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 = 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 = 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 = 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 = 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 = 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 = 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 = 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);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 137                         dups = isTrue(temp);
 138
 139                         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; }
 140
 141                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 142                 }
 143
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 147                 exit(1);
 148         }
 149 }
 150 //**********************************************************************************************************************
 151
 152 void RemoveSeqsCommand::help(){
 153         try {
 154                 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");
 155                 m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
 156                 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");
 157                 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=false. \n");
 158                 m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 159                 m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 160                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 161         }
 162         catch(exception& e) {
 163                 m->errorOut(e, "RemoveSeqsCommand", "help");
 164                 exit(1);
 165         }
 166 }
 167
 168 //**********************************************************************************************************************
 169
 170 int RemoveSeqsCommand::execute(){
 171         try {
 172
 173                 if (abort == true) { return 0; }
 174
 175                 //get names you want to keep
 176                 readAccnos();
 177
 178                 if (m->control_pressed) { return 0; }
 179
 180                 //read through the correct file and output lines you want to keep
 181                 if (namefile != "")                     {               readName();             }
 182                 if (fastafile != "")            {               readFasta();    }
 183                 if (groupfile != "")            {               readGroup();    }
 184                 if (alignfile != "")            {               readAlign();    }
 185                 if (listfile != "")                     {               readList();             }
 186                 if (taxfile != "")                      {               readTax();              }
 187
 188                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 189
 190                 if (outputNames.size() != 0) {
 191                         m->mothurOutEndLine();
 192                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 193                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 194                         m->mothurOutEndLine();
 195                 }
 196
 197                 return 0;
 198         }
 199
 200         catch(exception& e) {
 201                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 202                 exit(1);
 203         }
 204 }
 205
 206 //**********************************************************************************************************************
 207 int RemoveSeqsCommand::readFasta(){
 208         try {
 209                 if (outputDir == "") {  outputDir += hasPath(fastafile);  }
 210                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" + getExtension(fastafile);
 211                 ofstream out;
 212                 openOutputFile(outputFileName, out);
 213
 214                 ifstream in;
 215                 openInputFile(fastafile, in);
 216                 string name;
 217
 218                 bool wroteSomething = false;
 219
 220                 while(!in.eof()){
 221                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 222
 223                         Sequence currSeq(in);
 224                         name = currSeq.getName();
 225
 226                         if (name != "") {
 227                                 //if this name is in the accnos file
 228                                 if (names.count(name) == 0) {
 229                                         wroteSomething = true;
 230
 231                                         currSeq.printSequence(out);
 232                                 }
 233                         }
 234                         gobble(in);
 235                 }
 236                 in.close();
 237                 out.close();
 238
 239                 if (wroteSomething == false) {
 240                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 241                         remove(outputFileName.c_str());
 242                 }else { outputNames.push_back(outputFileName); }
 243
 244                 return 0;
 245
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 249                 exit(1);
 250         }
 251 }
 252 //**********************************************************************************************************************
 253 int RemoveSeqsCommand::readList(){
 254         try {
 255                 if (outputDir == "") {  outputDir += hasPath(listfile);  }
 256                 string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" +  getExtension(listfile);
 257                 ofstream out;
 258                 openOutputFile(outputFileName, out);
 259
 260                 ifstream in;
 261                 openInputFile(listfile, in);
 262
 263                 bool wroteSomething = false;
 264
 265                 while(!in.eof()){
 266                         //read in list vector
 267                         ListVector list(in);
 268
 269                         //make a new list vector
 270                         ListVector newList;
 271                         newList.setLabel(list.getLabel());
 272
 273                         //for each bin
 274                         for (int i = 0; i < list.getNumBins(); i++) {
 275                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 276
