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