getseqscommand.cpp

   1 /*
   2  *  getseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "getseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> GetSeqsCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pgroup);
  20                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  21                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  22                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(palignreport);
  23                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pqfile);
  24                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos);
  25                 CommandParameter pdups("dups", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pdups);
  26                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  27                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  28                 CommandParameter paccnos2("accnos2", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(paccnos2);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "GetSeqsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string GetSeqsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, list, taxonomy, quality or alignreport file.\n";
  44                 helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
  45                 helpString += "The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos unless you have a valid current accnos file, and at least one of the other parameters.\n";
  46                 helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n";
  47                 helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
  48                 helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
  49                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  50                 return helpString;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "GetSeqsCommand", "getHelpString");
  54                 exit(1);
  55         }
  56 }
  57
  58 //**********************************************************************************************************************
  59 GetSeqsCommand::GetSeqsCommand(){
  60         try {
  61                 abort = true; calledHelp = true;
  62                 setParameters();
  63                 vector<string> tempOutNames;
  64                 outputTypes["fasta"] = tempOutNames;
  65                 outputTypes["taxonomy"] = tempOutNames;
  66                 outputTypes["name"] = tempOutNames;
  67                 outputTypes["group"] = tempOutNames;
  68                 outputTypes["alignreport"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["qfile"] = tempOutNames;
  71                 outputTypes["accnosreport"] = tempOutNames;
  72         }
  73         catch(exception& e) {
  74                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  75                 exit(1);
  76         }
  77 }
  78 //**********************************************************************************************************************
  79 string GetSeqsCommand::getOutputFileNameTag(string type, string inputName=""){
  80         try {
  81         string outputFileName = "";
  82                 map<string, vector<string> >::iterator it;
  83
  84         //is this a type this command creates
  85         it = outputTypes.find(type);
  86         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  87         else {
  88             if (type == "fasta")            {   outputFileName =  "pick" + m->getExtension(inputName);   }
  89             else if (type == "taxonomy")    {   outputFileName =  "pick" + m->getExtension(inputName);   }
  90             else if (type == "name")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  91             else if (type == "group")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  92             else if (type == "list")        {   outputFileName =  "pick" + m->getExtension(inputName);   }
  93             else if (type == "qfile")       {   outputFileName =  "pick" + m->getExtension(inputName);   }
  94             else if (type == "accnosreport"){   outputFileName =  "accnos.report";                       }
  95             else if (type == "alignreport") {   outputFileName =  "pick.align.report";                   }
  96             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  97         }
  98         return outputFileName;
  99         }
 100         catch(exception& e) {
 101                 m->errorOut(e, "GetSeqsCommand", "getOutputFileNameTag");
 102                 exit(1);
 103         }
 104 }
 105 //**********************************************************************************************************************
 106 GetSeqsCommand::GetSeqsCommand(string option)  {
 107         try {
 108                 abort = false; calledHelp = false;
 109
 110                 //allow user to run help
 111                 if(option == "help") { help(); abort = true; calledHelp = true; }
 112                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
 113
 114                 else {
 115                         vector<string> myArray = setParameters();
 116
 117                         OptionParser parser(option);
 118                         map<string,string> parameters = parser.getParameters();
 119
 120                         ValidParameters validParameter;
 121                         map<string,string>::iterator it;
 122
 123                         //check to make sure all parameters are valid for command
 124                         for (it = parameters.begin(); it != parameters.end(); it++) {
 125                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 126                         }
 127
 128                         //initialize outputTypes
 129                         vector<string> tempOutNames;
 130                         outputTypes["fasta"] = tempOutNames;
 131                         outputTypes["taxonomy"] = tempOutNames;
 132                         outputTypes["name"] = tempOutNames;
 133                         outputTypes["group"] = tempOutNames;
 134                         outputTypes["alignreport"] = tempOutNames;
 135                         outputTypes["list"] = tempOutNames;
 136                         outputTypes["qfile"] = tempOutNames;
 137                         outputTypes["accnosreport"] = tempOutNames;
 138
 139                         //if the user changes the output directory command factory will send this info to us in the output parameter
 140                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 141
 142                         //if the user changes the input directory command factory will send this info to us in the output parameter
 143                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 144                         if (inputDir == "not found"){   inputDir = "";          }
 145                         else {
 146                                 string path;
 147                                 it = parameters.find("alignreport");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 153                                 }
 154
 155                                 it = parameters.find("fasta");
 156                                 //user has given a template file
 157                                 if(it != parameters.end()){
 158                                         path = m->hasPath(it->second);
 159                                         //if the user has not given a path then, add inputdir. else leave path alone.
