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") { fastafile = ""; 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") { namefile = ""; 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") { taxfile = ""; 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                         if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 254                                 vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 255                                 parser.getNameFile(files);
 256                         }
 257                 }
 258
 259         }
 260         catch(exception& e) {
 261                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 262                 exit(1);
 263         }
 264 }
 265 //**********************************************************************************************************************
 266
 267 int GetSeqsCommand::execute(){
 268         try {
 269
 270                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 271
 272                 //get names you want to keep
 273                 readAccnos();
 274
 275                 if (m->control_pressed) { return 0; }
 276
 277                 //read through the correct file and output lines you want to keep
 278                 if (namefile != "")                     {               readName();                     }
 279                 if (fastafile != "")            {               readFasta();            }
 280                 if (groupfile != "")            {               readGroup();            }
 281                 if (alignfile != "")            {               readAlign();            }
 282                 if (listfile != "")                     {               readList();                     }
 283                 if (taxfile != "")                      {               readTax();                      }
 284                 if (qualfile != "")                     {               readQual();                     }
 285                 if (accnosfile2 != "")          {               compareAccnos();        }
 286
 287                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);  } return 0; }
 288
 289
 290                 if (outputNames.size() != 0) {
 291                         m->mothurOutEndLine();
 292                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 293                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 294                         m->mothurOutEndLine();
 295
 296                         //set fasta file as new current fastafile
 297                         string current = "";
 298                         itTypes = outputTypes.find("fasta");
 299                         if (itTypes != outputTypes.end()) {
 300                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 301                         }
 302
 303                         itTypes = outputTypes.find("name");
 304                         if (itTypes != outputTypes.end()) {
 305                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 306                         }
 307
 308                         itTypes = outputTypes.find("group");
 309                         if (itTypes != outputTypes.end()) {
 310                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 311                         }
 312
 313                         itTypes = outputTypes.find("list");
 314                         if (itTypes != outputTypes.end()) {
 315                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 316                         }
 317
 318                         itTypes = outputTypes.find("taxonomy");
 319                         if (itTypes != outputTypes.end()) {
 320                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 321                         }
 322
 323                         itTypes = outputTypes.find("qfile");
 324                         if (itTypes != outputTypes.end()) {
 325                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 326                         }
 327
 328                 }
 329
 330                 return 0;
 331         }
 332
 333         catch(exception& e) {
 334                 m->errorOut(e, "GetSeqsCommand", "execute");
 335                 exit(1);
 336         }
 337 }
 338
 339 //**********************************************************************************************************************
 340 int GetSeqsCommand::readFasta(){
 341         try {
 342                 string thisOutputDir = outputDir;
 343                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 344                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 345                 ofstream out;
 346                 m->openOutputFile(outputFileName, out);
 347
 348
 349                 ifstream in;
 350                 m->openInputFile(fastafile, in);
 351                 string name;
 352
 353                 bool wroteSomething = false;
 354                 int selectedCount = 0;
 355
 356                 while(!in.eof()){
 357
 358                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 359
 360                         Sequence currSeq(in);
 361                         name = currSeq.getName();
 362
 363                         if (name != "") {
 364                                 //if this name is in the accnos file
 365                                 if (names.count(name) != 0) {
 366                                         wroteSomething = true;
 367
 368                                         currSeq.printSequence(out);
 369                                         selectedCount++;
 370                                 }
 371                         }
 372                         m->gobble(in);
 373                 }
 374                 in.close();
 375                 out.close();
 376
 377
 378                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 379                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 380
 381                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 382
 383                 return 0;
 384
 385         }
 386         catch(exception& e) {
 387                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 388                 exit(1);
 389         }
 390 }
 391 //**********************************************************************************************************************
 392 int GetSeqsCommand::readQual(){
 393         try {
 394                 string thisOutputDir = outputDir;
 395                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 396                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 397                 ofstream out;
 398                 m->openOutputFile(outputFileName, out);
 399
 400
 401                 ifstream in;
 402                 m->openInputFile(qualfile, in);
 403                 string name;
 404
 405                 bool wroteSomething = false;
 406                 int selectedCount = 0;
 407
 408
 409                 while(!in.eof()){
 410                         string saveName = "";
 411                         string name = "";
 412                         string scores = "";
 413
