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::getValidParameters(){
  16         try {
  17                 string Array[] =  {"fasta","name", "group", "qfile","alignreport", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "GetSeqsCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 GetSeqsCommand::GetSeqsCommand(){
  28         try {
  29                 abort = true;
  30                 //initialize outputTypes
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["taxonomy"] = tempOutNames;
  34                 outputTypes["name"] = tempOutNames;
  35                 outputTypes["group"] = tempOutNames;
  36                 outputTypes["alignreport"] = tempOutNames;
  37                 outputTypes["list"] = tempOutNames;
  38                 outputTypes["qfile"] = tempOutNames;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 vector<string> GetSeqsCommand::getRequiredParameters(){
  47         try {
  48                 string Array[] =  {"accnos"};
  49                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "GetSeqsCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> GetSeqsCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "GetSeqsCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 GetSeqsCommand::GetSeqsCommand(string option)  {
  70         try {
  71                 abort = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"fasta","name", "group", "alignreport", "qfile", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
  79                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["taxonomy"] = tempOutNames;
  96                         outputTypes["name"] = tempOutNames;
  97                         outputTypes["group"] = tempOutNames;
  98                         outputTypes["alignreport"] = tempOutNames;
  99                         outputTypes["list"] = tempOutNames;
 100                         outputTypes["qfile"] = tempOutNames;
 101
 102                         //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("alignreport");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 116                                 }
 117
 118                                 it = parameters.find("fasta");
 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["fasta"] = inputDir + it->second;            }
 124                                 }
 125
 126                                 it = parameters.find("accnos");
 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["accnos"] = inputDir + it->second;           }
 132                                 }
 133
 134                                 it = parameters.find("list");
 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["list"] = inputDir + it->second;             }
 140                                 }
 141
 142                                 it = parameters.find("name");
 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["name"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("group");
 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["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("taxonomy");
 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["taxonomy"] = inputDir + it->second;         }
 164                                 }
 165
 166                                 it = parameters.find("qfile");
 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["qfile"] = inputDir + it->second;            }
 172                                 }
 173                         }
 174
 175
 176                         //check for required parameters
 177                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 178                         if (accnosfile == "not open") { abort = true; }
 179                         else if (accnosfile == "not found") {  accnosfile = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 180
 181                         fastafile = validParameter.validFile(parameters, "fasta", true);
 182                         if (fastafile == "not open") { abort = true; }
 183                         else if (fastafile == "not found") {  fastafile = "";  }
 184
 185                         namefile = validParameter.validFile(parameters, "name", true);
 186                         if (namefile == "not open") { abort = true; }
 187                         else if (namefile == "not found") {  namefile = "";  }
 188
 189                         groupfile = validParameter.validFile(parameters, "group", true);
 190                         if (groupfile == "not open") { abort = true; }
 191                         else if (groupfile == "not found") {  groupfile = "";  }
 192
 193                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 194                         if (alignfile == "not open") { abort = true; }
 195                         else if (alignfile == "not found") {  alignfile = "";  }
 196
 197                         listfile = validParameter.validFile(parameters, "list", true);
 198                         if (listfile == "not open") { abort = true; }
 199                         else if (listfile == "not found") {  listfile = "";  }
 200
 201                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 202                         if (taxfile == "not open") { abort = true; }
 203                         else if (taxfile == "not found") {  taxfile = "";  }
 204
 205                         qualfile = validParameter.validFile(parameters, "qfile", true);
 206                         if (qualfile == "not open") { abort = true; }
 207                         else if (qualfile == "not found") {  qualfile = "";  }
 208
 209                         string usedDups = "true";
 210                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 211                         dups = m->isTrue(temp);
 212
 213                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy, quality or listfile."); m->mothurOutEndLine(); abort = true; }
 214
 215                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 216
 217                 }
 218
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 222                 exit(1);
 223         }
 224 }
 225 //**********************************************************************************************************************
 226
 227 void GetSeqsCommand::help(){
 228         try {
 229                 m->mothurOut("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");
 230                 m->mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
 231                 m->mothurOut("The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos and at least one of the other parameters.\n");
 232                 m->mothurOut("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");
 233                 m->mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 234                 m->mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 235                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 236         }
 237         catch(exception& e) {