 160                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 161                                 }
 162
 163                                 it = parameters.find("accnos");
 164                                 //user has given a template file
 165                                 if(it != parameters.end()){
 166                                         path = m->hasPath(it->second);
 167                                         //if the user has not given a path then, add inputdir. else leave path alone.
 168                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 169                                 }
 170
 171                                 it = parameters.find("accnos2");
 172                                 //user has given a template file
 173                                 if(it != parameters.end()){
 174                                         path = m->hasPath(it->second);
 175                                         //if the user has not given a path then, add inputdir. else leave path alone.
 176                                         if (path == "") {       parameters["accnos2"] = inputDir + it->second;          }
 177                                 }
 178
 179                                 it = parameters.find("list");
 180                                 //user has given a template file
 181                                 if(it != parameters.end()){
 182                                         path = m->hasPath(it->second);
 183                                         //if the user has not given a path then, add inputdir. else leave path alone.
 184                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 185                                 }
 186
 187                                 it = parameters.find("name");
 188                                 //user has given a template file
 189                                 if(it != parameters.end()){
 190                                         path = m->hasPath(it->second);
 191                                         //if the user has not given a path then, add inputdir. else leave path alone.
 192                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 193                                 }
 194
 195                                 it = parameters.find("group");
 196                                 //user has given a template file
 197                                 if(it != parameters.end()){
 198                                         path = m->hasPath(it->second);
 199                                         //if the user has not given a path then, add inputdir. else leave path alone.
 200                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 201                                 }
 202
 203                                 it = parameters.find("taxonomy");
 204                                 //user has given a template file
 205                                 if(it != parameters.end()){
 206                                         path = m->hasPath(it->second);
 207                                         //if the user has not given a path then, add inputdir. else leave path alone.
 208                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 209                                 }
 210
 211                                 it = parameters.find("qfile");
 212                                 //user has given a template file
 213                                 if(it != parameters.end()){
 214                                         path = m->hasPath(it->second);
 215                                         //if the user has not given a path then, add inputdir. else leave path alone.
 216                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 217                                 }
 218                         }
 219
 220
 221                         //check for required parameters
 222                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 223                         if (accnosfile == "not open") { abort = true; }
 224                         else if (accnosfile == "not found") {
 225                                 accnosfile = m->getAccnosFile();
 226                                 if (accnosfile != "") {  m->mothurOut("Using " + accnosfile + " as input file for the accnos parameter."); m->mothurOutEndLine(); }
 227                                 else {
 228                                         m->mothurOut("You have no valid accnos file and accnos is required."); m->mothurOutEndLine();
 229                                         abort = true;
 230                                 }
 231                         }else { m->setAccnosFile(accnosfile); }
 232
 233                         if (accnosfile2 == "not found") { accnosfile2 = ""; }
 234
 235                         fastafile = validParameter.validFile(parameters, "fasta", true);
 236                         if (fastafile == "not open") { fastafile = ""; abort = true; }
 237                         else if (fastafile == "not found") {  fastafile = "";  }
 238                         else { m->setFastaFile(fastafile); }
 239
 240                         namefile = validParameter.validFile(parameters, "name", true);
 241                         if (namefile == "not open") { namefile = ""; abort = true; }
 242                         else if (namefile == "not found") {  namefile = "";  }
 243                         else { m->setNameFile(namefile); }
 244
 245                         groupfile = validParameter.validFile(parameters, "group", true);
 246                         if (groupfile == "not open") { abort = true; }
 247                         else if (groupfile == "not found") {  groupfile = "";  }
 248                         else { m->setGroupFile(groupfile); }
 249
 250                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 251                         if (alignfile == "not open") { abort = true; }
 252                         else if (alignfile == "not found") {  alignfile = "";  }
 253
 254                         listfile = validParameter.validFile(parameters, "list", true);
 255                         if (listfile == "not open") { abort = true; }
 256                         else if (listfile == "not found") {  listfile = "";  }
 257                         else { m->setListFile(listfile); }
 258
 259                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 260                         if (taxfile == "not open") { taxfile = ""; abort = true; }
