getgroupscommand.cpp

   1 /*
   2  *  getgroupscommand.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 "getgroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14
  15 //**********************************************************************************************************************
  16 vector<string> GetGroupsCommand::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, "GetGroupsCommand", "getValidParameters");
  24                 exit(1);
  25         }
  26 }
  27 //**********************************************************************************************************************
  28 GetGroupsCommand::GetGroupsCommand(){
  29         try {
  30                 abort = true;
  31                 //initialize outputTypes
  32                 vector<string> tempOutNames;
  33                 outputTypes["fasta"] = tempOutNames;
  34                 outputTypes["taxonomy"] = tempOutNames;
  35                 outputTypes["name"] = tempOutNames;
  36                 outputTypes["group"] = tempOutNames;
  37                 outputTypes["list"] = tempOutNames;
  38         }
  39         catch(exception& e) {
  40                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
  41                 exit(1);
  42         }
  43 }
  44 //**********************************************************************************************************************
  45 vector<string> GetGroupsCommand::getRequiredParameters(){
  46         try {
  47                 string Array[] =  {"group"};
  48                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  49                 return myArray;
  50         }
  51         catch(exception& e) {
  52                 m->errorOut(e, "GetGroupsCommand", "getRequiredParameters");
  53                 exit(1);
  54         }
  55 }
  56 //**********************************************************************************************************************
  57 vector<string> GetGroupsCommand::getRequiredFiles(){
  58         try {
  59                 vector<string> myArray;
  60                 return myArray;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "GetGroupsCommand", "getRequiredFiles");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 GetGroupsCommand::GetGroupsCommand(string option)  {
  69         try {
  70                 abort = false;
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; }
  74
  75                 else {
  76                         //valid paramters for this command
  77                         string Array[] =  {"fasta","name", "group", "accnos", "groups", "list","taxonomy","outputdir","inputdir" };
  78                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  79
  80                         OptionParser parser(option);
  81                         map<string,string> parameters = parser.getParameters();
  82
  83                         ValidParameters validParameter;
  84                         map<string,string>::iterator it;
  85
  86                         //check to make sure all parameters are valid for command
  87                         for (it = parameters.begin(); it != parameters.end(); it++) {
  88                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  89                         }
  90
  91                         //initialize outputTypes
  92                         vector<string> tempOutNames;
  93                         outputTypes["fasta"] = tempOutNames;
  94                         outputTypes["taxonomy"] = tempOutNames;
  95                         outputTypes["name"] = tempOutNames;
  96                         outputTypes["group"] = tempOutNames;
  97                         outputTypes["list"] = tempOutNames;
  98
  99
 100                         //if the user changes the output directory command factory will send this info to us in the output parameter
 101                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 102
 103                         //if the user changes the input directory command factory will send this info to us in the output parameter
 104                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 105                         if (inputDir == "not found"){   inputDir = "";          }
 106                         else {
 107                                 string path;
 108                                 it = parameters.find("fasta");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 114                                 }
 115
 116                                 it = parameters.find("accnos");
 117                                 //user has given a template file
 118                                 if(it != parameters.end()){
 119                                         path = m->hasPath(it->second);
 120                                         //if the user has not given a path then, add inputdir. else leave path alone.
 121                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 122                                 }
 123
 124                                 it = parameters.find("list");
 125                                 //user has given a template file
 126                                 if(it != parameters.end()){
 127                                         path = m->hasPath(it->second);
 128                                         //if the user has not given a path then, add inputdir. else leave path alone.
 129                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 130                                 }
 131
 132                                 it = parameters.find("name");
 133                                 //user has given a template file
 134                                 if(it != parameters.end()){
 135                                         path = m->hasPath(it->second);
 136                                         //if the user has not given a path then, add inputdir. else leave path alone.
 137                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 138                                 }
 139
 140                                 it = parameters.find("group");
 141                                 //user has given a template file
 142                                 if(it != parameters.end()){
 143                                         path = m->hasPath(it->second);
 144                                         //if the user has not given a path then, add inputdir. else leave path alone.
 145                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 146                                 }
 147
 148                                 it = parameters.find("taxonomy");
 149                                 //user has given a template file
 150                                 if(it != parameters.end()){
 151                                         path = m->hasPath(it->second);
 152                                         //if the user has not given a path then, add inputdir. else leave path alone.
