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
  14 //**********************************************************************************************************************
  15 vector<string> GetSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  22                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
  23                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pqfile);
  24                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos);
  25                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
  26                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  27                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  28                 CommandParameter paccnos2("accnos2", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos2);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "GetSeqsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string GetSeqsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, list, taxonomy, quality or alignreport file.\n";
  44                 helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
  45                 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";
  46                 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";
  47                 helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
  48                 helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
  49                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  50                 return helpString;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "GetSeqsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57
  58 //**********************************************************************************************************************
  59 GetSeqsCommand::GetSeqsCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63                 vector<string> tempOutNames;
  64                 outputTypes["fasta"] = tempOutNames;
  65                 outputTypes["taxonomy"] = tempOutNames;
  66                 outputTypes["name"] = tempOutNames;
  67                 outputTypes["group"] = tempOutNames;
  68                 outputTypes["alignreport"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["qfile"] = tempOutNames;
  71                 outputTypes["accnosreport"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 GetSeqsCommand::GetSeqsCommand(string option)  {
  80         try {
  81                 abort = false; calledHelp = false;
  82
  83                 //allow user to run help
  84                 if(option == "help") { help(); abort = true; calledHelp = true; }
  85                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  86
  87                 else {
  88                         vector<string> myArray = setParameters();
  89
  90                         OptionParser parser(option);
  91                         map<string,string> parameters = parser.getParameters();
  92
  93                         ValidParameters validParameter;
  94                         map<string,string>::iterator it;
  95
  96                         //check to make sure all parameters are valid for command
  97                         for (it = parameters.begin(); it != parameters.end(); it++) {
  98                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  99                         }
 100
 101                         //initialize outputTypes
 102                         vector<string> tempOutNames;
 103                         outputTypes["fasta"] = tempOutNames;
 104                         outputTypes["taxonomy"] = tempOutNames;
 105                         outputTypes["name"] = tempOutNames;
 106                         outputTypes["group"] = tempOutNames;
 107                         outputTypes["alignreport"] = tempOutNames;
 108                         outputTypes["list"] = tempOutNames;
 109                         outputTypes["qfile"] = tempOutNames;
 110                         outputTypes["accnosreport"] = tempOutNames;
 111
 112                         //if the user changes the output directory command factory will send this info to us in the output parameter
 113                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 114
 115                         //if the user changes the input directory command factory will send this info to us in the output parameter
 116                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 117                         if (inputDir == "not found"){   inputDir = "";          }
 118                         else {
 119                                 string path;
 120                                 it = parameters.find("alignreport");
 121                                 //user has given a template file
 122                                 if(it != parameters.end()){
 123                                         path = m->hasPath(it->second);
 124                                         //if the user has not given a path then, add inputdir. else leave path alone.
 125                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 126                                 }
 127
 128                                 it = parameters.find("fasta");
 129                                 //user has given a template file
 130                                 if(it != parameters.end()){
 131                                         path = m->hasPath(it->second);
 132                                         //if the user has not given a path then, add inputdir. else leave path alone.
 133                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 134                                 }
 135
 136                                 it = parameters.find("accnos");
 137                                 //user has given a template file
 138                                 if(it != parameters.end()){
 139                                         path = m->hasPath(it->second);
 140                                         //if the user has not given a path then, add inputdir. else leave path alone.
 141                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 142                                 }
 143
 144                                 it = parameters.find("accnos2");
 145                                 //user has given a template file
 146                                 if(it != parameters.end()){
 147                                         path = m->hasPath(it->second);
 148                                         //if the user has not given a path then, add inputdir. else leave path alone.
 149                                         if (path == "") {       parameters["accnos2"] = inputDir + it->second;          }
 150                                 }
 151
 152                                 it = parameters.find("list");
 153                                 //user has given a template file
 154                                 if(it != parameters.end()){
 155                                         path = m->hasPath(it->second);
 156                                         //if the user has not given a path then, add inputdir. else leave path alone.
 157                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 158                                 }
 159
 160                                 it = parameters.find("name");
 161                                 //user has given a template file
 162                                 if(it != parameters.end()){
 163                                         path = m->hasPath(it->second);
 164                                         //if the user has not given a path then, add inputdir. else leave path alone.