 414                         in >> name;
 415
 416                         if (name.length() != 0) {
 417                                 saveName = name.substr(1);
 418                                 while (!in.eof())       {
 419                                         char c = in.get();
 420                                         if (c == 10 || c == 13){        break;  }
 421                                         else { name += c; }
 422                                 }
 423                                 m->gobble(in);
 424                         }
 425
 426                         while(in){
 427                                 char letter= in.get();
 428                                 if(letter == '>'){      in.putback(letter);     break;  }
 429                                 else{ scores += letter; }
 430                         }
 431
 432                         m->gobble(in);
 433
 434                         if (names.count(saveName) != 0) {
 435                                 wroteSomething = true;
 436
 437                                 out << name << endl << scores;
 438                                 selectedCount++;
 439                         }
 440
 441                         m->gobble(in);
 442                 }
 443                 in.close();
 444                 out.close();
 445
 446
 447                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 448                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 449
 450                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 451
 452
 453                 return 0;
 454
 455         }
 456         catch(exception& e) {
 457                 m->errorOut(e, "GetSeqsCommand", "readQual");
 458                 exit(1);
 459         }
 460 }
 461 //**********************************************************************************************************************
 462 int GetSeqsCommand::readList(){
 463         try {
 464                 string thisOutputDir = outputDir;
 465                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 466                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 467                 ofstream out;
 468                 m->openOutputFile(outputFileName, out);
 469
 470                 ifstream in;
 471                 m->openInputFile(listfile, in);
 472
 473                 bool wroteSomething = false;
 474                 int selectedCount = 0;
 475
 476                 while(!in.eof()){
 477
 478                         selectedCount = 0;
 479
 480                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 481
 482                         //read in list vector
 483                         ListVector list(in);
 484
 485                         //make a new list vector
 486                         ListVector newList;
 487                         newList.setLabel(list.getLabel());
 488
 489                         //for each bin
 490                         for (int i = 0; i < list.getNumBins(); i++) {
 491
 492                                 //parse out names that are in accnos file
 493                                 string binnames = list.get(i);
 494
 495                                 string newNames = "";
 496                                 while (binnames.find_first_of(',') != -1) {
 497                                         string name = binnames.substr(0,binnames.find_first_of(','));
 498                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 499
 500                                         //if that name is in the .accnos file, add it
 501                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++; }
 502                                 }
 503
 504                                 //get last name
 505                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++; }
 506
 507                                 //if there are names in this bin add to new list
 508                                 if (newNames != "") {
 509                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 510                                         newList.push_back(newNames);
 511                                 }
 512                         }
 513
 514                         //print new listvector
 515                         if (newList.getNumBins() != 0) {
 516                                 wroteSomething = true;
 517                                 newList.print(out);
 518                         }
 519
 520                         m->gobble(in);
 521                 }
 522                 in.close();
 523                 out.close();
 524
 525                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 526                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 527
 528                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 529
 530                 return 0;
 531
 532         }
 533         catch(exception& e) {
 534                 m->errorOut(e, "GetSeqsCommand", "readList");
 535                 exit(1);
 536         }
 537 }
 538 //**********************************************************************************************************************
 539 int GetSeqsCommand::readName(){
 540         try {
 541                 string thisOutputDir = outputDir;
 542                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 543                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 544                 ofstream out;
 545                 m->openOutputFile(outputFileName, out);
 546
 547
 548                 ifstream in;
 549                 m->openInputFile(namefile, in);
 550                 string name, firstCol, secondCol;
 551
 552                 bool wroteSomething = false;
 553                 int selectedCount = 0;
 554
 555                 while(!in.eof()){
 556
 557                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 558
 559                         in >> firstCol;
 560                         in >> secondCol;
 561
 562                         string hold = "";
 563                         if (dups) { hold = secondCol; }
 564
 565                         vector<string> parsedNames;
 566                         m->splitAtComma(secondCol, parsedNames);
 567
 568                         vector<string> validSecond;
 569                         for (int i = 0; i < parsedNames.size(); i++) {
 570                                 if (names.count(parsedNames[i]) != 0) {
 571                                         validSecond.push_back(parsedNames[i]);
 572                                 }
 573                         }
 574
 575                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 576                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 577                                 out << firstCol << '\t' << hold << endl;
 578                                 wroteSomething = true;
 579                                 selectedCount += parsedNames.size();
 580                         }else {
 581                                 selectedCount += validSecond.size();
 582
 583                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 584                                 if (names.count(firstCol) != 0) {