 261                         else if (taxfile == "not found") {  taxfile = "";  }
 262                         else { m->setTaxonomyFile(taxfile); }
 263
 264                         qualfile = validParameter.validFile(parameters, "qfile", true);
 265                         if (qualfile == "not open") { abort = true; }
 266                         else if (qualfile == "not found") {  qualfile = "";  }
 267                         else { m->setQualFile(qualfile); }
 268
 269                         accnosfile2 = validParameter.validFile(parameters, "accnos2", true);
 270                         if (accnosfile2 == "not open") { abort = true; }
 271                         else if (accnosfile2 == "not found") {  accnosfile2 = "";  }
 272
 273
 274                         string usedDups = "true";
 275                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "true"; usedDups = ""; }
 276                         dups = m->isTrue(temp);
 277
 278                         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; }
 279
 280                         if ((namefile == "") && ((fastafile != "") || (taxfile != ""))){
 281                                 vector<string> files; files.push_back(fastafile); files.push_back(taxfile);
 282                                 parser.getNameFile(files);
 283                         }
 284                 }
 285
 286         }
 287         catch(exception& e) {
 288                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 289                 exit(1);
 290         }
 291 }
 292 //**********************************************************************************************************************
 293
 294 int GetSeqsCommand::execute(){
 295         try {
 296
 297                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 298
 299                 //get names you want to keep
 300                 names = m->readAccnos(accnosfile);
 301
 302                 if (m->control_pressed) { return 0; }
 303
 304                 //read through the correct file and output lines you want to keep
 305                 if (namefile != "")                     {               readName();                     }
 306                 if (fastafile != "")            {               readFasta();            }
 307                 if (groupfile != "")            {               readGroup();            }
 308                 if (alignfile != "")            {               readAlign();            }
 309                 if (listfile != "")                     {               readList();                     }
 310                 if (taxfile != "")                      {               readTax();                      }
 311                 if (qualfile != "")                     {               readQual();                     }
 312                 if (accnosfile2 != "")          {               compareAccnos();        }
 313
 314         if (m->debug) { runSanityCheck(); }
 315
 316                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   m->mothurRemove(outputNames[i]);  } return 0; }
 317
 318
 319                 if (outputNames.size() != 0) {
 320                         m->mothurOutEndLine();
 321                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 322                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 323                         m->mothurOutEndLine();
 324
 325                         //set fasta file as new current fastafile
 326                         string current = "";
 327                         itTypes = outputTypes.find("fasta");
 328                         if (itTypes != outputTypes.end()) {
 329                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 330                         }
 331
 332                         itTypes = outputTypes.find("name");
 333                         if (itTypes != outputTypes.end()) {
 334                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 335                         }
 336
 337                         itTypes = outputTypes.find("group");
 338                         if (itTypes != outputTypes.end()) {
 339                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 340                         }
 341
 342                         itTypes = outputTypes.find("list");
 343                         if (itTypes != outputTypes.end()) {
 344                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 345                         }
 346
 347                         itTypes = outputTypes.find("taxonomy");
 348                         if (itTypes != outputTypes.end()) {
 349                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 350                         }
 351
 352                         itTypes = outputTypes.find("qfile");
 353                         if (itTypes != outputTypes.end()) {
 354                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 355                         }
 356
 357                 }
 358
 359                 return 0;
 360         }
 361
 362         catch(exception& e) {
 363                 m->errorOut(e, "GetSeqsCommand", "execute");
 364                 exit(1);
 365         }
 366 }
 367
 368 //**********************************************************************************************************************
 369 int GetSeqsCommand::readFasta(){
 370         try {
 371                 string thisOutputDir = outputDir;
 372                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 373                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
 374                 ofstream out;
 375                 m->openOutputFile(outputFileName, out);
 376
 377
 378                 ifstream in;
 379                 m->openInputFile(fastafile, in);
 380                 string name;
 381
 382                 bool wroteSomething = false;
 383                 int selectedCount = 0;
 384
 385         if (m->debug) { set<string> temp; sanity["fasta"] = temp; }
 386
 387                 while(!in.eof()){
 388
 389                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 390
 391                         Sequence currSeq(in);
 392                         name = currSeq.getName();
 393
 394                         if (name != "") {
 395                                 //if this name is in the accnos file
 396                                 if (names.count(name) != 0) {
 397                                         wroteSomething = true;
 398
