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