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