 399                                         currSeq.printSequence(out);
 400                                         selectedCount++;
 401
 402                     if (m->debug) { sanity["fasta"].insert(name); }
 403                                 }
 404                         }
 405                         m->gobble(in);
 406                 }
 407                 in.close();
 408                 out.close();
 409
 410
 411                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 412                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 413
 414                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 415
 416                 return 0;
 417
 418         }
 419         catch(exception& e) {
 420                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 421                 exit(1);
 422         }
 423 }
 424 //**********************************************************************************************************************
 425 int GetSeqsCommand::readQual(){
 426         try {
 427                 string thisOutputDir = outputDir;
 428                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 429                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + getOutputFileNameTag("qfile", qualfile);
 430                 ofstream out;
 431                 m->openOutputFile(outputFileName, out);
 432
 433
 434                 ifstream in;
 435                 m->openInputFile(qualfile, in);
 436                 string name;
 437
 438                 bool wroteSomething = false;
 439                 int selectedCount = 0;
 440
 441         if (m->debug) { set<string> temp; sanity["qual"] = temp; }
 442
 443                 while(!in.eof()){
 444                         string saveName = "";
 445                         string name = "";
 446                         string scores = "";
 447
 448                         in >> name;
 449
 450                         if (name.length() != 0) {
 451                                 saveName = name.substr(1);
 452                                 while (!in.eof())       {
 453                                         char c = in.get();
 454                                         if (c == 10 || c == 13){        break;  }
 455                                         else { name += c; }
 456                                 }
 457                                 m->gobble(in);
 458                         }
 459
 460                         while(in){
 461                                 char letter= in.get();
 462                                 if(letter == '>'){      in.putback(letter);     break;  }
 463                                 else{ scores += letter; }
 464                         }
 465
 466                         m->gobble(in);
 467
 468                         if (names.count(saveName) != 0) {
 469                                 wroteSomething = true;
 470
 471                                 out << name << endl << scores;
 472                                 selectedCount++;
 473                 if (m->debug) { sanity["qual"].insert(name); }
 474                         }
 475
 476                         m->gobble(in);
 477                 }
 478                 in.close();
 479                 out.close();
 480
 481
 482                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 483                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 484
 485                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your quality file."); m->mothurOutEndLine();
 486
 487
 488                 return 0;
 489
 490         }
 491         catch(exception& e) {
 492                 m->errorOut(e, "GetSeqsCommand", "readQual");
 493                 exit(1);
 494         }
 495 }
 496 //**********************************************************************************************************************
 497 int GetSeqsCommand::readList(){
 498         try {
 499                 string thisOutputDir = outputDir;
 500                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 501                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + getOutputFileNameTag("list", listfile);
 502                 ofstream out;
 503                 m->openOutputFile(outputFileName, out);
 504
 505                 ifstream in;
 506                 m->openInputFile(listfile, in);
 507
 508                 bool wroteSomething = false;
 509                 int selectedCount = 0;
 510
 511         if (m->debug) { set<string> temp; sanity["list"] = temp; }
 512
 513                 while(!in.eof()){
 514
 515                         selectedCount = 0;
 516
 517                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 518
 519                         //read in list vector
 520                         ListVector list(in);
 521
 522                         //make a new list vector
 523                         ListVector newList;
 524                         newList.setLabel(list.getLabel());
 525
 526                         //for each bin
 527                         for (int i = 0; i < list.getNumBins(); i++) {
 528
 529                                 //parse out names that are in accnos file
 530                                 string binnames = list.get(i);
 531
 532                                 string newNames = "";
 533                                 while (binnames.find_first_of(',') != -1) {
 534                                         string name = binnames.substr(0,binnames.find_first_of(','));
 535                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 536
 537                                         //if that name is in the .accnos file, add it
 538                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++; if (m->debug) { sanity["list"].insert(name); } }
 539                                 }
 540
 541                                 //get last name
 542                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++;  if (m->debug) { sanity["list"].insert(binnames); } }
 543
 544                                 //if there are names in this bin add to new list
 545                                 if (newNames != "") {
 546                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 547                                         newList.push_back(newNames);
 548                                 }
 549                         }
 550
 551                         //print new listvector
 552                         if (newList.getNumBins() != 0) {
