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
  14 //**********************************************************************************************************************
  15
  16 RemoveSeqsCommand::RemoveSeqsCommand(string option)  {
  17         try {
  18                 abort = false;
  19
  20                 //allow user to run help
  21                 if(option == "help") { help(); abort = true; }
  22
  23                 else {
  24                         //valid paramters for this command
  25                         string Array[] =  {"fasta","name", "group", "alignreport", "accnos", "list","taxonomy","outputdir","inputdir", "dups" };
  26                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  27
  28                         OptionParser parser(option);
  29                         map<string,string> parameters = parser.getParameters();
  30
  31                         ValidParameters validParameter;
  32                         map<string,string>::iterator it;
  33
  34                         //check to make sure all parameters are valid for command
  35                         for (it = parameters.begin(); it != parameters.end(); it++) {
  36                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  37                         }
  38
  39                         //if the user changes the output directory command factory will send this info to us in the output parameter
  40                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
  41
  42                         //if the user changes the input directory command factory will send this info to us in the output parameter
  43                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  44                         if (inputDir == "not found"){   inputDir = "";          }
  45                         else {
  46                                 string path;
  47                                 it = parameters.find("alignreport");
  48                                 //user has given a template file
  49                                 if(it != parameters.end()){
  50                                         path = m->hasPath(it->second);
  51                                         //if the user has not given a path then, add inputdir. else leave path alone.
  52                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
  53                                 }
  54
  55                                 it = parameters.find("fasta");
  56                                 //user has given a template file
  57                                 if(it != parameters.end()){
  58                                         path = m->hasPath(it->second);
  59                                         //if the user has not given a path then, add inputdir. else leave path alone.
  60                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
  61                                 }
  62
  63                                 it = parameters.find("accnos");
  64                                 //user has given a template file
  65                                 if(it != parameters.end()){
  66                                         path = m->hasPath(it->second);
  67                                         //if the user has not given a path then, add inputdir. else leave path alone.
  68                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
  69                                 }
  70
  71                                 it = parameters.find("list");
  72                                 //user has given a template file
  73                                 if(it != parameters.end()){
  74                                         path = m->hasPath(it->second);
  75                                         //if the user has not given a path then, add inputdir. else leave path alone.
  76                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  77                                 }
  78
  79                                 it = parameters.find("name");
  80                                 //user has given a template file
  81                                 if(it != parameters.end()){
  82                                         path = m->hasPath(it->second);
  83                                         //if the user has not given a path then, add inputdir. else leave path alone.
  84                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  85                                 }
  86
  87                                 it = parameters.find("group");
  88                                 //user has given a template file
  89                                 if(it != parameters.end()){
  90                                         path = m->hasPath(it->second);
  91                                         //if the user has not given a path then, add inputdir. else leave path alone.
  92                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  93                                 }
  94
  95                                 it = parameters.find("taxonomy");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 101                                 }
 102                         }
 103
 104
 105                         //check for required parameters
 106                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 107                         if (accnosfile == "not open") { abort = true; }
 108                         else if (accnosfile == "not found") {  accnosfile = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 109
 110                         fastafile = validParameter.validFile(parameters, "fasta", true);
 111                         if (fastafile == "not open") { abort = true; }
 112                         else if (fastafile == "not found") {  fastafile = "";  }
 113
 114                         namefile = validParameter.validFile(parameters, "name", true);
 115                         if (namefile == "not open") { abort = true; }
 116                         else if (namefile == "not found") {  namefile = "";  }
 117
 118                         groupfile = validParameter.validFile(parameters, "group", true);
 119                         if (groupfile == "not open") { abort = true; }
 120                         else if (groupfile == "not found") {  groupfile = "";  }
 121
 122                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 123                         if (alignfile == "not open") { abort = true; }
 124                         else if (alignfile == "not found") {  alignfile = "";  }
 125
 126                         listfile = validParameter.validFile(parameters, "list", true);
 127                         if (listfile == "not open") { abort = true; }
