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                 if (outputDir == "") {  outputDir += m->hasPath(fastafile);  }
 214                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 215                 ofstream out;
 216                 m->openOutputFile(outputFileName, out);
 217
 218                 ifstream in;
 219                 m->openInputFile(fastafile, in);
 220                 string name;
 221
 222                 bool wroteSomething = false;
 223
 224                 while(!in.eof()){
 225                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 226
 227                         Sequence currSeq(in);
 228                         name = currSeq.getName();
 229
 230                         if (name != "") {
 231                                 //if this name is in the accnos file
 232                                 if (names.count(name) == 0) {
 233                                         wroteSomething = true;
 234
 235                                         currSeq.printSequence(out);
 236                                 }
 237                         }
 238                         m->gobble(in);
 239                 }
 240                 in.close();
 241                 out.close();
 242
 243                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 244                 outputNames.push_back(outputFileName);
 245
 246                 return 0;
 247
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 251                 exit(1);
 252         }
 253 }
 254 //**********************************************************************************************************************
 255 int RemoveSeqsCommand::readList(){
 256         try {
 257                 if (outputDir == "") {  outputDir += m->hasPath(listfile);  }
 258                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 259                 ofstream out;
 260                 m->openOutputFile(outputFileName, out);
 261
 262                 ifstream in;
 263                 m->openInputFile(listfile, in);
 264
 265                 bool wroteSomething = false;
 266
 267                 while(!in.eof()){
 268                         //read in list vector
 269                         ListVector list(in);
 270
 271                         //make a new list vector
 272                         ListVector newList;
 273                         newList.setLabel(list.getLabel());
 274
 275                         //for each bin
 276                         for (int i = 0; i < list.getNumBins(); i++) {
 277                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 278
 279                                 //parse out names that are in accnos file
 280                                 string binnames = list.get(i);
 281
 282                                 string newNames = "";
 283                                 while (binnames.find_first_of(',') != -1) {
 284                                         string name = binnames.substr(0,binnames.find_first_of(','));
 285                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 286
 287                                         //if that name is in the .accnos file, add it
 288                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 289                                 }
 290
 291                                 //get last name
 292                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 293
 294                                 //if there are names in this bin add to new list
 295                                 if (newNames != "") {
 296                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 297                                         newList.push_back(newNames);
 298                                 }
 299                         }
 300
 301                         //print new listvector
 302                         if (newList.getNumBins() != 0) {
 303                                 wroteSomething = true;
 304                                 newList.print(out);
 305                         }
 306
 307                         m->gobble(in);
 308                 }
 309                 in.close();
 310                 out.close();
 311
 312                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 313                 outputNames.push_back(outputFileName);
 314
 315                 return 0;
 316
 317         }
 318         catch(exception& e) {
 319                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 320                 exit(1);
 321         }
 322 }
 323 //**********************************************************************************************************************
 324 int RemoveSeqsCommand::readName(){
 325         try {
 326                 if (outputDir == "") {  outputDir += m->hasPath(namefile);  }
 327                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 328
 329                 ofstream out;
 330                 m->openOutputFile(outputFileName, out);
 331
 332                 ifstream in;
 333                 m->openInputFile(namefile, in);
 334                 string name, firstCol, secondCol;
 335
 336                 bool wroteSomething = false;
 337
 338                 while(!in.eof()){
 339                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 340
 341                         in >> firstCol;
 342                         in >> secondCol;
 343
 344                         vector<string> parsedNames;
 345                         //parse second column saving each name
 346                         while (secondCol.find_first_of(',') != -1) {
 347                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 348                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 349                                 parsedNames.push_back(name);
 350                         }
 351
 352                         //get name after last ,
 353                         parsedNames.push_back(secondCol);
 354
 355                         vector<string> validSecond;  validSecond.clear();
 356                         for (int i = 0; i < parsedNames.size(); i++) {
 357                                 if (names.count(parsedNames[i]) == 0) {
 358                                         validSecond.push_back(parsedNames[i]);
 359                                 }
 360                         }
 361
 362                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 363                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 364                         }else {
 365                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 366                                 if (names.count(firstCol) == 0) {
 367
 368                                         wroteSomething = true;
 369
 370                                         out << firstCol << '\t';
 371
 372                                         //you know you have at least one valid second since first column is valid
 373                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 374                                         out << validSecond[validSecond.size()-1] << endl;
 375
 376                                         //make first name in set you come to first column and then add the remaining names to second column
 377                                 }else {
 378
 379                                         //you want part of this row
 380                                         if (validSecond.size() != 0) {
 381
 382                                                 wroteSomething = true;
 383
 384                                                 out << validSecond[0] << '\t';
 385
 386                                                 //you know you have at least one valid second since first column is valid