 165                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 166                                 }
 167
 168                                 it = parameters.find("group");
 169                                 //user has given a template file
 170                                 if(it != parameters.end()){
 171                                         path = m->hasPath(it->second);
 172                                         //if the user has not given a path then, add inputdir. else leave path alone.
 173                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 174                                 }
 175
 176                                 it = parameters.find("taxonomy");
 177                                 //user has given a template file
 178                                 if(it != parameters.end()){
 179                                         path = m->hasPath(it->second);
 180                                         //if the user has not given a path then, add inputdir. else leave path alone.
 181                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 182                                 }
 183
 184                                 it = parameters.find("qfile");
 185                                 //user has given a template file
 186                                 if(it != parameters.end()){
 187                                         path = m->hasPath(it->second);
 188                                         //if the user has not given a path then, add inputdir. else leave path alone.
 189                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 190                                 }
 191                         }
 192
 193
 194                         //check for required parameters
 195                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 196                         if (accnosfile == "not open") { abort = true; }
 197                         else if (accnosfile == "not found") {
 198                                 accnosfile = m->getAccnosFile();
 199                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 200                                 else {
 201                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 202                                         abort = true;
 203                                 }
 204                         }else { m->setAccnosFile(accnosfile); }
 205
 206                         if (accnosfile2 == "not found") { accnosfile2 = ""; }
 207
 208                         fastafile = validParameter.validFile(parameters, "fasta", true);
 209                         if (fastafile == "not open") { abort = true; }
 210                         else if (fastafile == "not found") {  fastafile = "";  }
 211                         else { m->setFastaFile(fastafile); }
 212
 213                         namefile = validParameter.validFile(parameters, "name", true);
 214                         if (namefile == "not open") { abort = true; }
 215                         else if (namefile == "not found") {  namefile = "";  }
 216                         else { m->setNameFile(namefile); }
 217
 218                         groupfile = validParameter.validFile(parameters, "group", true);
 219                         if (groupfile == "not open") { abort = true; }
 220                         else if (groupfile == "not found") {  groupfile = "";  }
 221                         else { m->setGroupFile(groupfile); }
 222
 223                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 224                         if (alignfile == "not open") { abort = true; }
 225                         else if (alignfile == "not found") {  alignfile = "";  }
 226
 227                         listfile = validParameter.validFile(parameters, "list", true);
 228                         if (listfile == "not open") { abort = true; }
 229                         else if (listfile == "not found") {  listfile = "";  }
 230                         else { m->setListFile(listfile); }
 231
 232                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 233                         if (taxfile == "not open") { abort = true; }
 234                         else if (taxfile == "not found") {  taxfile = "";  }
 235                         else { m->setTaxonomyFile(taxfile); }
 236
 237                         qualfile = validParameter.validFile(parameters, "qfile", true);
 238                         if (qualfile == "not open") { abort = true; }
 239                         else if (qualfile == "not found") {  qualfile = "";  }
 240                         else { m->setQualFile(qualfile); }
 241
 242                         accnosfile2 = validParameter.validFile(parameters, "accnos2", true);
 243                         if (accnosfile2 == "not open") { abort = true; }
 244                         else if (accnosfile2 == "not found") {  accnosfile2 = "";  }
 245
 246
 247                         string usedDups = "true";
 248                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "true"; usedDups = ""; }
 249                         dups = m->isTrue(temp);
 250
 251                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == "") && (accnosfile2 == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy, quality or listfile."); m->mothurOutEndLine(); abort = true; }
 252                 }
 253
 254         }
 255         catch(exception& e) {
 256                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 257                 exit(1);
 258         }
 259 }
 260 //**********************************************************************************************************************
 261
 262 int GetSeqsCommand::execute(){
 263         try {
 264
 265                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 266
 267                 //get names you want to keep
 268                 readAccnos();
 269
 270                 if (m->control_pressed) { return 0; }
 271
 272                 //read through the correct file and output lines you want to keep
 273                 if (namefile != "")                     {               readName();                     }
 274                 if (fastafile != "")            {               readFasta();            }
 275                 if (groupfile != "")            {               readGroup();            }
 276                 if (alignfile != "")            {               readAlign();            }
 277                 if (listfile != "")                     {               readList();                     }
 278                 if (taxfile != "")                      {               readTax();                      }