 277                                 //parse out names that are in accnos file
 278                                 string binnames = list.get(i);
 279
 280                                 string newNames = "";
 281                                 while (binnames.find_first_of(',') != -1) {
 282                                         string name = binnames.substr(0,binnames.find_first_of(','));
 283                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 284
 285                                         //if that name is in the .accnos file, add it
 286                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 287                                 }
 288
 289                                 //get last name
 290                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 291
 292                                 //if there are names in this bin add to new list
 293                                 if (newNames != "") {
 294                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 295                                         newList.push_back(newNames);
 296                                 }
 297                         }
 298
 299                         //print new listvector
 300                         if (newList.getNumBins() != 0) {
 301                                 wroteSomething = true;
 302                                 newList.print(out);
 303                         }
 304
 305                         gobble(in);
 306                 }
 307                 in.close();
 308                 out.close();
 309
 310                 if (wroteSomething == false) {
 311                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 312                         remove(outputFileName.c_str());
 313                 }else { 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 += hasPath(namefile);  }
 327                 string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" + getExtension(namefile);
 328
 329                 ofstream out;
 330                 openOutputFile(outputFileName, out);
 331
 332                 ifstream in;
 333                 openInputFile(namefile, in);
 334                 string name, firstCol, secondCol;
 335
 336                 bool wroteSomething = false;
 337         int numSeqsInNames = 0;
 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
 353                         //get name after last ,
 354                         parsedNames.push_back(secondCol);
 355
 356                         vector<string> validSecond;  validSecond.clear();
 357                         for (int i = 0; i < parsedNames.size(); i++) {
 358                                 if (names.count(parsedNames[i]) == 0) {
 359                                         validSecond.push_back(parsedNames[i]);
 360                                 }
 361                         }
 362                 numSeqsInNames += parsedNames.size();
 363                         if ((dups) && (validSecond.size() != parsedNames.size())) {  //if dups is true and we want to get rid of anyone, get rid of everyone
 364                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 365                         }else {
 366                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 367                                 if (names.count(firstCol) == 0) {
 368
 369                                         wroteSomething = true;
 370
 371                                         out << firstCol << '\t';
 372
 373                                         //you know you have at least one valid second since first column is valid
 374                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 375                                         out << validSecond[validSecond.size()-1] << endl;
 376
 377                                         //make first name in set you come to first column and then add the remaining names to second column
 378                                 }else {
 379
 380                                         //you want part of this row
 381                                         if (validSecond.size() != 0) {
 382
 383                                                 wroteSomething = true;
 384
 385                                                 out << validSecond[0] << '\t';
 386
 387                                                 //you know you have at least one valid second since first column is valid
 388                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 389                                                 out << validSecond[validSecond.size()-1] << endl;
 390                                         }
 391                                 }
 392                         }
 393                         gobble(in);
 394                 }
 395                 in.close();
 396                 out.close();
 397 cout << "your names file contains " << numSeqsInNames << endl;
 398                 if (wroteSomething == false) {
 399                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 400                         remove(outputFileName.c_str());
 401                 }else { outputNames.push_back(outputFileName); }
 402
 403                 return 0;
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 407                 exit(1);
 408         }
 409 }
 410
 411 //**********************************************************************************************************************
 412 int RemoveSeqsCommand::readGroup(){
 413         try {
 414                 if (outputDir == "") {  outputDir += hasPath(groupfile);  }
 415                 string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
 416                 ofstream out;
 417                 openOutputFile(outputFileName, out);
 418
 419                 ifstream in;
 420                 openInputFile(groupfile, in);
 421                 string name, group;
 422
 423                 bool wroteSomething = false;
 424
 425                 while(!in.eof()){
 426                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 427
 428                         in >> name;                             //read from first column
 429                         in >> group;                    //read from second column
 430
 431                         //if this name is in the accnos file
 432                         if (names.count(name) == 0) {
 433                                 wroteSomething = true;
