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 = "false"; 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
 287                 return 0;
 288         }
 289
 290         catch(exception& e) {
 291                 m->errorOut(e, "GetSeqsCommand", "execute");
 292                 exit(1);
 293         }
 294 }
 295
 296 //**********************************************************************************************************************
 297 int GetSeqsCommand::readFasta(){
 298         try {
 299                 string thisOutputDir = outputDir;
 300                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 301                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 302                 ofstream out;
 303                 m->openOutputFile(outputFileName, out);
 304
 305
 306                 ifstream in;
 307                 m->openInputFile(fastafile, in);
 308                 string name;
 309
 310                 bool wroteSomething = false;
 311
 312                 while(!in.eof()){
 313
 314                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 315
 316                         Sequence currSeq(in);
 317                         name = currSeq.getName();
 318
 319                         if (name != "") {
 320                                 //if this name is in the accnos file
 321                                 if (names.count(name) != 0) {
 322                                         wroteSomething = true;
 323
 324                                         currSeq.printSequence(out);
 325                                 }
 326                         }
 327                         m->gobble(in);
 328                 }
 329                 in.close();
 330                 out.close();
 331
 332
 333                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 334                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 335
 336                 return 0;
 337
 338         }
 339         catch(exception& e) {
 340                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 341                 exit(1);
 342         }
 343 }
 344 //**********************************************************************************************************************
 345 int GetSeqsCommand::readQual(){
 346         try {
 347                 string thisOutputDir = outputDir;
 348                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 349                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 350                 ofstream out;
 351                 m->openOutputFile(outputFileName, out);
 352
 353
 354                 ifstream in;
 355                 m->openInputFile(qualfile, in);
 356                 string name;
 357
 358                 bool wroteSomething = false;
 359
 360
 361                 while(!in.eof()){
 362                         string saveName = "";
 363                         string name = "";
 364                         string scores = "";
 365
 366                         in >> name;
 367
 368                         if (name.length() != 0) {
 369                                 saveName = name.substr(1);
 370                                 while (!in.eof())       {
 371                                         char c = in.get();
 372                                         if (c == 10 || c == 13){        break;  }
 373                                         else { name += c; }
 374                                 }
 375                                 m->gobble(in);
 376                         }
 377
 378                         while(in){
 379                                 char letter= in.get();
 380                                 if(letter == '>'){      in.putback(letter);     break;  }
 381                                 else{ scores += letter; }
 382                         }
 383
 384                         m->gobble(in);
 385
 386                         if (names.count(saveName) != 0) {
 387                                 wroteSomething = true;
 388
 389                                 out << name << endl << scores;
 390                         }
 391
 392                         m->gobble(in);
 393                 }
 394                 in.close();
 395                 out.close();
 396
 397
 398                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 399                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 400
 401                 return 0;
 402
 403         }
 404         catch(exception& e) {
 405                 m->errorOut(e, "GetSeqsCommand", "readQual");
 406                 exit(1);
 407         }
 408 }
 409 //**********************************************************************************************************************
 410 int GetSeqsCommand::readList(){
 411         try {
 412                 string thisOutputDir = outputDir;
 413                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 414                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 415                 ofstream out;
 416                 m->openOutputFile(outputFileName, out);
 417
 418                 ifstream in;
 419                 m->openInputFile(listfile, in);
 420
 421                 bool wroteSomething = false;
 422
 423                 while(!in.eof()){
 424
 425                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 426
 427                         //read in list vector
 428                         ListVector list(in);
 429
 430                         //make a new list vector
 431                         ListVector newList;
 432                         newList.setLabel(list.getLabel());
 433
 434                         //for each bin
 435                         for (int i = 0; i < list.getNumBins(); i++) {
 436
 437                                 //parse out names that are in accnos file
 438                                 string binnames = list.get(i);
 439
 440                                 string newNames = "";
 441                                 while (binnames.find_first_of(',') != -1) {
 442                                         string name = binnames.substr(0,binnames.find_first_of(','));
 443                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 444
 445                                         //if that name is in the .accnos file, add it
 446                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 447                                 }
 448
 449                                 //get last name
 450                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 451
 452                                 //if there are names in this bin add to new list
 453                                 if (newNames != "") {
 454                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 455                                         newList.push_back(newNames);
 456                                 }
 457                         }
 458
 459                         //print new listvector
 460                         if (newList.getNumBins() != 0) {
 461                                 wroteSomething = true;
 462                                 newList.print(out);
 463                         }
 464
 465                         m->gobble(in);
 466                 }
 467                 in.close();
 468                 out.close();
 469
 470                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 471                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 472
 473                 return 0;
 474
 475         }
 476         catch(exception& e) {
 477                 m->errorOut(e, "GetSeqsCommand", "readList");
 478                 exit(1);
 479         }
 480 }