 279                 if (qualfile != "")                     {               readQual();                     }
 280                 if (accnosfile2 != "")          {               compareAccnos();        }
 281
 282                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);  } return 0; }
 283
 284
 285                 if (outputNames.size() != 0) {
 286                         m->mothurOutEndLine();
 287                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 288                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 289                         m->mothurOutEndLine();
 290
 291                         //set fasta file as new current fastafile
 292                         string current = "";
 293                         itTypes = outputTypes.find("fasta");
 294                         if (itTypes != outputTypes.end()) {
 295                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 296                         }
 297
 298                         itTypes = outputTypes.find("name");
 299                         if (itTypes != outputTypes.end()) {
 300                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 301                         }
 302
 303                         itTypes = outputTypes.find("group");
 304                         if (itTypes != outputTypes.end()) {
 305                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 306                         }
 307
 308                         itTypes = outputTypes.find("list");
 309                         if (itTypes != outputTypes.end()) {
 310                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 311                         }
 312
 313                         itTypes = outputTypes.find("taxonomy");
 314                         if (itTypes != outputTypes.end()) {
 315                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 316                         }
 317
 318                         itTypes = outputTypes.find("qfile");
 319                         if (itTypes != outputTypes.end()) {
 320                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 321                         }
 322
 323                 }
 324
 325                 return 0;
 326         }
 327
 328         catch(exception& e) {
 329                 m->errorOut(e, "GetSeqsCommand", "execute");
 330                 exit(1);
 331         }
 332 }
 333
 334 //**********************************************************************************************************************
 335 int GetSeqsCommand::readFasta(){
 336         try {
 337                 string thisOutputDir = outputDir;
 338                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 339                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 340                 ofstream out;
 341                 m->openOutputFile(outputFileName, out);
 342
 343
 344                 ifstream in;
 345                 m->openInputFile(fastafile, in);
 346                 string name;
 347
 348                 bool wroteSomething = false;
 349                 int selectedCount = 0;
 350
 351                 while(!in.eof()){
 352
 353                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 354
 355                         Sequence currSeq(in);
 356                         name = currSeq.getName();
 357
 358                         if (name != "") {
 359                                 //if this name is in the accnos file
 360                                 if (names.count(name) != 0) {
 361                                         wroteSomething = true;
 362
 363                                         currSeq.printSequence(out);
 364                                         selectedCount++;
 365                                 }
 366                         }
 367                         m->gobble(in);
 368                 }
 369                 in.close();
 370                 out.close();
 371
 372
 373                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 374                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 375
 376                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 377
 378                 return 0;
 379
 380         }
 381         catch(exception& e) {
 382                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 383                 exit(1);
 384         }
 385 }
 386 //**********************************************************************************************************************
 387 int GetSeqsCommand::readQual(){
 388         try {
 389                 string thisOutputDir = outputDir;
 390                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 391                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 392                 ofstream out;
 393                 m->openOutputFile(outputFileName, out);
 394
 395
 396                 ifstream in;
 397                 m->openInputFile(qualfile, in);
 398                 string name;
 399
 400                 bool wroteSomething = false;
 401                 int selectedCount = 0;
 402
 403
 404                 while(!in.eof()){
 405                         string saveName = "";
 406                         string name = "";
 407                         string scores = "";
 408
 409                         in >> name;
 410
 411                         if (name.length() != 0) {
 412                                 saveName = name.substr(1);
 413                                 while (!in.eof())       {
 414                                         char c = in.get();
 415                                         if (c == 10 || c == 13){        break;  }
 416                                         else { name += c; }
 417                                 }
 418                                 m->gobble(in);
 419                         }
 420
 421                         while(in){
 422                                 char letter= in.get();
 423                                 if(letter == '>'){      in.putback(letter);     break;  }
 424                                 else{ scores += letter; }
 425                         }
 426
 427                         m->gobble(in);
 428
 429                         if (names.count(saveName) != 0) {
 430                                 wroteSomething = true;
 431
 432                                 out << name << endl << scores;
 433                                 selectedCount++;
 434                         }
 435
 436                         m->gobble(in);
 437                 }
 438                 in.close();
 439                 out.close();
 440
 441
 442                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 443                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 444
