removegroupscommand.cpp

   1 /*
   2  *  removegroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "removegroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> RemoveGroupsCommand::getValidParameters(){
  17         try {
  18                 string Array[] =  {"fasta","name", "group", "accnos", "groups","list","taxonomy","outputdir","inputdir" };
  19                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  20                 return myArray;
  21         }
  22         catch(exception& e) {
  23                 m->errorOut(e, "RemoveGroupsCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 RemoveGroupsCommand::RemoveGroupsCommand(){
  29         try {
  30                 abort = true; calledHelp = true;
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["taxonomy"] = tempOutNames;
  34                 outputTypes["name"] = tempOutNames;
  35                 outputTypes["group"] = tempOutNames;
  36                 outputTypes["list"] = tempOutNames;
  37         }
  38         catch(exception& e) {
  39                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
  40                 exit(1);
  41         }
  42 }
  43 //**********************************************************************************************************************
  44 vector<string> RemoveGroupsCommand::getRequiredParameters(){
  45         try {
  46                 string Array[] =  {"group"};
  47                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  48                 return myArray;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "RemoveGroupsCommand", "getRequiredParameters");
  52                 exit(1);
  53         }
  54 }
  55 //**********************************************************************************************************************
  56 vector<string> RemoveGroupsCommand::getRequiredFiles(){
  57         try {
  58                 vector<string> myArray;
  59                 return myArray;
  60         }
  61         catch(exception& e) {
  62                 m->errorOut(e, "RemoveGroupsCommand", "getRequiredFiles");
  63                 exit(1);
  64         }
  65 }
  66 //**********************************************************************************************************************
  67 RemoveGroupsCommand::RemoveGroupsCommand(string option)  {
  68         try {
  69                 abort = false; calledHelp = false;
  70
  71                 //allow user to run help
  72                 if(option == "help") { help(); abort = true; calledHelp = true; }
  73
  74                 else {
  75                         //valid paramters for this command
  76                         string Array[] =  {"fasta","name", "group", "accnos", "groups", "list","taxonomy","outputdir","inputdir" };
  77                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  78
  79                         OptionParser parser(option);
  80                         map<string,string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83                         map<string,string>::iterator it;
  84
  85                         //check to make sure all parameters are valid for command
  86                         for (it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["fasta"] = tempOutNames;
  93                         outputTypes["taxonomy"] = tempOutNames;
  94                         outputTypes["name"] = tempOutNames;
  95                         outputTypes["group"] = tempOutNames;
  96                         outputTypes["list"] = tempOutNames;
  97
  98
  99                         //if the user changes the output directory command factory will send this info to us in the output parameter
 100                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 101
 102                         //if the user changes the input directory command factory will send this info to us in the output parameter
 103                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 104                         if (inputDir == "not found"){   inputDir = "";          }
 105                         else {
 106                                 string path;
 107                                 it = parameters.find("fasta");
 108                                 //user has given a template file
 109                                 if(it != parameters.end()){
 110                                         path = m->hasPath(it->second);
 111                                         //if the user has not given a path then, add inputdir. else leave path alone.
 112                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 113                                 }
 114
 115                                 it = parameters.find("accnos");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 121                                 }
 122
 123                                 it = parameters.find("list");
 124                                 //user has given a template file
 125                                 if(it != parameters.end()){
 126                                         path = m->hasPath(it->second);
 127                                         //if the user has not given a path then, add inputdir. else leave path alone.
 128                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 129                                 }
 130
 131                                 it = parameters.find("name");
 132                                 //user has given a template file
 133                                 if(it != parameters.end()){
 134                                         path = m->hasPath(it->second);
 135                                         //if the user has not given a path then, add inputdir. else leave path alone.
 136                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 137                                 }
 138
 139                                 it = parameters.find("group");
 140                                 //user has given a template file
 141                                 if(it != parameters.end()){
 142                                         path = m->hasPath(it->second);
 143                                         //if the user has not given a path then, add inputdir. else leave path alone.
 144                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 145                                 }
 146
 147                                 it = parameters.find("taxonomy");
 148                                 //user has given a template file
 149                                 if(it != parameters.end()){
 150                                         path = m->hasPath(it->second);
 151                                         //if the user has not given a path then, add inputdir. else leave path alone.