 585
 586                                         wroteSomething = true;
 587
 588                                         out << firstCol << '\t';
 589
 590                                         //you know you have at least one valid second since first column is valid
 591                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 592                                         out << validSecond[validSecond.size()-1] << endl;
 593
 594
 595                                 //make first name in set you come to first column and then add the remaining names to second column
 596                                 }else {
 597                                         //you want part of this row
 598                                         if (validSecond.size() != 0) {
 599
 600                                                 wroteSomething = true;
 601
 602                                                 out << validSecond[0] << '\t';
 603
 604                                                 //you know you have at least one valid second since first column is valid
 605                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 606                                                 out << validSecond[validSecond.size()-1] << endl;
 607                                         }
 608                                 }
 609                         }
 610                         m->gobble(in);
 611                 }
 612                 in.close();
 613                 out.close();
 614
 615                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 616                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 617
 618                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 619
 620                 return 0;
 621
 622         }
 623         catch(exception& e) {
 624                 m->errorOut(e, "GetSeqsCommand", "readName");
 625                 exit(1);
 626         }
 627 }
 628
 629 //**********************************************************************************************************************
 630 int GetSeqsCommand::readGroup(){
 631         try {
 632                 string thisOutputDir = outputDir;
 633                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 634                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 635                 ofstream out;
 636                 m->openOutputFile(outputFileName, out);
 637
 638
 639                 ifstream in;
 640                 m->openInputFile(groupfile, in);
 641                 string name, group;
 642
 643                 bool wroteSomething = false;
 644                 int selectedCount = 0;
 645
 646                 while(!in.eof()){
 647
 648                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 649
 650
 651                         in >> name;                             //read from first column
 652                         in >> group;                    //read from second column
 653
 654                         //if this name is in the accnos file
 655                         if (names.count(name) != 0) {
 656                                 wroteSomething = true;
 657
 658                                 out << name << '\t' << group << endl;
 659                                 selectedCount++;
 660                         }
 661
 662                         m->gobble(in);
 663                 }
 664                 in.close();
 665                 out.close();
 666
 667                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 668                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 669
 670                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 671
 672
 673                 return 0;
 674
 675         }
 676         catch(exception& e) {
 677                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 678                 exit(1);
 679         }
 680 }
 681 //**********************************************************************************************************************
 682 int GetSeqsCommand::readTax(){
 683         try {
 684                 string thisOutputDir = outputDir;
 685                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 686                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 687                 ofstream out;
 688                 m->openOutputFile(outputFileName, out);
 689
 690                 ifstream in;
 691                 m->openInputFile(taxfile, in);
 692                 string name, tax;
 693
 694                 bool wroteSomething = false;
 695                 int selectedCount = 0;
 696
 697                 while(!in.eof()){
 698
 699                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 700
 701                         in >> name;                             //read from first column
 702                         in >> tax;                      //read from second column
 703
 704                         //if this name is in the accnos file
 705                         if (names.count(name) != 0) {
 706                                 wroteSomething = true;
 707
 708                                 out << name << '\t' << tax << endl;
 709                                 selectedCount++;
 710                         }
 711
 712                         m->gobble(in);
 713                 }
 714                 in.close();
 715                 out.close();
 716
 717                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 718                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 719
 720                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 721
 722                 return 0;
 723
 724         }
 725         catch(exception& e) {
 726                 m->errorOut(e, "GetSeqsCommand", "readTax");
 727                 exit(1);
 728         }
 729 }
 730 //**********************************************************************************************************************
 731 //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
 732 int GetSeqsCommand::readAlign(){
 733         try {
 734                 string thisOutputDir = outputDir;
 735                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 736                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 737                 ofstream out;
 738                 m->openOutputFile(outputFileName, out);
 739
 740
 741                 ifstream in;
 742                 m->openInputFile(alignfile, in);
 743                 string name, junk;
 744
 745                 bool wroteSomething = false;
 746                 int selectedCount = 0;
 747
 748                 //read column headers
 749                 for (int i = 0; i < 16; i++) {
 750                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 751                         else                    {       break;                  }
 752                 }
 753                 out << endl;
 754
 755                 while(!in.eof()){
 756
 757                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 758
 759
 760                         in >> name;                             //read from first column