 445                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 446
 447
 448                 return 0;
 449
 450         }
 451         catch(exception& e) {
 452                 m->errorOut(e, "GetSeqsCommand", "readQual");
 453                 exit(1);
 454         }
 455 }
 456 //**********************************************************************************************************************
 457 int GetSeqsCommand::readList(){
 458         try {
 459                 string thisOutputDir = outputDir;
 460                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 461                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 462                 ofstream out;
 463                 m->openOutputFile(outputFileName, out);
 464
 465                 ifstream in;
 466                 m->openInputFile(listfile, in);
 467
 468                 bool wroteSomething = false;
 469                 int selectedCount = 0;
 470
 471                 while(!in.eof()){
 472
 473                         selectedCount = 0;
 474
 475                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 476
 477                         //read in list vector
 478                         ListVector list(in);
 479
 480                         //make a new list vector
 481                         ListVector newList;
 482                         newList.setLabel(list.getLabel());
 483
 484                         //for each bin
 485                         for (int i = 0; i < list.getNumBins(); i++) {
 486
 487                                 //parse out names that are in accnos file
 488                                 string binnames = list.get(i);
 489
 490                                 string newNames = "";
 491                                 while (binnames.find_first_of(',') != -1) {
 492                                         string name = binnames.substr(0,binnames.find_first_of(','));
 493                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 494
 495                                         //if that name is in the .accnos file, add it
 496                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++; }
 497                                 }
 498
 499                                 //get last name
 500                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++; }
 501
 502                                 //if there are names in this bin add to new list
 503                                 if (newNames != "") {
 504                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 505                                         newList.push_back(newNames);
 506                                 }
 507                         }
 508
 509                         //print new listvector
 510                         if (newList.getNumBins() != 0) {
 511                                 wroteSomething = true;
 512                                 newList.print(out);
 513                         }
 514
 515                         m->gobble(in);
 516                 }
 517                 in.close();
 518                 out.close();
 519
 520                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 521                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 522
 523                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 524
 525                 return 0;
 526
 527         }
 528         catch(exception& e) {
 529                 m->errorOut(e, "GetSeqsCommand", "readList");
 530                 exit(1);
 531         }
 532 }
 533 //**********************************************************************************************************************
 534 int GetSeqsCommand::readName(){
 535         try {
 536                 string thisOutputDir = outputDir;
 537                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 538                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 539                 ofstream out;
 540                 m->openOutputFile(outputFileName, out);
 541
 542
 543                 ifstream in;
 544                 m->openInputFile(namefile, in);
 545                 string name, firstCol, secondCol;
 546
 547                 bool wroteSomething = false;
 548                 int selectedCount = 0;
 549
 550                 while(!in.eof()){
 551
 552                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 553
 554                         in >> firstCol;
 555                         in >> secondCol;
 556
 557                         string hold = "";
 558                         if (dups) { hold = secondCol; }
 559
 560                         vector<string> parsedNames;
 561                         m->splitAtComma(secondCol, parsedNames);
 562
 563                         vector<string> validSecond;
 564                         for (int i = 0; i < parsedNames.size(); i++) {
 565                                 if (names.count(parsedNames[i]) != 0) {
 566                                         validSecond.push_back(parsedNames[i]);
 567                                 }
 568                         }
 569
 570                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 571                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 572                                 out << firstCol << '\t' << hold << endl;
 573                                 wroteSomething = true;
 574                                 selectedCount += parsedNames.size();
 575                         }else {
 576                                 selectedCount += validSecond.size();
 577
 578                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 579                                 if (names.count(firstCol) != 0) {
 580
 581                                         wroteSomething = true;
 582
 583                                         out << firstCol << '\t';
 584
 585                                         //you know you have at least one valid second since first column is valid
 586                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 587                                         out << validSecond[validSecond.size()-1] << endl;
 588
 589
 590                                 //make first name in set you come to first column and then add the remaining names to second column
 591                                 }else {
 592                                         //you want part of this row
 593                                         if (validSecond.size() != 0) {
 594
 595                                                 wroteSomething = true;
 596