 152                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 153                                 }
 154                         }
 155
 156
 157                         //check for required parameters
 158                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 159                         if (accnosfile == "not open") { abort = true; }
 160                         else if (accnosfile == "not found") {  accnosfile = ""; }
 161
 162                         fastafile = validParameter.validFile(parameters, "fasta", true);
 163                         if (fastafile == "not open") { abort = true; }
 164                         else if (fastafile == "not found") {  fastafile = "";  }
 165
 166                         namefile = validParameter.validFile(parameters, "name", true);
 167                         if (namefile == "not open") { abort = true; }
 168                         else if (namefile == "not found") {  namefile = "";  }
 169
 170                         groupfile = validParameter.validFile(parameters, "group", true);
 171                         if (groupfile == "not open") { abort = true; }
 172                         else if (groupfile == "not found") {  groupfile = "";  m->mothurOut("You must provide a group file."); m->mothurOutEndLine(); abort = true; }
 173
 174                         listfile = validParameter.validFile(parameters, "list", true);
 175                         if (listfile == "not open") { abort = true; }
 176                         else if (listfile == "not found") {  listfile = "";  }
 177
 178                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 179                         if (taxfile == "not open") { abort = true; }
 180                         else if (taxfile == "not found") {  taxfile = "";  }
 181
 182                         groups = validParameter.validFile(parameters, "groups", false);
 183                         if (groups == "not found") { groups = ""; }
 184                         else {
 185                                 m->splitAtDash(groups, Groups);
 186                         }
 187
 188                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 189
 190                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group or list."); m->mothurOutEndLine(); abort = true; }
 191                 }
 192
 193         }
 194         catch(exception& e) {
 195                 m->errorOut(e, "RemoveGroupsCommand", "RemoveGroupsCommand");
 196                 exit(1);
 197         }
 198 }
 199 //**********************************************************************************************************************
 200
 201 void RemoveGroupsCommand::help(){
 202         try {
 203                 m->mothurOut("The remove.groups command removes sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy.\n");
 204                 m->mothurOut("It outputs a file containing the sequences NOT in the those specified groups.\n");
 205                 m->mothurOut("The remove.groups command parameters are accnos, fasta, name, group, list, taxonomy and groups. The group parameter is required.\n");
 206                 m->mothurOut("You must also provide an accnos containing the list of groups to remove or set the groups parameter to the groups you wish to remove.\n");
 207                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like removed.  You can separate group names with dashes.\n");
 208                 m->mothurOut("The remove.groups command should be in the following format: remove.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n");
 209                 m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n");
 210                 m->mothurOut("or remove.seqs(groups=pasture, fasta=amazon.fasta, amazon.groups).\n");
 211                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 212         }
 213         catch(exception& e) {
 214                 m->errorOut(e, "RemoveGroupsCommand", "help");
 215                 exit(1);
 216         }
 217 }
 218
 219 //**********************************************************************************************************************
 220
 221 int RemoveGroupsCommand::execute(){
 222         try {
 223
 224                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 225
 226                 groupMap = new GroupMap(groupfile);
 227                 groupMap->readMap();
 228
 229                 //get groups you want to remove
 230                 if (accnosfile != "") { readAccnos(); }
 231
 232                 //make sure groups are valid
 233                 //takes care of user setting groupNames that are invalid or setting groups=all
 234                 SharedUtil* util = new SharedUtil();
 235                 util->setGroups(Groups, groupMap->namesOfGroups);
 236                 delete util;
 237
 238                 //fill names with names of sequences that are from the groups we want to remove
 239                 fillNames();
 240
 241                 if (m->control_pressed) { delete groupMap; return 0; }
 242
 243                 //read through the correct file and output lines you want to keep
 244                 if (namefile != "")                     {               readName();             }
 245                 if (fastafile != "")            {               readFasta();    }
 246                 if (groupfile != "")            {               readGroup();    }
 247                 if (listfile != "")                     {               readList();             }
 248                 if (taxfile != "")                      {               readTax();              }
 249
 250                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 251
 252                 m->mothurOut("Removed " + toString(names.size()) + " sequences. From the groups: "); m->mothurOutEndLine();
 253                 for (int i = 0; i < Groups.size(); i++) {       m->mothurOut(Groups[i]); m->mothurOutEndLine(); }
 254                 m->mothurOutEndLine();
 255
 256                 if (outputNames.size() != 0) {
 257                         m->mothurOutEndLine();
 258                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 259                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 260                         m->mothurOutEndLine();
 261                 }
 262
 263                 return 0;
 264         }
 265
 266         catch(exception& e) {
 267                 m->errorOut(e, "RemoveGroupsCommand", "execute");
 268                 exit(1);
 269         }
 270 }
 271
 272 //**********************************************************************************************************************
 273 int RemoveGroupsCommand::readFasta(){
 274         try {