 481 //**********************************************************************************************************************
 482 int GetSeqsCommand::readName(){
 483         try {
 484                 string thisOutputDir = outputDir;
 485                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 486                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 487                 ofstream out;
 488                 m->openOutputFile(outputFileName, out);
 489
 490
 491                 ifstream in;
 492                 m->openInputFile(namefile, in);
 493                 string name, firstCol, secondCol;
 494
 495                 bool wroteSomething = false;
 496
 497
 498                 while(!in.eof()){
 499
 500                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 501
 502                         in >> firstCol;
 503                         in >> secondCol;
 504
 505                         string hold = "";
 506                         if (dups) { hold = secondCol; }
 507
 508                         vector<string> parsedNames;
 509                         //parse second column saving each name
 510                         while (secondCol.find_first_of(',') != -1) {
 511                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 512                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 513                                 parsedNames.push_back(name);
 514                         }
 515
 516                         //get name after last ,
 517                         parsedNames.push_back(secondCol);
 518
 519                         vector<string> validSecond;
 520                         for (int i = 0; i < parsedNames.size(); i++) {
 521                                 if (names.count(parsedNames[i]) != 0) {
 522                                         validSecond.push_back(parsedNames[i]);
 523                                 }
 524                         }
 525
 526                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 527                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 528                                 out << firstCol << '\t' << hold << endl;
 529                                 wroteSomething = true;
 530                         }else {
 531                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 532                                 if (names.count(firstCol) != 0) {
 533
 534                                         wroteSomething = true;
 535
 536                                         out << firstCol << '\t';
 537
 538                                         //you know you have at least one valid second since first column is valid
 539                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 540                                         out << validSecond[validSecond.size()-1] << endl;
 541
 542
 543                                 //make first name in set you come to first column and then add the remaining names to second column
 544                                 }else {
 545                                         //you want part of this row
 546                                         if (validSecond.size() != 0) {
 547
 548                                                 wroteSomething = true;
 549
 550                                                 out << validSecond[0] << '\t';
 551
 552                                                 //you know you have at least one valid second since first column is valid
 553                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 554                                                 out << validSecond[validSecond.size()-1] << endl;
 555                                         }
 556                                 }
 557                         }
 558                         m->gobble(in);
 559                 }
 560                 in.close();
 561                 out.close();
 562
 563                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 564                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 565
 566                 return 0;
 567
 568         }
 569         catch(exception& e) {
 570                 m->errorOut(e, "GetSeqsCommand", "readName");
 571                 exit(1);
 572         }
 573 }
 574
 575 //**********************************************************************************************************************
 576 int GetSeqsCommand::readGroup(){
 577         try {
 578                 string thisOutputDir = outputDir;
 579                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 580                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 581                 ofstream out;
 582                 m->openOutputFile(outputFileName, out);
 583
 584
 585                 ifstream in;
 586                 m->openInputFile(groupfile, in);
 587                 string name, group;
 588
 589                 bool wroteSomething = false;
 590
 591                 while(!in.eof()){
 592
 593                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 594
 595
 596                         in >> name;                             //read from first column
 597                         in >> group;                    //read from second column
 598
 599                         //if this name is in the accnos file
 600                         if (names.count(name) != 0) {
 601                                 wroteSomething = true;
 602
 603                                 out << name << '\t' << group << endl;
 604                         }
 605
 606                         m->gobble(in);
 607                 }
 608                 in.close();
 609                 out.close();
 610
 611                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 612                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 613
 614                 return 0;
 615
 616         }
 617         catch(exception& e) {
 618                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 619                 exit(1);
 620         }
 621 }
 622 //**********************************************************************************************************************
 623 int GetSeqsCommand::readTax(){
 624         try {
 625                 string thisOutputDir = outputDir;
 626                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 627                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 628                 ofstream out;
 629                 m->openOutputFile(outputFileName, out);
 630
 631                 ifstream in;
 632                 m->openInputFile(taxfile, in);
 633                 string name, tax;
 634
 635                 bool wroteSomething = false;
 636
 637                 while(!in.eof()){
 638
 639                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 640
 641                         in >> name;                             //read from first column
 642                         in >> tax;                      //read from second column
 643
 644                         //if this name is in the accnos file
 645                         if (names.count(name) != 0) {
 646                                 wroteSomething = true;
 647
 648                                 out << name << '\t' << tax << endl;
 649                         }
 650
 651                         m->gobble(in);
 652                 }
 653                 in.close();
 654                 out.close();
 655
 656                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 657                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 658
 659                 return 0;
 660
 661         }
 662         catch(exception& e) {
 663                 m->errorOut(e, "GetSeqsCommand", "readTax");