 387                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 388                                                 out << validSecond[validSecond.size()-1] << endl;
 389                                         }
 390                                 }
 391                         }
 392                         m->gobble(in);
 393                 }
 394                 in.close();
 395                 out.close();
 396
 397                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 398                 outputNames.push_back(outputFileName);
 399
 400                 return 0;
 401         }
 402         catch(exception& e) {
 403                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 404                 exit(1);
 405         }
 406 }
 407
 408 //**********************************************************************************************************************
 409 int RemoveSeqsCommand::readGroup(){
 410         try {
 411                 if (outputDir == "") {  outputDir += m->hasPath(groupfile);  }
 412                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 413                 ofstream out;
 414                 m->openOutputFile(outputFileName, out);
 415
 416                 ifstream in;
 417                 m->openInputFile(groupfile, in);
 418                 string name, group;
 419
 420                 bool wroteSomething = false;
 421
 422                 while(!in.eof()){
 423                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 424
 425                         in >> name;                             //read from first column
 426                         in >> group;                    //read from second column
 427
 428                         //if this name is in the accnos file
 429                         if (names.count(name) == 0) {
 430                                 wroteSomething = true;
 431                                 out << name << '\t' << group << endl;
 432                         }
 433
 434                         m->gobble(in);
 435                 }
 436                 in.close();
 437                 out.close();
 438
 439                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 440                 outputNames.push_back(outputFileName);
 441
 442                 return 0;
 443         }
 444         catch(exception& e) {
 445                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 446                 exit(1);
 447         }
 448 }
 449 //**********************************************************************************************************************
 450 int RemoveSeqsCommand::readTax(){
 451         try {
 452                 if (outputDir == "") {  outputDir += m->hasPath(taxfile);  }
 453                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 454                 ofstream out;
 455                 m->openOutputFile(outputFileName, out);
 456
 457                 ifstream in;
 458                 m->openInputFile(taxfile, in);
 459                 string name, tax;
 460
 461                 bool wroteSomething = false;
 462
 463                 while(!in.eof()){
 464                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 465
 466                         in >> name;                             //read from first column
 467                         in >> tax;                      //read from second column
 468
 469                         //if this name is in the accnos file
 470                         if (names.count(name) == 0) {
 471                                 wroteSomething = true;
 472                                 out << name << '\t' << tax << endl;
 473                         }
 474
 475                         m->gobble(in);
 476                 }
 477                 in.close();
 478                 out.close();
 479
 480                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 481                 outputNames.push_back(outputFileName);
 482
 483                 return 0;
 484         }
 485         catch(exception& e) {
 486                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 487                 exit(1);
 488         }
 489 }
 490 //**********************************************************************************************************************
 491 //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
 492 int RemoveSeqsCommand::readAlign(){
 493         try {
 494                 if (outputDir == "") {  outputDir += m->hasPath(alignfile);  }
 495                 string outputFileName = outputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 496                 ofstream out;
 497                 m->openOutputFile(outputFileName, out);
 498
 499                 ifstream in;
 500                 m->openInputFile(alignfile, in);
 501                 string name, junk;
 502
 503                 bool wroteSomething = false;
 504
 505                 //read column headers
 506                 for (int i = 0; i < 16; i++) {
 507                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 508                         else                    {       break;                  }
 509                 }
 510                 out << endl;
 511
 512                 while(!in.eof()){
 513                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 514
 515                         in >> name;                             //read from first column
 516
 517                         //if this name is in the accnos file
 518                         if (names.count(name) == 0) {
 519                                 wroteSomething = true;
 520
 521                                 out << name << '\t';
 522
 523                                 //read rest
 524                                 for (int i = 0; i < 15; i++) {
 525                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 526                                         else                    {       break;                  }
 527                                 }
 528                                 out << endl;
 529
 530                         }else {//still read just don't do anything with it
 531
 532                                 //read rest
 533                                 for (int i = 0; i < 15; i++) {
 534                                         if (!in.eof())  {       in >> junk;             }
 535                                         else                    {       break;                  }
 536                                 }
 537                         }
 538
 539                         m->gobble(in);
 540                 }
 541                 in.close();
 542                 out.close();
 543
 544                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();  }
 545                 outputNames.push_back(outputFileName);
 546
 547                 return 0;
 548
 549         }
 550         catch(exception& e) {
 551                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 552                 exit(1);
 553         }
 554 }
 555 //**********************************************************************************************************************
 556 void RemoveSeqsCommand::readAccnos(){
 557         try {
 558
 559                 ifstream in;
 560                 m->openInputFile(accnosfile, in);
 561                 string name;
 562
 563                 while(!in.eof()){
 564                         in >> name;
 565
 566                         names.insert(name);
 567
 568                         m->gobble(in);
 569                 }
 570                 in.close();
 571
 572         }
 573         catch(exception& e) {
 574                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 575                 exit(1);
 576         }
 577 }
 578
 579 //**********************************************************************************************************************
 580
 581