 553                                 wroteSomething = true;
 554                                 newList.print(out);
 555                         }
 556
 557                         m->gobble(in);
 558                 }
 559                 in.close();
 560                 out.close();
 561
 562                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 563                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 564
 565                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 566
 567                 return 0;
 568
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "GetSeqsCommand", "readList");
 572                 exit(1);
 573         }
 574 }
 575 //**********************************************************************************************************************
 576 int GetSeqsCommand::readName(){
 577         try {
 578                 string thisOutputDir = outputDir;
 579                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 580                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
 581                 ofstream out;
 582                 m->openOutputFile(outputFileName, out);
 583
 584
 585                 ifstream in;
 586                 m->openInputFile(namefile, in);
 587                 string name, firstCol, secondCol;
 588
 589                 bool wroteSomething = false;
 590                 int selectedCount = 0;
 591
 592         if (m->debug) { set<string> temp; sanity["name"] = temp; }
 593         if (m->debug) { set<string> temp; sanity["dupname"] = temp; }
 594
 595                 while(!in.eof()){
 596
 597                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 598
 599                         in >> firstCol;
 600                         in >> secondCol;
 601
 602                         string hold = "";
 603                         if (dups) { hold = secondCol; }
 604
 605                         vector<string> parsedNames;
 606                         m->splitAtComma(secondCol, parsedNames);
 607
 608                         vector<string> validSecond;
 609                         for (int i = 0; i < parsedNames.size(); i++) {
 610                                 if (names.count(parsedNames[i]) != 0) {
 611                                         validSecond.push_back(parsedNames[i]);
 612                     if (m->debug) { sanity["dupname"].insert(parsedNames[i]); }
 613                                 }
 614                         }
 615
 616                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 617                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]); if (m->debug) { sanity["dupname"].insert(parsedNames[i]); } }
 618                                 out << firstCol << '\t' << hold << endl;
 619                                 wroteSomething = true;
 620                                 selectedCount += parsedNames.size();
 621                 if (m->debug) { sanity["name"].insert(firstCol); }
 622                         }else {
 623                                 selectedCount += validSecond.size();
 624
 625                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 626                                 if (names.count(firstCol) != 0) {
 627
 628                                         wroteSomething = true;
 629
 630                                         out << firstCol << '\t';
 631
 632                                         //you know you have at least one valid second since first column is valid
 633                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 634                                         out << validSecond[validSecond.size()-1] << endl;
 635
 636                     if (m->debug) { sanity["name"].insert(firstCol); }
 637
 638
 639                                 //make first name in set you come to first column and then add the remaining names to second column
 640                                 }else {
 641                                         //you want part of this row
 642                                         if (validSecond.size() != 0) {
 643
 644                                                 wroteSomething = true;
 645
 646                                                 out << validSecond[0] << '\t';
 647
 648                                                 //you know you have at least one valid second since first column is valid
 649                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 650                                                 out << validSecond[validSecond.size()-1] << endl;
 651
 652                         if (m->debug) { sanity["name"].insert(validSecond[0]); }
 653                                         }
 654                                 }
 655                         }
 656                         m->gobble(in);
 657                 }
 658                 in.close();
 659                 out.close();
 660
 661                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 662                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 663
 664                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 665
 666                 return 0;
 667
 668         }
 669         catch(exception& e) {
 670                 m->errorOut(e, "GetSeqsCommand", "readName");
 671                 exit(1);
 672         }
 673 }
 674
 675 //**********************************************************************************************************************
 676 int GetSeqsCommand::readGroup(){
 677         try {
 678                 string thisOutputDir = outputDir;
 679                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 680                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
 681                 ofstream out;
 682                 m->openOutputFile(outputFileName, out);
 683
 684
 685                 ifstream in;
 686                 m->openInputFile(groupfile, in);
 687                 string name, group;
 688
 689                 bool wroteSomething = false;
 690                 int selectedCount = 0;
 691
 692         if (m->debug) { set<string> temp; sanity["group"] = temp; }
 693
 694                 while(!in.eof()){
 695
 696                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 697
 698
 699                         in >> name;                             //read from first column
 700                         in >> group;                    //read from second column
 701