 153                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 154                                 }
 155                         }
 156
 157
 158                         //check for required parameters
 159                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 160                         if (accnosfile == "not open") { abort = true; }
 161                         else if (accnosfile == "not found") {  accnosfile = ""; }
 162
 163                         fastafile = validParameter.validFile(parameters, "fasta", true);
 164                         if (fastafile == "not open") { abort = true; }
 165                         else if (fastafile == "not found") {  fastafile = "";  }
 166
 167                         namefile = validParameter.validFile(parameters, "name", true);
 168                         if (namefile == "not open") { abort = true; }
 169                         else if (namefile == "not found") {  namefile = "";  }
 170
 171                         groupfile = validParameter.validFile(parameters, "group", true);
 172                         if (groupfile == "not open") { abort = true; }
 173                         else if (groupfile == "not found") {  groupfile = "";  m->mothurOut("You must provide a group file."); m->mothurOutEndLine(); abort = true; }
 174
 175                         listfile = validParameter.validFile(parameters, "list", true);
 176                         if (listfile == "not open") { abort = true; }
 177                         else if (listfile == "not found") {  listfile = "";  }
 178
 179                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 180                         if (taxfile == "not open") { abort = true; }
 181                         else if (taxfile == "not found") {  taxfile = "";  }
 182
 183                         groups = validParameter.validFile(parameters, "groups", false);
 184                         if (groups == "not found") { groups = ""; }
 185                         else {
 186                                 m->splitAtDash(groups, Groups);
 187                         }
 188
 189                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 190
 191                         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; }
 192                 }
 193
 194         }
 195         catch(exception& e) {
 196                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
 197                 exit(1);
 198         }
 199 }
 200 //**********************************************************************************************************************
 201
 202 void GetGroupsCommand::help(){
 203         try {
 204                 m->mothurOut("The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy.\n");
 205                 m->mothurOut("It outputs a file containing the sequences in the those specified groups.\n");
 206                 m->mothurOut("The get.groups command parameters are accnos, fasta, name, group, list, taxonomy and groups. The group parameter is required.\n");
 207                 m->mothurOut("You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n");
 208                 m->mothurOut("The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n");
 209                 m->mothurOut("The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n");
 210                 m->mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n");
 211                 m->mothurOut("or get.seqs(groups=pasture, fasta=amazon.fasta, amazon.groups).\n");
 212                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 213         }
 214         catch(exception& e) {
 215                 m->errorOut(e, "GetGroupsCommand", "help");
 216                 exit(1);
 217         }
 218 }
 219
 220 //**********************************************************************************************************************
 221
 222 int GetGroupsCommand::execute(){
 223         try {
 224
 225                 if (abort == true) { return 0; }
 226
 227                 groupMap = new GroupMap(groupfile);
 228                 groupMap->readMap();
 229
 230                 //get groups you want to remove
 231                 if (accnosfile != "") { readAccnos(); }
 232
 233                 //make sure groups are valid
 234                 //takes care of user setting groupNames that are invalid or setting groups=all
 235                 SharedUtil* util = new SharedUtil();
 236                 util->setGroups(Groups, groupMap->namesOfGroups);
 237                 delete util;
 238
 239                 //fill names with names of sequences that are from the groups we want to remove
 240                 fillNames();
 241
 242                 if (m->control_pressed) { delete groupMap; return 0; }
 243
 244                 //read through the correct file and output lines you want to keep
 245                 if (namefile != "")                     {               readName();             }
 246                 if (fastafile != "")            {               readFasta();    }
 247                 if (groupfile != "")            {               readGroup();    }
 248                 if (listfile != "")                     {               readList();             }
 249                 if (taxfile != "")                      {               readTax();              }
 250
 251                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 252
 253                 m->mothurOut("Selected " + toString(names.size()) + " sequences. From the groups: "); m->mothurOutEndLine();
 254                 for (int i = 0; i < Groups.size(); i++) {       m->mothurOut(Groups[i]); m->mothurOutEndLine(); }
 255                 m->mothurOutEndLine();
 256
 257                 if (outputNames.size() != 0) {
 258                         m->mothurOutEndLine();
 259                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 260                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 261                         m->mothurOutEndLine();
 262                 }
 263
 264                 return 0;
 265         }
 266
 267         catch(exception& e) {
 268                 m->errorOut(e, "GetGroupsCommand", "execute");
 269                 exit(1);
 270         }
 271 }
 272
 273 //**********************************************************************************************************************
 274 int GetGroupsCommand::readFasta(){
 275         try {
 276                 string thisOutputDir = outputDir;
 277                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 278                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 279
 280                 ofstream out;
 281                 m->openOutputFile(outputFileName, out);
 282
 283                 ifstream in;
 284                 m->openInputFile(fastafile, in);
 285                 string name;
 286
 287                 bool wroteSomething = false;
 288
 289                 while(!in.eof()){
 290                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 291
 292                         Sequence currSeq(in);
 293                         name = currSeq.getName();
 294
