removeseqscommand.cpp

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