 702                         //if this name is in the accnos file
 703                         if (names.count(name) != 0) {
 704                                 wroteSomething = true;
 705
 706                                 out << name << '\t' << group << endl;
 707                                 selectedCount++;
 708
 709                 if (m->debug) {  sanity["group"].insert(name); }
 710                         }
 711
 712                         m->gobble(in);
 713                 }
 714                 in.close();
 715                 out.close();
 716
 717                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 718                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 719
 720                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 721
 722
 723                 return 0;
 724
 725         }
 726         catch(exception& e) {
 727                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 728                 exit(1);
 729         }
 730 }
 731 //**********************************************************************************************************************
 732 int GetSeqsCommand::readTax(){
 733         try {
 734                 string thisOutputDir = outputDir;
 735                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 736                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + getOutputFileNameTag("taxonomy", taxfile);
 737                 ofstream out;
 738                 m->openOutputFile(outputFileName, out);
 739
 740                 ifstream in;
 741                 m->openInputFile(taxfile, in);
 742                 string name, tax;
 743
 744                 bool wroteSomething = false;
 745                 int selectedCount = 0;
 746
 747         if (m->debug) { set<string> temp; sanity["tax"] = temp; }
 748
 749                 while(!in.eof()){
 750
 751                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 752
 753                         in >> name;                             //read from first column
 754                         in >> tax;                      //read from second column
 755
 756                         //if this name is in the accnos file
 757                         if (names.count(name) != 0) {
 758                                 wroteSomething = true;
 759
 760                                 out << name << '\t' << tax << endl;
 761                                 selectedCount++;
 762
 763                 if (m->debug) { sanity["tax"].insert(name); }
 764                         }
 765
 766                         m->gobble(in);
 767                 }
 768                 in.close();
 769                 out.close();
 770
 771                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 772                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 773
 774                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your taxonomy file."); m->mothurOutEndLine();
 775
 776                 return 0;
 777
 778         }
 779         catch(exception& e) {
 780                 m->errorOut(e, "GetSeqsCommand", "readTax");
 781                 exit(1);
 782         }
 783 }
 784 //**********************************************************************************************************************
 785 //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
 786 int GetSeqsCommand::readAlign(){
 787         try {
 788                 string thisOutputDir = outputDir;
 789                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 790                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + getOutputFileNameTag("alignreport");
 791                 ofstream out;
 792                 m->openOutputFile(outputFileName, out);
 793
 794
 795                 ifstream in;
 796                 m->openInputFile(alignfile, in);
 797                 string name, junk;
 798
 799                 bool wroteSomething = false;
 800                 int selectedCount = 0;
 801
 802                 //read column headers
 803                 for (int i = 0; i < 16; i++) {
 804                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 805                         else                    {       break;                  }
 806                 }
 807                 out << endl;
 808
 809                 while(!in.eof()){
 810
 811                         if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 812
 813
 814                         in >> name;                             //read from first column
 815
 816                         //if this name is in the accnos file
 817                         if (names.count(name) != 0) {
 818                                 wroteSomething = true;
 819                                 selectedCount++;
 820
 821                                 out << name << '\t';
 822
 823                                 //read rest
 824                                 for (int i = 0; i < 15; i++) {
 825                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 826                                         else                    {       break;                  }
 827                                 }
 828                                 out << endl;
 829
 830                         }else {//still read just don't do anything with it
 831                                 //read rest
 832                                 for (int i = 0; i < 15; i++) {
 833                                         if (!in.eof())  {       in >> junk;             }
 834                                         else                    {       break;                  }
 835                                 }
 836                         }
 837
 838                         m->gobble(in);
 839                 }
 840                 in.close();
 841                 out.close();
 842
 843                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 844                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 845
 846                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your alignreport file."); m->mothurOutEndLine();
 847
 848                 return 0;
 849
 850         }
 851         catch(exception& e) {
 852                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 853                 exit(1);
 854         }
 855 }
 856 //**********************************************************************************************************************
 857 //just looking at common mistakes.