 295                         if (name != "") {
 296                                 //if this name is in the accnos file
 297                                 if (names.count(name) != 0) {
 298                                         wroteSomething = true;
 299
 300                                         currSeq.printSequence(out);
 301                                 }
 302                         }
 303                         m->gobble(in);
 304                 }
 305                 in.close();
 306                 out.close();
 307
 308                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 309                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 310
 311                 return 0;
 312
 313         }
 314         catch(exception& e) {
 315                 m->errorOut(e, "GetGroupsCommand", "readFasta");
 316                 exit(1);
 317         }
 318 }
 319
 320 //**********************************************************************************************************************
 321 int GetGroupsCommand::readList(){
 322         try {
 323                 string thisOutputDir = outputDir;
 324                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 325                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 326
 327                 ofstream out;
 328                 m->openOutputFile(outputFileName, out);
 329
 330                 ifstream in;
 331                 m->openInputFile(listfile, in);
 332
 333                 bool wroteSomething = false;
 334
 335                 while(!in.eof()){
 336                         //read in list vector
 337                         ListVector list(in);
 338
 339                         //make a new list vector
 340                         ListVector newList;
 341                         newList.setLabel(list.getLabel());
 342
 343                         //for each bin
 344                         for (int i = 0; i < list.getNumBins(); i++) {
 345                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 346
 347                                 //parse out names that are in accnos file
 348                                 string binnames = list.get(i);
 349
 350                                 string newNames = "";
 351                                 while (binnames.find_first_of(',') != -1) {
 352                                         string name = binnames.substr(0,binnames.find_first_of(','));
 353                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 354
 355                                         //if that name is in the .accnos file, add it
 356                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 357                                 }
 358
 359                                 //get last name
 360                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 361
 362                                 //if there are names in this bin add to new list
 363                                 if (newNames != "") {
 364                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 365                                         newList.push_back(newNames);
 366                                 }
 367                         }
 368
 369                         //print new listvector
 370                         if (newList.getNumBins() != 0) {
 371                                 wroteSomething = true;
 372                                 newList.print(out);
 373                         }
 374
 375                         m->gobble(in);
 376                 }
 377                 in.close();
 378                 out.close();
 379
 380                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 381                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 382
 383                 return 0;
 384
 385         }
 386         catch(exception& e) {
 387                 m->errorOut(e, "GetGroupsCommand", "readList");
 388                 exit(1);
 389         }
 390 }
 391 //**********************************************************************************************************************
 392 int GetGroupsCommand::readName(){
 393         try {
 394                 string thisOutputDir = outputDir;
 395                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 396                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 397
 398                 ofstream out;
 399                 m->openOutputFile(outputFileName, out);
 400
 401                 ifstream in;
 402                 m->openInputFile(namefile, in);
 403                 string name, firstCol, secondCol;
 404
 405                 bool wroteSomething = false;
 406
 407                 while(!in.eof()){
 408                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 409
 410                         in >> firstCol;         m->gobble(in);
 411                         in >> secondCol;
 412
 413                         vector<string> parsedNames;
 414                         m->splitAtComma(secondCol, parsedNames);
 415
 416                         vector<string> validSecond;  validSecond.clear();
 417                         for (int i = 0; i < parsedNames.size(); i++) {
 418                                 if (names.count(parsedNames[i]) != 0) {
 419                                         validSecond.push_back(parsedNames[i]);
 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 does NOT contain sequences from the groups you wish to get."); 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, "GetGroupsCommand", "readName");
 462                 exit(1);
 463         }
 464 }
 465
 466 //**********************************************************************************************************************
 467 int GetGroupsCommand::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 does NOT contain sequences from the groups you wish to get."); 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, "GetGroupsCommand", "readGroup");
 506                 exit(1);
 507         }
 508 }
 509 //**********************************************************************************************************************
 510 int GetGroupsCommand::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 does NOT contain sequences from the groups you wish to get."); 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, "GetGroupsCommand", "readTax");
 548                 exit(1);
 549         }
 550 }
 551 //**********************************************************************************************************************
 552 void GetGroupsCommand::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, "GetGroupsCommand", "readAccnos");
 572                 exit(1);
 573         }
 574 }
 575 //**********************************************************************************************************************
 576 int GetGroupsCommand::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, "GetGroupsCommand", "fillNames");
 595                 exit(1);
 596         }
 597 }
 598
 599 //**********************************************************************************************************************
 600
 601