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