 128                         else if (listfile == "not found") {  listfile = "";  }
 129
 130                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 131                         if (taxfile == "not open") { abort = true; }
 132                         else if (taxfile == "not found") {  taxfile = "";  }
 133
 134
 135                         string usedDups = "true";
 136                         string temp = validParameter.validFile(parameters, "dups", false);
 137                         if (temp == "not found") {
 138                                 if (namefile != "") {  temp = "true";                                   }
 139                                 else                            {  temp = "false"; usedDups = "";       }
 140                         }
 141                         dups = m->isTrue(temp);
 142
 143                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, group, taxonomy, alignreport or list."); m->mothurOutEndLine(); abort = true; }
 144
 145                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 146                 }
 147
 148         }
 149         catch(exception& e) {
 150                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 151                 exit(1);
 152         }
 153 }
 154 //**********************************************************************************************************************
 155
 156 void RemoveSeqsCommand::help(){
 157         try {
 158                 m->mothurOut("The remove.seqs command reads an .accnos file and at least one of the following file types: fasta, name, group, list, taxonomy or alignreport file.\n");
 159                 m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
 160                 m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide accnos and at least one of the file parameters.\n");
 161                 m->mothurOut("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");
 162                 m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 163                 m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 164                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 165         }
 166         catch(exception& e) {
 167                 m->errorOut(e, "RemoveSeqsCommand", "help");
 168                 exit(1);
 169         }
 170 }
 171
 172 //**********************************************************************************************************************
 173
 174 int RemoveSeqsCommand::execute(){
 175         try {
 176
 177                 if (abort == true) { return 0; }
 178
 179                 //get names you want to keep
 180                 readAccnos();
 181
 182                 if (m->control_pressed) { return 0; }
 183
 184                 //read through the correct file and output lines you want to keep
 185                 if (namefile != "")                     {               readName();             }
 186                 if (fastafile != "")            {               readFasta();    }
 187                 if (groupfile != "")            {               readGroup();    }
 188                 if (alignfile != "")            {               readAlign();    }
 189                 if (listfile != "")                     {               readList();             }
 190                 if (taxfile != "")                      {               readTax();              }
 191
 192                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 193
 194                 if (outputNames.size() != 0) {
 195                         m->mothurOutEndLine();
 196                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 197                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 198                         m->mothurOutEndLine();
 199                 }
 200
 201                 return 0;
 202         }
 203
 204         catch(exception& e) {
 205                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 206                 exit(1);
 207         }
 208 }
 209
 210 //**********************************************************************************************************************
 211 int RemoveSeqsCommand::readFasta(){
 212         try {
 213                 string thisOutputDir = outputDir;
 214                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 215                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 216
 217                 ofstream out;
 218                 m->openOutputFile(outputFileName, out);
 219
 220                 ifstream in;
 221                 m->openInputFile(fastafile, in);
 222                 string name;
 223
 224                 bool wroteSomething = false;
 225
 226                 while(!in.eof()){
 227                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 228
 229                         Sequence currSeq(in);
 230                         name = currSeq.getName();
 231
 232                         if (name != "") {
 233                                 //if this name is in the accnos file
 234                                 if (names.count(name) == 0) {
 235                                         wroteSomething = true;
 236
 237                                         currSeq.printSequence(out);
 238                                 }
 239                         }
 240                         m->gobble(in);
 241                 }
 242                 in.close();
 243                 out.close();
 244
 245                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 246                 outputNames.push_back(outputFileName);
 247
 248                 return 0;
 249
 250         }
 251         catch(exception& e) {
 252                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 253                 exit(1);
 254         }
 255 }
 256 //**********************************************************************************************************************
 257 int RemoveSeqsCommand::readList(){
 258         try {
 259                 string thisOutputDir = outputDir;
 260                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 261                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 262
 263                 ofstream out;
 264                 m->openOutputFile(outputFileName, out);
 265
 266                 ifstream in;
 267                 m->openInputFile(listfile, in);
 268
 269                 bool wroteSomething = false;
 270
 271                 while(!in.eof()){
 272                         //read in list vector
 273                         ListVector list(in);
 274
 275                         //make a new list vector
 276                         ListVector newList;