 858 int GetSeqsCommand::runSanityCheck(){
 859         try {
 860         string thisOutputDir = outputDir;
 861         if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 862         string filename = outputDir + "get.seqs.debug.report";
 863
 864         ofstream out;
 865                 m->openOutputFile(filename, out);
 866
 867
 868         //compare fasta, name, qual and taxonomy if given to make sure they contain the same seqs
 869         if (fastafile != "") {
 870             if (namefile != "") { //compare with fasta
 871                 if (sanity["fasta"] != sanity["name"]) { //create mismatch file
 872                     createMisMatchFile(out, fastafile, namefile, sanity["fasta"], sanity["name"]);
 873                 }
 874             }
 875             if (qualfile != "") {
 876                 if (sanity["fasta"] != sanity["qual"]) { //create mismatch file
 877                     createMisMatchFile(out, fastafile, qualfile, sanity["fasta"], sanity["qual"]);
 878                 }
 879             }
 880             if (taxfile != "") {
 881                 if (sanity["fasta"] != sanity["tax"]) { //create mismatch file
 882                     createMisMatchFile(out, fastafile, taxfile, sanity["fasta"], sanity["tax"]);
 883                 }
 884             }
 885         }
 886
 887         //compare dupnames, groups and list if given to make sure they match
 888         if (namefile != "") {
 889             if (groupfile != "") {
 890                 if (sanity["dupname"] != sanity["group"]) { //create mismatch file
 891                     createMisMatchFile(out, namefile, groupfile, sanity["dupname"], sanity["group"]);
 892                 }
 893             }
 894             if (listfile != "") {
 895                 if (sanity["dupname"] != sanity["list"]) { //create mismatch file
 896                     createMisMatchFile(out, namefile, listfile, sanity["dupname"], sanity["list"]);
 897                 }
 898             }
 899         }else{
 900
 901             if ((groupfile != "") && (fastafile != "")) {
 902                 if (sanity["fasta"] != sanity["group"]) { //create mismatch file
 903                     createMisMatchFile(out, fastafile, groupfile, sanity["fasta"], sanity["group"]);
 904                 }
 905             }
 906         }
 907
 908         out.close();
 909
 910         if (m->isBlank(filename)) { m->mothurRemove(filename); }
 911         else { m->mothurOut("\n[DEBUG]: " + filename + " contains the file mismatches.\n");outputNames.push_back(filename); outputTypes["debug"].push_back(filename); }
 912
 913         return 0;
 914     }
 915         catch(exception& e) {
 916                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
 917                 exit(1);
 918         }
 919 }
 920 //**********************************************************************************************************************
 921 //just looking at common mistakes.