 597                                                 out << validSecond[0] << '\t';
 598
 599                                                 //you know you have at least one valid second since first column is valid
 600                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 601                                                 out << validSecond[validSecond.size()-1] << endl;
 602                                         }
 603                                 }
 604                         }
 605                         m->gobble(in);
 606                 }
 607                 in.close();
 608                 out.close();
 609
 610                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 611                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 612
 613                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 614
 615                 return 0;
 616
 617         }
 618         catch(exception& e) {
 619                 m->errorOut(e, "GetSeqsCommand", "readName");
 620                 exit(1);
 621         }
 622 }
 623
 624 //**********************************************************************************************************************
 625 int GetSeqsCommand::readGroup(){
 626         try {
 627                 string thisOutputDir = outputDir;
 628                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 629                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 630                 ofstream out;
 631                 m->openOutputFile(outputFileName, out);
 632
 633
 634                 ifstream in;
 635                 m->openInputFile(groupfile, in);
 636                 string name, group;
 637
 638                 bool wroteSomething = false;
 639                 int selectedCount = 0;
 640
 641                 while(!in.eof()){
 642
 643                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 644
 645
 646                         in >> name;                             //read from first column
 647                         in >> group;                    //read from second column
 648
 649                         //if this name is in the accnos file
 650                         if (names.count(name) != 0) {
 651                                 wroteSomething = true;
 652
 653                                 out << name << '\t' << group << endl;
 654                                 selectedCount++;
 655                         }
 656
 657                         m->gobble(in);
 658                 }
 659                 in.close();
 660                 out.close();
 661
 662                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 663                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 664
 665                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 666
 667
 668                 return 0;
 669
 670         }
 671         catch(exception& e) {
 672                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 673                 exit(1);
 674         }
 675 }
 676 //**********************************************************************************************************************
 677 int GetSeqsCommand::readTax(){
 678         try {
 679                 string thisOutputDir = outputDir;
 680                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 681                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 682                 ofstream out;
 683                 m->openOutputFile(outputFileName, out);
 684
 685                 ifstream in;
 686                 m->openInputFile(taxfile, in);
 687                 string name, tax;
 688
 689                 bool wroteSomething = false;
 690                 int selectedCount = 0;
 691
 692                 while(!in.eof()){
 693
 694                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 695
 696                         in >> name;                             //read from first column
 697                         in >> tax;                      //read from second column
 698
 699                         //if this name is in the accnos file
 700                         if (names.count(name) != 0) {
 701                                 wroteSomething = true;
 702
 703                                 out << name << '\t' << tax << endl;
 704                                 selectedCount++;
 705                         }
 706
 707                         m->gobble(in);
 708                 }
 709                 in.close();
 710                 out.close();
 711
 712                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 713                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 714
 715                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 716
 717                 return 0;
 718
 719         }
 720         catch(exception& e) {
 721                 m->errorOut(e, "GetSeqsCommand", "readTax");
 722                 exit(1);
 723         }
 724 }
 725 //**********************************************************************************************************************
 726 //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
 727 int GetSeqsCommand::readAlign(){
 728         try {
 729                 string thisOutputDir = outputDir;
 730                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 731                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 732                 ofstream out;
 733                 m->openOutputFile(outputFileName, out);
 734
 735
 736                 ifstream in;
 737                 m->openInputFile(alignfile, in);
 738                 string name, junk;
 739
 740                 bool wroteSomething = false;
 741                 int selectedCount = 0;
 742
 743                 //read column headers
 744                 for (int i = 0; i < 16; i++) {
 745                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 746                         else                    {       break;                  }
 747                 }
 748                 out << endl;
 749
 750                 while(!in.eof()){
 751
 752                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 753
 754
 755                         in >> name;                             //read from first column
 756
 757                         //if this name is in the accnos file
 758                         if (names.count(name) != 0) {
 759                                 wroteSomething = true;
 760                                 selectedCount++;
 761
 762                                 out << name << '\t';
 763
 764                                 //read rest
 765                                 for (int i = 0; i < 15; i++) {