 238                 m->errorOut(e, "GetSeqsCommand", "help");
 239                 exit(1);
 240         }
 241 }
 242
 243 //**********************************************************************************************************************
 244
 245 int GetSeqsCommand::execute(){
 246         try {
 247
 248                 if (abort == true) { return 0; }
 249
 250                 //get names you want to keep
 251                 readAccnos();
 252
 253                 if (m->control_pressed) { return 0; }
 254
 255                 //read through the correct file and output lines you want to keep
 256                 if (namefile != "")                     {               readName();             }
 257                 if (fastafile != "")            {               readFasta();    }
 258                 if (groupfile != "")            {               readGroup();    }
 259                 if (alignfile != "")            {               readAlign();    }
 260                 if (listfile != "")                     {               readList();             }
 261                 if (taxfile != "")                      {               readTax();              }
 262                 if (qualfile != "")                     {               readQual();             }
 263
 264                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());  } return 0; }
 265
 266                 m->mothurOut("Selected " + toString(names.size()) + " sequences."); m->mothurOutEndLine();
 267
 268                 if (outputNames.size() != 0) {
 269                         m->mothurOutEndLine();
 270                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 271                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 272                         m->mothurOutEndLine();
 273                 }
 274
 275                 return 0;
 276         }
 277
 278         catch(exception& e) {
 279                 m->errorOut(e, "GetSeqsCommand", "execute");
 280                 exit(1);
 281         }
 282 }
 283
 284 //**********************************************************************************************************************
 285 int GetSeqsCommand::readFasta(){
 286         try {
 287                 string thisOutputDir = outputDir;
 288                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 289                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 290                 ofstream out;
 291                 m->openOutputFile(outputFileName, out);
 292
 293
 294                 ifstream in;
 295                 m->openInputFile(fastafile, in);
 296                 string name;
 297
 298                 bool wroteSomething = false;
 299
 300                 while(!in.eof()){
 301
 302                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 303
 304                         Sequence currSeq(in);
 305                         name = currSeq.getName();
 306
 307                         if (name != "") {
 308                                 //if this name is in the accnos file
 309                                 if (names.count(name) != 0) {
 310                                         wroteSomething = true;
 311
 312                                         currSeq.printSequence(out);
 313                                 }
 314                         }
 315                         m->gobble(in);
 316                 }
 317                 in.close();
 318                 out.close();
 319
 320
 321                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 322                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 323
 324                 return 0;
 325
 326         }
 327         catch(exception& e) {
 328                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 329                 exit(1);
 330         }
 331 }
 332 //**********************************************************************************************************************
 333 int GetSeqsCommand::readQual(){
 334         try {
 335                 string thisOutputDir = outputDir;
 336                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 337                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 338                 ofstream out;
 339                 m->openOutputFile(outputFileName, out);
 340
 341
 342                 ifstream in;
 343                 m->openInputFile(qualfile, in);
 344                 string name;
 345
 346                 bool wroteSomething = false;
 347
 348
 349                 while(!in.eof()){
 350                         string saveName = "";
 351                         string name = "";
 352                         string scores = "";
 353
 354                         in >> name;
 355
 356                         if (name.length() != 0) {
 357                                 saveName = name.substr(1);
 358                                 while (!in.eof())       {
 359                                         char c = in.get();
 360                                         if (c == 10 || c == 13){        break;  }
 361                                         else { name += c; }
 362                                 }
 363                                 m->gobble(in);
 364                         }
 365
 366                         while(in){
 367                                 char letter= in.get();
 368                                 if(letter == '>'){      in.putback(letter);     break;  }
 369                                 else{ scores += letter; }
 370                         }
 371
 372                         m->gobble(in);
 373
 374                         if (names.count(saveName) != 0) {
 375                                 wroteSomething = true;
 376
 377                                 out << name << endl << scores;
 378                         }
 379
 380                         m->gobble(in);
 381                 }
 382                 in.close();
 383                 out.close();
 384
 385
 386                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 387                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 388
 389                 return 0;
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "GetSeqsCommand", "readQual");
 394                 exit(1);
 395         }
 396 }
 397 //**********************************************************************************************************************
 398 int GetSeqsCommand::readList(){
 399         try {
 400                 string thisOutputDir = outputDir;
 401                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 402                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 403                 ofstream out;
 404                 m->openOutputFile(outputFileName, out);
 405
 406                 ifstream in;
 407                 m->openInputFile(listfile, in);
 408
 409                 bool wroteSomething = false;
 410
 411                 while(!in.eof()){
 412