 277                         newList.setLabel(list.getLabel());
 278
 279                         //for each bin
 280                         for (int i = 0; i < list.getNumBins(); i++) {
 281                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 282
 283                                 //parse out names that are in accnos file
 284                                 string binnames = list.get(i);
 285
 286                                 string newNames = "";
 287                                 while (binnames.find_first_of(',') != -1) {
 288                                         string name = binnames.substr(0,binnames.find_first_of(','));
 289                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 290
 291                                         //if that name is in the .accnos file, add it
 292                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 293                                 }
 294
 295                                 //get last name
 296                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 297
 298                                 //if there are names in this bin add to new list
 299                                 if (newNames != "") {
 300                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 301                                         newList.push_back(newNames);
 302                                 }
 303                         }
 304
 305                         //print new listvector
 306                         if (newList.getNumBins() != 0) {
 307                                 wroteSomething = true;
 308                                 newList.print(out);
 309                         }
 310
 311                         m->gobble(in);
 312                 }
 313                 in.close();
 314                 out.close();
 315
 316                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 317                 outputNames.push_back(outputFileName);
 318
 319                 return 0;
 320
 321         }
 322         catch(exception& e) {
 323                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 324                 exit(1);
 325         }
 326 }
 327 //**********************************************************************************************************************
 328 int RemoveSeqsCommand::readName(){
 329         try {
 330                 string thisOutputDir = outputDir;
 331                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 332                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 333
 334                 ofstream out;
 335                 m->openOutputFile(outputFileName, out);
 336
 337                 ifstream in;
 338                 m->openInputFile(namefile, in);
 339                 string name, firstCol, secondCol;
 340
 341                 bool wroteSomething = false;
 342
 343                 while(!in.eof()){
 344                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 345
 346                         in >> firstCol;
 347                         in >> secondCol;
 348
 349                         vector<string> parsedNames;
 350                         //parse second column saving each name
 351                         while (secondCol.find_first_of(',') != -1) {
 352                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 353                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 354                                 parsedNames.push_back(name);
 355                         }
 356
 357                         //get name after last ,
 358                         parsedNames.push_back(secondCol);
 359
 360                         vector<string> validSecond;  validSecond.clear();
 361                         for (int i = 0; i < parsedNames.size(); i++) {
 362                                 if (names.count(parsedNames[i]) == 0) {
 363                                         validSecond.push_back(parsedNames[i]);
 364                                 }
 365                         }
 366
 367                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 368                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 369                         }else {
 370                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 371                                 if (names.count(firstCol) == 0) {
 372
 373                                         wroteSomething = true;
 374
 375                                         out << firstCol << '\t';
 376
 377                                         //you know you have at least one valid second since first column is valid
 378                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 379                                         out << validSecond[validSecond.size()-1] << endl;
 380
 381                                         //make first name in set you come to first column and then add the remaining names to second column
 382                                 }else {
 383
 384                                         //you want part of this row
 385                                         if (validSecond.size() != 0) {
 386
 387                                                 wroteSomething = true;
 388
 389                                                 out << validSecond[0] << '\t';
 390
 391                                                 //you know you have at least one valid second since first column is valid
 392                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 393                                                 out << validSecond[validSecond.size()-1] << endl;
 394                                         }
 395                                 }
 396                         }
 397                         m->gobble(in);
 398                 }
 399                 in.close();
 400                 out.close();
 401
 402                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 403                 outputNames.push_back(outputFileName);
 404
 405                 return 0;
 406         }
 407         catch(exception& e) {
 408                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 409                 exit(1);
 410         }
 411 }
 412
 413 //**********************************************************************************************************************
 414 int RemoveSeqsCommand::readGroup(){
 415         try {
 416                 string thisOutputDir = outputDir;
 417                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 418                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 419
 420                 ofstream out;
 421                 m->openOutputFile(outputFileName, out);
 422
 423                 ifstream in;