 434                                 out << name << '\t' << group << endl;
 435                         }
 436
 437                         gobble(in);
 438                 }
 439                 in.close();
 440                 out.close();
 441
 442                 if (wroteSomething == false) {
 443                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 444                         remove(outputFileName.c_str());
 445                 }else { outputNames.push_back(outputFileName); }
 446
 447                 return 0;
 448         }
 449         catch(exception& e) {
 450                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 451                 exit(1);
 452         }
 453 }
 454 //**********************************************************************************************************************
 455 int RemoveSeqsCommand::readTax(){
 456         try {
 457                 if (outputDir == "") {  outputDir += hasPath(taxfile);  }
 458                 string outputFileName = outputDir + getRootName(getSimpleName(taxfile)) + "pick" + getExtension(taxfile);
 459                 ofstream out;
 460                 openOutputFile(outputFileName, out);
 461
 462                 ifstream in;
 463                 openInputFile(taxfile, in);
 464                 string name, tax;
 465
 466                 bool wroteSomething = false;
 467
 468                 while(!in.eof()){
 469                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 470
 471                         in >> name;                             //read from first column
 472                         in >> tax;                      //read from second column
 473
 474                         //if this name is in the accnos file
 475                         if (names.count(name) == 0) {
 476                                 wroteSomething = true;
 477                                 out << name << '\t' << tax << endl;
 478                         }
 479
 480                         gobble(in);
 481                 }
 482                 in.close();
 483                 out.close();
 484
 485                 if (wroteSomething == false) {
 486                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 487                         remove(outputFileName.c_str());
 488                 }else { outputNames.push_back(outputFileName); }
 489
 490                 return 0;
 491         }
 492         catch(exception& e) {
 493                 m->errorOut(e, "RemoveSeqsCommand", "readTax");
 494                 exit(1);
 495         }
 496 }
 497 //**********************************************************************************************************************
 498 //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
 499 int RemoveSeqsCommand::readAlign(){
 500         try {
 501                 if (outputDir == "") {  outputDir += hasPath(alignfile);  }
 502                 string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
 503                 ofstream out;
 504                 openOutputFile(outputFileName, out);
 505
 506                 ifstream in;
 507                 openInputFile(alignfile, in);
 508                 string name, junk;
 509
 510                 bool wroteSomething = false;
 511
 512                 //read column headers
 513                 for (int i = 0; i < 16; i++) {
 514                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 515                         else                    {       break;                  }
 516                 }
 517                 out << endl;
 518
 519                 while(!in.eof()){
 520                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 521
 522                         in >> name;                             //read from first column
 523
 524                         //if this name is in the accnos file
 525                         if (names.count(name) == 0) {
 526                                 wroteSomething = true;
 527
 528                                 out << name << '\t';
 529
 530                                 //read rest
 531                                 for (int i = 0; i < 15; i++) {
 532                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 533                                         else                    {       break;                  }
 534                                 }
 535                                 out << endl;
 536
 537                         }else {//still read just don't do anything with it
 538
 539                                 //read rest
 540                                 for (int i = 0; i < 15; i++) {
 541                                         if (!in.eof())  {       in >> junk;             }
 542                                         else                    {       break;                  }
 543                                 }
 544                         }
 545
 546                         gobble(in);
 547                 }
 548                 in.close();
 549                 out.close();
 550
 551                 if (wroteSomething == false) {
 552                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 553                         remove(outputFileName.c_str());
 554                 }else { outputNames.push_back(outputFileName); }
 555
 556                 return 0;
 557
 558         }
 559         catch(exception& e) {
 560                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 561                 exit(1);
 562         }
 563 }
 564 //**********************************************************************************************************************
 565 void RemoveSeqsCommand::readAccnos(){
 566         try {
 567
 568                 ifstream in;
 569                 openInputFile(accnosfile, in);
 570                 string name;
 571
 572                 while(!in.eof()){
 573                         in >> name;
 574
 575                         names.insert(name);
 576
 577                         gobble(in);
 578                 }
 579                 in.close();
 580
 581         }
 582         catch(exception& e) {
 583                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 584                 exit(1);
 585         }
 586 }
 587
 588 //**********************************************************************************************************************
 589
 590