 275                 string thisOutputDir = outputDir;
 276                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 277                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 278
 279                 ofstream out;
 280                 m->openOutputFile(outputFileName, out);
 281
 282                 ifstream in;
 283                 m->openInputFile(fastafile, in);
 284                 string name;
 285
 286                 bool wroteSomething = false;
 287
 288                 while(!in.eof()){
 289                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 290
 291                         Sequence currSeq(in);
 292                         name = currSeq.getName();
 293
 294                         if (name != "") {
 295                                 //if this name is in the accnos file
 296                                 if (names.count(name) == 0) {
 297                                         wroteSomething = true;
 298
 299                                         currSeq.printSequence(out);
 300                                 }
 301                         }
 302                         m->gobble(in);
 303                 }
 304                 in.close();
 305                 out.close();
 306
 307                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 308                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 309
 310                 return 0;
 311
 312         }
 313         catch(exception& e) {
 314                 m->errorOut(e, "RemoveGroupsCommand", "readFasta");
 315                 exit(1);
 316         }
 317 }
 318
 319 //**********************************************************************************************************************
 320 int RemoveGroupsCommand::readList(){
 321         try {
 322                 string thisOutputDir = outputDir;
 323                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 324                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 325
 326                 ofstream out;
 327                 m->openOutputFile(outputFileName, out);
 328
 329                 ifstream in;
 330                 m->openInputFile(listfile, in);
 331
 332                 bool wroteSomething = false;
 333
 334                 while(!in.eof()){
 335                         //read in list vector
 336                         ListVector list(in);
 337
 338                         //make a new list vector
 339                         ListVector newList;
 340                         newList.setLabel(list.getLabel());
 341
 342                         //for each bin
 343                         for (int i = 0; i < list.getNumBins(); i++) {
 344                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 345
 346                                 //parse out names that are in accnos file
 347                                 string binnames = list.get(i);
 348
 349                                 string newNames = "";
 350                                 while (binnames.find_first_of(',') != -1) {
 351                                         string name = binnames.substr(0,binnames.find_first_of(','));
 352                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 353
 354                                         //if that name is in the .accnos file, add it
 355                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 356                                 }
 357
 358                                 //get last name
 359                                 if (names.count(binnames) == 0) {  newNames += binnames + ",";  }
 360
 361                                 //if there are names in this bin add to new list
 362                                 if (newNames != "") {
 363                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 364                                         newList.push_back(newNames);
 365                                 }
 366                         }
 367
 368                         //print new listvector
 369                         if (newList.getNumBins() != 0) {
 370                                 wroteSomething = true;
 371                                 newList.print(out);
 372                         }
 373
 374                         m->gobble(in);
 375                 }
 376                 in.close();
 377                 out.close();
 378
 379                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 380                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 381
 382                 return 0;
 383
 384         }
 385         catch(exception& e) {
 386                 m->errorOut(e, "RemoveGroupsCommand", "readList");
 387                 exit(1);
 388         }
 389 }
 390 //**********************************************************************************************************************
 391 int RemoveGroupsCommand::readName(){
 392         try {
 393                 string thisOutputDir = outputDir;
 394                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 395                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 396
 397                 ofstream out;
 398                 m->openOutputFile(outputFileName, out);
 399
 400                 ifstream in;
 401                 m->openInputFile(namefile, in);
 402                 string name, firstCol, secondCol;
 403
 404                 bool wroteSomething = false;
 405
 406                 while(!in.eof()){
 407                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 408
 409                         in >> firstCol;         m->gobble(in);
 410                         in >> secondCol;
 411
 412                         vector<string> parsedNames;
 413                         m->splitAtComma(secondCol, parsedNames);
 414
 415                         vector<string> validSecond;  validSecond.clear();
 416                         for (int i = 0; i < parsedNames.size(); i++) {
 417                                 if (names.count(parsedNames[i]) == 0) {
 418                                         validSecond.push_back(parsedNames[i]);
 419                                 }
 420                         }
 421
 422
 423                         //if the name in the first column is in the set then print it and any other names in second column also in set
 424                         if (names.count(firstCol) == 0) {
 425
 426                                 wroteSomething = true;
 427
 428                                 out << firstCol << '\t';
 429
 430                                 //you know you have at least one valid second since first column is valid
 431                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 432                                 out << validSecond[validSecond.size()-1] << endl;