 413                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 414
 415                         //read in list vector
 416                         ListVector list(in);
 417
 418                         //make a new list vector
 419                         ListVector newList;
 420                         newList.setLabel(list.getLabel());
 421
 422                         //for each bin
 423                         for (int i = 0; i < list.getNumBins(); i++) {
 424
 425                                 //parse out names that are in accnos file
 426                                 string binnames = list.get(i);
 427
 428                                 string newNames = "";
 429                                 while (binnames.find_first_of(',') != -1) {
 430                                         string name = binnames.substr(0,binnames.find_first_of(','));
 431                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 432
 433                                         //if that name is in the .accnos file, add it
 434                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 435                                 }
 436
 437                                 //get last name
 438                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 439
 440                                 //if there are names in this bin add to new list
 441                                 if (newNames != "") {
 442                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 443                                         newList.push_back(newNames);
 444                                 }
 445                         }
 446
 447                         //print new listvector
 448                         if (newList.getNumBins() != 0) {
 449                                 wroteSomething = true;
 450                                 newList.print(out);
 451                         }
 452
 453                         m->gobble(in);
 454                 }
 455                 in.close();
 456                 out.close();
 457
 458                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 459                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 460
 461                 return 0;
 462
 463         }
 464         catch(exception& e) {
 465                 m->errorOut(e, "GetSeqsCommand", "readList");
 466                 exit(1);
 467         }
 468 }
 469 //**********************************************************************************************************************
 470 int GetSeqsCommand::readName(){
 471         try {
 472                 string thisOutputDir = outputDir;
 473                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 474                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 475                 ofstream out;
 476                 m->openOutputFile(outputFileName, out);
 477
 478
 479                 ifstream in;
 480                 m->openInputFile(namefile, in);
 481                 string name, firstCol, secondCol;
 482
 483                 bool wroteSomething = false;
 484
 485
 486                 while(!in.eof()){
 487
 488                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 489
 490                         in >> firstCol;
 491                         in >> secondCol;
 492
 493                         string hold = "";
 494                         if (dups) { hold = secondCol; }
 495
 496                         vector<string> parsedNames;
 497                         //parse second column saving each name
 498                         while (secondCol.find_first_of(',') != -1) {
 499                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 500                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 501                                 parsedNames.push_back(name);
 502                         }
 503
 504                         //get name after last ,
 505                         parsedNames.push_back(secondCol);
 506
 507                         vector<string> validSecond;
 508                         for (int i = 0; i < parsedNames.size(); i++) {
 509                                 if (names.count(parsedNames[i]) != 0) {
 510                                         validSecond.push_back(parsedNames[i]);
 511                                 }
 512                         }
 513
 514                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 515                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 516                                 out << firstCol << '\t' << hold << endl;
 517                                 wroteSomething = true;
 518                         }else {
 519                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 520                                 if (names.count(firstCol) != 0) {
 521
 522                                         wroteSomething = true;
 523
 524                                         out << firstCol << '\t';
 525
 526                                         //you know you have at least one valid second since first column is valid
 527                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 528                                         out << validSecond[validSecond.size()-1] << endl;
 529
 530
 531                                 //make first name in set you come to first column and then add the remaining names to second column
 532                                 }else {
 533                                         //you want part of this row
 534                                         if (validSecond.size() != 0) {
 535
 536                                                 wroteSomething = true;
 537
 538                                                 out << validSecond[0] << '\t';
 539
 540                                                 //you know you have at least one valid second since first column is valid
 541                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 542                                                 out << validSecond[validSecond.size()-1] << endl;
 543                                         }
 544                                 }
 545                         }
 546                         m->gobble(in);
 547                 }
 548                 in.close();
 549                 out.close();
 550
 551                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 552                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 553
 554                 return 0;
 555
 556         }
 557         catch(exception& e) {
 558                 m->errorOut(e, "GetSeqsCommand", "readName");
 559                 exit(1);
 560         }
 561 }
 562
 563 //**********************************************************************************************************************
 564 int GetSeqsCommand::readGroup(){
 565         try {
 566                 string thisOutputDir = outputDir;
 567                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 568                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 569                 ofstream out;
