG]: " + filename + " contains the file mismatches.\n");outputNames.push_back(filename); outputTypes["debug"].push_back(filename); }
1024
1025         return 0;
1026     }
1027         catch(exception& e) {
1028                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
1029                 exit(1);
1030         }
1031 }
1032 //**********************************************************************************************************************
1033 //just looking at common mistakes.
1034 int GetSeqsCommand::createMisMatchFile(ofstream& out, string filename1, string filename2, set<string> set1, set<string> set2){
1035         try {
1036         out << "****************************************" << endl << endl;
1037         out << "Names unique to " << filename1 << ":\n";
1038
1039         //remove names in set1 that are also in set2
1040         for (set<string>::iterator it = set1.begin(); it != set1.end();) {
1041             string name = *it;
1042
1043             if (set2.count(name) == 0)  { out << name << endl;  } //name unique to set1
1044             else                        { set2.erase(name);     } //you are in both so erase
1045             set1.erase(it++);
1046         }
1047
1048         out << "\nNames unique to " << filename2 << ":\n";
1049         //output results
1050         for (set<string>::iterator it = set2.begin(); it != set2.end(); it++) {  out << *it << endl;  }
1051
1052         out << "****************************************" << endl << endl;
1053
1054         return 0;
1055     }
1056         catch(exception& e) {
1057                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
1058                 exit(1);
1059         }
1060 }
1061 //**********************************************************************************************************************
1062
1063 int GetSeqsCommand::compareAccnos(){
1064         try {
1065
1066                 string thisOutputDir = outputDir;
1067                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
1068         map<string, string> variables;
1069                 variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile));
1070                 string outputFileName = getOutputFileName("accnosreport", variables);
1071
1072                 ofstream out;
1073                 m->openOutputFile(outputFileName, out);
1074
1075                 ifstream in;
1076                 m->openInputFile(accnosfile2, in);
1077                 string name;
1078
1079                 set<string> namesAccnos2;
1080                 set<string> namesDups;
1081                 set<string> namesAccnos = names;
1082
1083                 map<string, int> nameCount;
1084
1085                 if (namefile != "") {
1086                         ifstream inName;
1087                         m->openInputFile(namefile, inName);
1088
1089
1090                         while(!inName.eof()){
1091
1092                                 if (m->control_pressed) { inName.close(); return 0; }
1093
1094                                 string thisname, repnames;
1095
1096                                 inName >> thisname;             m->gobble(inName);              //read from first column
1097                                 inName >> repnames;                     //read from second column
1098
1099                                 int num = m->getNumNames(repnames);
1100                                 nameCount[thisname] = num;
1101
1102                                 m->gobble(inName);
1103                         }
1104                         inName.close();
1105                 }
1106
1107                 while(!in.eof()){
1108                         in >> name;
1109
1110                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
1111                                 int pos = name.find_last_of('_');
1112                                 string tempName = name;
1113                                 if (pos != string::npos) {  tempName = tempName.substr(pos+1); cout << tempName << endl; }
1114                                 if (namesAccnos.count(tempName) == 0){
1115                                         namesAccnos2.insert(name);
1116                                 }else { //you are in both so erase
1117                                         namesAccnos.erase(name);
1118                                         namesDups.insert(name);
1119                                 }
1120                         }else { //you are in both so erase
1121                                 namesAccnos.erase(name);
1122                                 namesDups.insert(name);
1123                         }
1124
1125                         m->gobble(in);
1126                 }
1127                 in.close();
1128
1129                 out << "Names in both files : " + toString(namesDups.size()) << endl;
1130                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
1131
1132                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
1133                         out << (*it);
1134                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1135                         out << endl;
1136                 }
1137
1138                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
1139                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
1140
1141                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
1142                         out << (*it);
1143                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1144                         out << endl;
1145                 }
1146
1147                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
1148                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
1149
1150                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
1151                         out << (*it);
1152                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1153                         out << endl;
1154                 }
1155
1156                 out.close();
1157
1158                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
1159
1160                 return 0;
1161
1162         }
1163         catch(exception& e) {
1164                 m->errorOut(e, "GetSeqsCommand", "compareAccnos");
1165                 exit(1);
1166         }
1167 }
1168
1169
1170 //**********************************************************************************************************************
1171