 433
 434                                 //make first name in set you come to first column and then add the remaining names to second column
 435                         }else {
 436
 437                                 //you want part of this row
 438                                 if (validSecond.size() != 0) {
 439
 440                                         wroteSomething = true;
 441
 442                                         out << validSecond[0] << '\t';
 443
 444                                         //you know you have at least one valid second since first column is valid
 445                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 446                                         out << validSecond[validSecond.size()-1] << endl;
 447                                 }
 448                         }
 449
 450                         m->gobble(in);
 451                 }
 452                 in.close();
 453                 out.close();
 454
 455                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 456                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 457
 458                 return 0;
 459         }
 460         catch(exception& e) {
 461                 m->errorOut(e, "RemoveGroupsCommand", "readName");
 462                 exit(1);
 463         }
 464 }
 465
 466 //**********************************************************************************************************************
 467 int RemoveGroupsCommand::readGroup(){
 468         try {
 469                 string thisOutputDir = outputDir;
 470                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 471                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 472
 473                 ofstream out;
 474                 m->openOutputFile(outputFileName, out);
 475
 476                 ifstream in;
 477                 m->openInputFile(groupfile, in);
 478                 string name, group;
 479
 480                 bool wroteSomething = false;
 481
 482                 while(!in.eof()){
 483                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 484
 485                         in >> name;                             //read from first column
 486                         in >> group;                    //read from second column
 487
 488                         //if this name is in the accnos file
 489                         if (names.count(name) == 0) {
 490                                 wroteSomething = true;
 491                                 out << name << '\t' << group << endl;
 492                         }
 493
 494                         m->gobble(in);
 495                 }
 496                 in.close();
 497                 out.close();
 498
 499                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 500                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 501
 502                 return 0;
 503         }
 504         catch(exception& e) {
 505                 m->errorOut(e, "RemoveGroupsCommand", "readGroup");
 506                 exit(1);
 507         }
 508 }
 509 //**********************************************************************************************************************
 510 int RemoveGroupsCommand::readTax(){
 511         try {
 512                 string thisOutputDir = outputDir;
 513                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 514                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 515                 ofstream out;
 516                 m->openOutputFile(outputFileName, out);
 517
 518                 ifstream in;
 519                 m->openInputFile(taxfile, in);
 520                 string name, tax;
 521
 522                 bool wroteSomething = false;
 523
 524                 while(!in.eof()){
 525                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 526
 527                         in >> name;                             //read from first column
 528                         in >> tax;                      //read from second column
 529
 530                         //if this name is in the accnos file
 531                         if (names.count(name) == 0) {
 532                                 wroteSomething = true;
 533                                 out << name << '\t' << tax << endl;
 534                         }
 535
 536                         m->gobble(in);
 537                 }
 538                 in.close();
 539                 out.close();
 540
 541                 if (wroteSomething == false) {  m->mothurOut("Your file contains only sequences from the groups you wish to remove."); m->mothurOutEndLine();  }
 542                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 543
 544                 return 0;
 545         }
 546         catch(exception& e) {
 547                 m->errorOut(e, "RemoveGroupsCommand", "readTax");
 548                 exit(1);
 549         }
 550 }
 551 //**********************************************************************************************************************
 552 void RemoveGroupsCommand::readAccnos(){
 553         try {
 554                 Groups.clear();
 555
 556                 ifstream in;
 557                 m->openInputFile(accnosfile, in);
 558                 string name;
 559
 560                 while(!in.eof()){
 561                         in >> name;
 562
 563                         Groups.push_back(name);
 564
 565                         m->gobble(in);
 566                 }
 567                 in.close();
 568
 569         }
 570         catch(exception& e) {
 571                 m->errorOut(e, "RemoveGroupsCommand", "readAccnos");
 572                 exit(1);
 573         }
 574 }
 575 //**********************************************************************************************************************
 576 int RemoveGroupsCommand::fillNames(){
 577         try {
 578                 vector<string> seqs = groupMap->getNamesSeqs();
 579
 580                 for (int i = 0; i < seqs.size(); i++) {
 581
 582                         if (m->control_pressed) { return 0; }
 583
 584                         string group = groupMap->getGroup(seqs[i]);
 585
 586                         if (m->inUsersGroups(group, Groups)) {
 587                                 names.insert(seqs[i]);
 588                         }
 589                 }
 590
 591                 return 0;
 592         }
 593         catch(exception& e) {
 594                 m->errorOut(e, "RemoveGroupsCommand", "fillNames");
 595                 exit(1);
 596         }
 597 }
 598
 599 //**********************************************************************************************************************
 600
 601
 602