 761
 762                         //if this name is in the accnos file
 763                         if (names.count(name) != 0) {
 764                                 wroteSomething = true;
 765                                 selectedCount++;
 766
 767                                 out << name << '\t';
 768
 769                                 //read rest
 770                                 for (int i = 0; i < 15; i++) {
 771                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 772                                         else                    {       break;                  }
 773                                 }
 774                                 out << endl;
 775
 776                         }else {//still read just don't do anything with it
 777                                 //read rest
 778                                 for (int i = 0; i < 15; i++) {
 779                                         if (!in.eof())  {       in >> junk;             }
 780                                         else                    {       break;                  }
 781                                 }
 782                         }
 783
 784                         m->gobble(in);
 785                 }
 786                 in.close();
 787                 out.close();
 788
 789                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 790                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 791
 792                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 793
 794                 return 0;
 795
 796         }
 797         catch(exception& e) {
 798                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 799                 exit(1);
 800         }
 801 }
 802 //**********************************************************************************************************************
 803
 804 int GetSeqsCommand::readAccnos(){
 805         try {
 806
 807                 ifstream in;
 808                 m->openInputFile(accnosfile, in);
 809                 string name;
 810
 811                 while(!in.eof()){
 812                         in >> name;
 813
 814                         names.insert(name);
 815
 816                         m->gobble(in);
 817                 }
 818                 in.close();
 819
 820                 return 0;
 821
 822         }
 823         catch(exception& e) {
 824                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 825                 exit(1);
 826         }
 827 }
 828 //**********************************************************************************************************************
 829
 830 int GetSeqsCommand::compareAccnos(){
 831         try {
 832
 833                 string thisOutputDir = outputDir;
 834                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
 835                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile)) + "accnos.report";
 836                 ofstream out;
 837                 m->openOutputFile(outputFileName, out);
 838
 839                 ifstream in;
 840                 m->openInputFile(accnosfile2, in);
 841                 string name;
 842
 843                 set<string> namesAccnos2;
 844                 set<string> namesDups;
 845                 set<string> namesAccnos = names;
 846
 847                 map<string, int> nameCount;
 848
 849                 if (namefile != "") {
 850                         ifstream inName;
 851                         m->openInputFile(namefile, inName);
 852
 853
 854                         while(!inName.eof()){
 855
 856                                 if (m->control_pressed) { inName.close(); return 0; }
 857
 858                                 string thisname, repnames;
 859
 860                                 inName >> thisname;             m->gobble(inName);              //read from first column
 861                                 inName >> repnames;                     //read from second column
 862
 863                                 int num = m->getNumNames(repnames);
 864                                 nameCount[thisname] = num;
 865
 866                                 m->gobble(inName);
 867                         }
 868                         inName.close();
 869                 }
 870
 871                 while(!in.eof()){
 872                         in >> name;
 873
 874                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
 875                                 int pos = name.find_last_of('_');
 876                                 string tempName = name;
 877                                 if (pos != string::npos) {  tempName = tempName.substr(pos+1); cout << tempName << endl; }
 878                                 if (namesAccnos.count(tempName) == 0){
 879                                         namesAccnos2.insert(name);
 880                                 }else { //you are in both so erase
 881                                         namesAccnos.erase(name);
 882                                         namesDups.insert(name);
 883                                 }
 884                         }else { //you are in both so erase
 885                                 namesAccnos.erase(name);
 886                                 namesDups.insert(name);
 887                         }
 888
 889                         m->gobble(in);
 890                 }
 891                 in.close();
 892
 893                 out << "Names in both files : " + toString(namesDups.size()) << endl;
 894                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
 895
 896                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
 897                         out << (*it);
 898                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 899                         out << endl;
 900                 }
 901
 902                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
 903                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
 904
 905                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
 906                         out << (*it);
 907                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 908                         out << endl;
 909                 }
 910
 911                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
 912                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
 913
 914                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
 915                         out << (*it);
 916                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 917                         out << endl;
 918                 }
 919
 920                 out.close();
 921
 922                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
 923
 924                 return 0;
 925
 926         }
 927         catch(exception& e) {
 928                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 929                 exit(1);
 930         }
 931 }
 932
 933
 934 //**********************************************************************************************************************
 935