 570                 m->openOutputFile(outputFileName, out);
 571
 572
 573                 ifstream in;
 574                 m->openInputFile(groupfile, in);
 575                 string name, group;
 576
 577                 bool wroteSomething = false;
 578
 579                 while(!in.eof()){
 580
 581                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 582
 583
 584                         in >> name;                             //read from first column
 585                         in >> group;                    //read from second column
 586
 587                         //if this name is in the accnos file
 588                         if (names.count(name) != 0) {
 589                                 wroteSomething = true;
 590
 591                                 out << name << '\t' << group << endl;
 592                         }
 593
 594                         m->gobble(in);
 595                 }
 596                 in.close();
 597                 out.close();
 598
 599                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 600                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 601
 602                 return 0;
 603
 604         }
 605         catch(exception& e) {
 606                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 607                 exit(1);
 608         }
 609 }
 610 //**********************************************************************************************************************
 611 int GetSeqsCommand::readTax(){
 612         try {
 613                 string thisOutputDir = outputDir;
 614                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 615                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 616                 ofstream out;
 617                 m->openOutputFile(outputFileName, out);
 618
 619                 ifstream in;
 620                 m->openInputFile(taxfile, in);
 621                 string name, tax;
 622
 623                 bool wroteSomething = false;
 624
 625                 while(!in.eof()){
 626
 627                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 628
 629                         in >> name;                             //read from first column
 630                         in >> tax;                      //read from second column
 631
 632                         //if this name is in the accnos file
 633                         if (names.count(name) != 0) {
 634                                 wroteSomething = true;
 635
 636                                 out << name << '\t' << tax << endl;
 637                         }
 638
 639                         m->gobble(in);
 640                 }
 641                 in.close();
 642                 out.close();
 643
 644                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 645                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 646
 647                 return 0;
 648
 649         }
 650         catch(exception& e) {
 651                 m->errorOut(e, "GetSeqsCommand", "readTax");
 652                 exit(1);
 653         }
 654 }
 655 //**********************************************************************************************************************
 656 //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
 657 int GetSeqsCommand::readAlign(){
 658         try {
 659                 string thisOutputDir = outputDir;
 660                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 661                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 662                 ofstream out;
 663                 m->openOutputFile(outputFileName, out);
 664
 665
 666                 ifstream in;
 667                 m->openInputFile(alignfile, in);
 668                 string name, junk;
 669
 670                 bool wroteSomething = false;
 671
 672                 //read column headers
 673                 for (int i = 0; i < 16; i++) {
 674                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 675                         else                    {       break;                  }
 676                 }
 677                 out << endl;
 678
 679                 while(!in.eof()){
 680
 681                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 682
 683
 684                         in >> name;                             //read from first column
 685
 686                         //if this name is in the accnos file
 687                         if (names.count(name) != 0) {
 688                                 wroteSomething = true;
 689
 690                                 out << name << '\t';
 691
 692                                 //read rest
 693                                 for (int i = 0; i < 15; i++) {
 694                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 695                                         else                    {       break;                  }
 696                                 }
 697                                 out << endl;
 698
 699                         }else {//still read just don't do anything with it
 700                                 //read rest
 701                                 for (int i = 0; i < 15; i++) {
 702                                         if (!in.eof())  {       in >> junk;             }
 703                                         else                    {       break;                  }
 704                                 }
 705                         }
 706
 707                         m->gobble(in);
 708                 }
 709                 in.close();
 710                 out.close();
 711
 712                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 713                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 714
 715                 return 0;
 716
 717         }
 718         catch(exception& e) {
 719                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 720                 exit(1);
 721         }
 722 }
 723 //**********************************************************************************************************************
 724
 725 int GetSeqsCommand::readAccnos(){
 726         try {
 727
 728                 ifstream in;
 729                 m->openInputFile(accnosfile, in);
 730                 string name;
 731
 732                 while(!in.eof()){
 733                         in >> name;
 734
 735                         names.insert(name);
 736
 737                         m->gobble(in);
 738                 }
 739                 in.close();
 740
 741                 return 0;
 742
 743         }
 744         catch(exception& e) {
 745                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 746                 exit(1);
 747         }
 748 }
 749
 750 //**********************************************************************************************************************
 751