 664                 exit(1);
 665         }
 666 }
 667 //**********************************************************************************************************************
 668 //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
 669 int GetSeqsCommand::readAlign(){
 670         try {
 671                 string thisOutputDir = outputDir;
 672                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 673                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 674                 ofstream out;
 675                 m->openOutputFile(outputFileName, out);
 676
 677
 678                 ifstream in;
 679                 m->openInputFile(alignfile, in);
 680                 string name, junk;
 681
 682                 bool wroteSomething = false;
 683
 684                 //read column headers
 685                 for (int i = 0; i < 16; i++) {
 686                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 687                         else                    {       break;                  }
 688                 }
 689                 out << endl;
 690
 691                 while(!in.eof()){
 692
 693                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 694
 695
 696                         in >> name;                             //read from first column
 697
 698                         //if this name is in the accnos file
 699                         if (names.count(name) != 0) {
 700                                 wroteSomething = true;
 701
 702                                 out << name << '\t';
 703
 704                                 //read rest
 705                                 for (int i = 0; i < 15; i++) {
 706                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 707                                         else                    {       break;                  }
 708                                 }
 709                                 out << endl;
 710
 711                         }else {//still read just don't do anything with it
 712                                 //read rest
 713                                 for (int i = 0; i < 15; i++) {
 714                                         if (!in.eof())  {       in >> junk;             }
 715                                         else                    {       break;                  }
 716                                 }
 717                         }
 718
 719                         m->gobble(in);
 720                 }
 721                 in.close();
 722                 out.close();
 723
 724                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 725                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 726
 727                 return 0;
 728
 729         }
 730         catch(exception& e) {
 731                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 732                 exit(1);
 733         }
 734 }
 735 //**********************************************************************************************************************
 736
 737 int GetSeqsCommand::readAccnos(){
 738         try {
 739
 740                 ifstream in;
 741                 m->openInputFile(accnosfile, in);
 742                 string name;
 743
 744                 while(!in.eof()){
 745                         in >> name;
 746
 747                         names.insert(name);
 748
 749                         m->gobble(in);
 750                 }
 751                 in.close();
 752
 753                 return 0;
 754
 755         }
 756         catch(exception& e) {
 757                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 758                 exit(1);
 759         }
 760 }
 761 //**********************************************************************************************************************
 762
 763 int GetSeqsCommand::compareAccnos(){
 764         try {
 765
 766                 string thisOutputDir = outputDir;
 767                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
 768                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile)) + "accnos.report";
 769                 ofstream out;
 770                 m->openOutputFile(outputFileName, out);
 771
 772                 ifstream in;
 773                 m->openInputFile(accnosfile2, in);
 774                 string name;
 775
 776                 set<string> namesAccnos2;
 777                 set<string> namesDups;
 778                 set<string> namesAccnos = names;
 779
 780                 map<string, int> nameCount;
 781
 782                 if (namefile != "") {
 783                         ifstream inName;
 784                         m->openInputFile(namefile, inName);
 785
 786
 787                         while(!inName.eof()){
 788
 789                                 if (m->control_pressed) { inName.close(); return 0; }
 790
 791                                 string thisname, repnames;
 792
 793                                 inName >> thisname;             m->gobble(inName);              //read from first column
 794                                 inName >> repnames;                     //read from second column
 795
 796                                 int num = m->getNumNames(repnames);
 797                                 nameCount[thisname] = num;
 798
 799                                 m->gobble(inName);
 800                         }
 801                         inName.close();
 802                 }
 803
 804                 while(!in.eof()){
 805                         in >> name;
 806
 807                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
 808                                 namesAccnos2.insert(name);
 809                         }else { //you are in both so erase
 810                                 namesAccnos.erase(name);
 811                                 namesDups.insert(name);
 812                         }
 813
 814                         m->gobble(in);
 815                 }
 816                 in.close();
 817
 818                 out << "Names in both files : " + toString(namesDups.size()) << endl;
 819                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
 820
 821                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
 822                         out << (*it);
 823                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 824                         out << endl;
 825                 }
 826
 827                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
 828                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
 829
 830                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
 831                         out << (*it);
 832                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 833                         out << endl;
 834                 }
 835
 836                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
 837                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
 838
 839                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
 840                         out << (*it);
 841                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
 842                         out << endl;
 843                 }
 844
 845                 out.close();
 846
 847                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
 848
 849                 return 0;
 850
 851         }
 852         catch(exception& e) {
 853                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 854                 exit(1);
 855         }
 856 }
 857
 858
 859 //**********************************************************************************************************************
 860