 424                 m->openInputFile(groupfile, in);
 425                 string name, group;
 426
 427                 bool wroteSomething = false;
 428
 429                 while(!in.eof()){
 430                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 431
 432                         in >> name;                             //read from first column
 433                         in >> group;                    //read from second column
 434
 435                         //if this name is in the accnos file
 436                         if (names.count(name) == 0) {
 437                                 wroteSomething = true;
 438                                 out << name << '\t' << group << endl;
 439                         }
 440
 441                         m->gobble(in);
 442                 }
 443                 in.close();
 444                 out.close();
 445
 446                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 447                 outputNames.push_back(outputFileName);
 448
 449                 return 0;
 450         }
 451         catch(exception& e) {
 452                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 453                 exit(1);
 454         }
 455 }
 456 //**********************************************************************************************************************
 457 int RemoveSeqsCommand::readTax(){
 458         try {
 459                 string thisOutputDir = outputDir;
 460                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 461                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 462                 ofstream out;
 463                 m->openOutputFile(outputFileName, out);
 464
 465                 ifstream in;
 466                 m->openInputFile(taxfile, in);
 467                 string name, tax;
 468
 469                 bool wroteSomething = false;
 470
 471                 while(!in.eof()){
 472                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 473
 474                         in >> name;                             //read from first column
 475                         in >> tax;                      //read from second column
 476
 477                         //if this name is in the accnos file
 478                         if (names.count(name) == 0) {
 479                                 wroteSomething = true;
 480                                 out << name << '\t' << tax << endl;
 481                         }
 482
 483                         m->gobble(in);
 484                 }
 485                 in.close();
 486                 out.close();
 487
 488                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 489                 outputNames.push_back(outputFileName);
 490
 491                 return 0;
 492         }
 493         catch(exception& e) {
 494                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 495                 exit(1);
 496         }
 497 }
 498 //**********************************************************************************************************************
 499 //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
 500 int RemoveSeqsCommand::readAlign(){
 501         try {
 502                 string thisOutputDir = outputDir;
 503                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 504                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 505
 506                 ofstream out;
 507                 m->openOutputFile(outputFileName, out);
 508
 509                 ifstream in;
 510                 m->openInputFile(alignfile, in);
 511                 string name, junk;
 512
 513                 bool wroteSomething = false;
 514
 515                 //read column headers
 516                 for (int i = 0; i < 16; i++) {
 517                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 518                         else                    {       break;                  }
 519                 }
 520                 out << endl;
 521
 522                 while(!in.eof()){
 523                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 524
 525                         in >> name;                             //read from first column
 526
 527                         //if this name is in the accnos file
 528                         if (names.count(name) == 0) {
 529                                 wroteSomething = true;
 530
 531                                 out << name << '\t';
 532
 533                                 //read rest
 534                                 for (int i = 0; i < 15; i++) {
 535                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 536                                         else                    {       break;                  }
 537                                 }
 538                                 out << endl;
 539
 540                         }else {//still read just don't do anything with it
 541
 542                                 //read rest
 543                                 for (int i = 0; i < 15; i++) {
 544                                         if (!in.eof())  {       in >> junk;             }
 545                                         else                    {       break;                  }
 546                                 }
 547                         }
 548
 549                         m->gobble(in);
 550                 }
 551                 in.close();
 552                 out.close();
 553
 554                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 555                 outputNames.push_back(outputFileName);
 556
 557                 return 0;
 558
 559         }
 560         catch(exception& e) {
 561                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 562                 exit(1);
 563         }
 564 }
 565 //**********************************************************************************************************************
 566 void RemoveSeqsCommand::readAccnos(){
 567         try {
 568
 569                 ifstream in;
 570                 m->openInputFile(accnosfile, in);
 571                 string name;
 572
 573                 while(!in.eof()){
 574                         in >> name;
 575
 576                         names.insert(name);
 577
 578                         m->gobble(in);
 579                 }
 580                 in.close();
 581
 582         }
 583         catch(exception& e) {
 584                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 585                 exit(1);
 586         }
 587 }
 588
 589 //**********************************************************************************************************************
 590
 591