 766                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 767                                         else                    {       break;                  }
 768                                 }
 769                                 out << endl;
 770
 771                         }else {//still read just don't do anything with it
 772                                 //read rest
 773                                 for (int i = 0; i < 15; i++) {
 774                                         if (!in.eof())  {       in >> junk;             }
 775                                         else                    {       break;                  }
 776                                 }
 777                         }
 778
 779                         m->gobble(in);
 780                 }
 781                 in.close();
 782                 out.close();
 783
 784                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 785                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 786
 787                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 788
 789                 return 0;
 790
 791         }
 792         catch(exception& e) {
 793                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 794                 exit(1);
 795         }
 796 }
 797 //**********************************************************************************************************************
 798
 799 int GetSeqsCommand::readAccnos(){
 800         try {
 801
 802                 ifstream in;
 803                 m->openInputFile(accnosfile, in);
 804                 string name;
 805
 806                 while(!in.eof()){
 807                         in >> name;
 808
 809                         names.insert(name);
 810
 811                         m->gobble(in);
 812                 }
 813                 in.close();
 814
 815                 return 0;
 816
 817         }
 818         catch(exception& e) {
 819                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 820                 exit(1);
 821         }
 822 }
 823 //**********************************************************************************************************************
 824
 825 int GetSeqsCommand::compareAccnos(){
 826         try {
 827
 828                 string thisOutputDir = outputDir;
 829                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
 830                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile)) + "accnos.report";
 831                 ofstream out;
 832                 m->openOutputFile(outputFileName, out);
 833
 834                 ifstream in;
 835                 m->openInputFile(accnosfile2, in);
 836                 string name;
 837
 838                 set<string> namesAccnos2;
 839                 set<string> namesDups;
 840                 set<string> namesAccnos = names;
 841
 842                 map<string, int> nameCount;
 843
 844                 if (namefile != "") {
 845                         ifstream inName;
 846                         m->openInputFile(namefile, inName);
 847
 848
 849                         while(!inName.eof()){
 850
 851                                 if (m->control_pressed) { inName.close(); return 0; }
 852
 853                                 string thisname, repnames;
 854
 855                                 inName >> thisname;             m->gobble(inName);              //read from first column
 856                                 inName >> repnames;                     //read from second column
 857
 858                                 int num = m->getNumNames(repnames);
 859                                 nameCount[thisname] = num;
 860
 861                                 m->gobble(inName);
 862                         }
 863                         inName.close();
 864                 }
 865
 866                 while(!in.eof()){
 867                         in >> name;
 868
 869                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
 870                                 int pos = name.find_last_of('_');
 871                                 string tempName = name;
 872                                 if (pos != string::npos) {  tempName = tempName.substr(pos+1); cout << tempName << endl; }
 873                                 if (namesAccnos.count(tempName) == 0){
 874                                         namesAccnos2.insert(name);
 875                                 }else { //you are in both so erase
 876                                         namesAccnos.erase(name);
 877                                         namesDups.insert(name);
 878                                 }
 879                         }else { //you are in both so erase
 880                                 namesAccnos.erase(name);
 881                                 namesDups.insert(name);
 882                         }
 883
 884                         m->gobble(in);
 885                 }
 886                 in.close();
 887
 888                 out << "Names in both files : " + toString(namesDups.size()) << endl;
 889                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
 890
 891                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
 892                         out << (*it);
 893                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 894                         out << endl;
 895                 }
 896
 897                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
 898                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
 899
 900                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
 901                         out << (*it);
 902                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 903                         out << endl;
 904                 }
 905
 906                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
 907                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
 908
 909                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
 910                         out << (*it);
 911                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 912                         out << endl;
 913                 }
 914
 915                 out.close();
 916
 917                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
 918
 919                 return 0;
 920
 921         }
 922         catch(exception& e) {
 923                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 924                 exit(1);
 925         }
 926 }
 927
 928
 929 //**********************************************************************************************************************
 930