 922 int GetSeqsCommand::createMisMatchFile(ofstream& out, string filename1, string filename2, set<string> set1, set<string> set2){
 923         try {
 924         out << "****************************************" << endl << endl;
 925         out << "Names unique to " << filename1 << ":\n";
 926
 927         //remove names in set1 that are also in set2
 928         for (set<string>::iterator it = set1.begin(); it != set1.end();) {
 929             string name = *it;
 930
 931             if (set2.count(name) == 0)  { out << name << endl;  } //name unique to set1
 932             else                        { set2.erase(name);     } //you are in both so erase
 933             set1.erase(it++);
 934         }
 935
 936         out << "\nNames unique to " << filename2 << ":\n";
 937         //output results
 938         for (set<string>::iterator it = set2.begin(); it != set2.end(); it++) {  out << *it << endl;  }
 939
 940         out << "****************************************" << endl << endl;
 941
 942         return 0;
 943     }
 944         catch(exception& e) {
 945                 m->errorOut(e, "GetSeqsCommand", "runSanityCheck");
 946                 exit(1);
 947         }
 948 }
 949 //**********************************************************************************************************************
 950
 951 int GetSeqsCommand::compareAccnos(){
 952         try {
 953
 954                 string thisOutputDir = outputDir;
 955                 if (outputDir == "") {  thisOutputDir += m->hasPath(accnosfile);  }
 956                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(accnosfile)) + getOutputFileNameTag("accnosreport");
 957                 ofstream out;
 958                 m->openOutputFile(outputFileName, out);
 959
 960                 ifstream in;
 961                 m->openInputFile(accnosfile2, in);
 962                 string name;
 963
 964                 set<string> namesAccnos2;
 965                 set<string> namesDups;
 966                 set<string> namesAccnos = names;
 967
 968                 map<string, int> nameCount;
 969
 970                 if (namefile != "") {
 971                         ifstream inName;
 972                         m->openInputFile(namefile, inName);
 973
 974
 975                         while(!inName.eof()){
 976
 977                                 if (m->control_pressed) { inName.close(); return 0; }
 978
 979                                 string thisname, repnames;
 980
 981                                 inName >> thisname;             m->gobble(inName);              //read from first column
 982                                 inName >> repnames;                     //read from second column
 983
 984                                 int num = m->getNumNames(repnames);
 985                                 nameCount[thisname] = num;
 986
 987                                 m->gobble(inName);
 988                         }
 989                         inName.close();
 990                 }
 991
 992                 while(!in.eof()){
 993                         in >> name;
 994
 995                         if (namesAccnos.count(name) == 0){ //name unique to accnos2
 996                                 int pos = name.find_last_of('_');
 997                                 string tempName = name;
 998                                 if (pos != string::npos) {  tempName = tempName.substr(pos+1); cout << tempName << endl; }
 999                                 if (namesAccnos.count(tempName) == 0){
1000                                         namesAccnos2.insert(name);
1001                                 }else { //you are in both so erase
1002                                         namesAccnos.erase(name);
1003                                         namesDups.insert(name);
1004                                 }
1005                         }else { //you are in both so erase
1006                                 namesAccnos.erase(name);
1007                                 namesDups.insert(name);
1008                         }
1009
1010                         m->gobble(in);
1011                 }
1012                 in.close();
1013
1014                 out << "Names in both files : " + toString(namesDups.size()) << endl;
1015                 m->mothurOut("Names in both files : " + toString(namesDups.size())); m->mothurOutEndLine();
1016
1017                 for (set<string>::iterator it = namesDups.begin(); it != namesDups.end(); it++) {
1018                         out << (*it);
1019                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1020                         out << endl;
1021                 }
1022
1023                 out << "Names unique to " + accnosfile + " : " + toString(namesAccnos.size()) << endl;
1024                 m->mothurOut("Names unique to " + accnosfile + " : " + toString(namesAccnos.size())); m->mothurOutEndLine();
1025
1026                 for (set<string>::iterator it = namesAccnos.begin(); it != namesAccnos.end(); it++) {
1027                         out << (*it);
1028                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1029                         out << endl;
1030                 }
1031
1032                 out << "Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size()) << endl;
1033                 m->mothurOut("Names unique to " + accnosfile2 + " : " + toString(namesAccnos2.size())); m->mothurOutEndLine();
1034
1035                 for (set<string>::iterator it = namesAccnos2.begin(); it != namesAccnos2.end(); it++) {
1036                         out << (*it);
1037                         if (namefile != "") { out << '\t' << nameCount[(*it)]; }
1038                         out << endl;
1039                 }
1040
1041                 out.close();
1042
1043                 outputNames.push_back(outputFileName);  outputTypes["accnosreport"].push_back(outputFileName);
1044
1045                 return 0;
1046
1047         }
1048         catch(exception& e) {
1049                 m->errorOut(e, "GetSeqsCommand", "compareAccnos");
1050                 exit(1);
1051         }
1052 }
1053
1054
1055 //**********************************************************************************************************************
1056