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","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
  96
  97                         //check for required parameters
  98                         accnosfile = validParameter.validFile(parameters, "accnos", true);
  99                         if (accnosfile == "not open") { abort = true; }
 100                         else if (accnosfile == "not found") {  accnosfile = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 101
 102                         fastafile = validParameter.validFile(parameters, "fasta", true);
 103                         if (fastafile == "not open") { abort = true; }
 104                         else if (fastafile == "not found") {  fastafile = "";  }
 105
 106                         namefile = validParameter.validFile(parameters, "name", true);
 107                         if (namefile == "not open") { abort = true; }
 108                         else if (namefile == "not found") {  namefile = "";  }
 109
 110                         groupfile = validParameter.validFile(parameters, "group", true);
 111                         if (groupfile == "not open") { abort = true; }
 112                         else if (groupfile == "not found") {  groupfile = "";  }
 113
 114                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 115                         if (alignfile == "not open") { abort = true; }
 116                         else if (alignfile == "not found") {  alignfile = "";  }
 117
 118                         listfile = validParameter.validFile(parameters, "list", true);
 119                         if (listfile == "not open") { abort = true; }
 120                         else if (listfile == "not found") {  listfile = "";  }
 121
 122                         string usedDups = "true";
 123                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 124                         dups = isTrue(temp);
 125
 126                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport or list."); m->mothurOutEndLine(); abort = true; }
 127
 128                         int okay = 2;
 129                         if (outputDir != "") { okay++; }
 130                         if (usedDups != "") { okay++;  }
 131
 132                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 133
 134                         if (parameters.size() > okay) { m->mothurOut("You may only enter one of the following: fasta, name, group, alignreport, or list."); m->mothurOutEndLine(); abort = true;  }
 135                 }
 136
 137         }
 138         catch(exception& e) {
 139                 m->errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
 140                 exit(1);
 141         }
 142 }
 143 //**********************************************************************************************************************
 144
 145 void RemoveSeqsCommand::help(){
 146         try {
 147                 m->mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
 148                 m->mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
 149                 m->mothurOut("The remove.seqs command parameters are accnos, fasta, name, group, list, alignreport and dups.  You must provide accnos and one of the file parameters.\n");
 150                 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. If dups=true, then remove.seqs outputs a new .accnos file containing all the sequences removed. \n");
 151                 m->mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 152                 m->mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 153                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 154         }
 155         catch(exception& e) {
 156                 m->errorOut(e, "RemoveSeqsCommand", "help");
 157                 exit(1);
 158         }
 159 }
 160
 161 //**********************************************************************************************************************
 162
 163 int RemoveSeqsCommand::execute(){
 164         try {
 165
 166                 if (abort == true) { return 0; }
 167
 168                 //get names you want to keep
 169                 readAccnos();
 170
 171                 //read through the correct file and output lines you want to keep
 172                 if (fastafile != "")            {               readFasta();    }
 173                 else if (namefile != "")        {               readName();             }
 174                 else if (groupfile != "")       {               readGroup();    }
 175                 else if (alignfile != "")       {               readAlign();    }
 176                 else if (listfile != "")        {               readList();             }
 177
 178                 if (outputNames.size() != 0) {
 179                         m->mothurOutEndLine();
 180                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 181                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 182                         m->mothurOutEndLine();
 183                 }
 184
 185                 return 0;
 186         }
 187
 188         catch(exception& e) {
 189                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 190                 exit(1);
 191         }
 192 }
 193
 194 //**********************************************************************************************************************
 195 void RemoveSeqsCommand::readFasta(){
 196         try {
 197                 if (outputDir == "") {  outputDir += hasPath(fastafile);  }
 198                 string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
 199                 ofstream out;
 200                 openOutputFile(outputFileName, out);
 201
 202                 ifstream in;
 203                 openInputFile(fastafile, in);
 204                 string name;
 205
 206                 bool wroteSomething = false;
 207
 208                 while(!in.eof()){
 209                         Sequence currSeq(in);
 210                         name = currSeq.getName();
 211
 212                         if (name != "") {
 213                                 //if this name is in the accnos file
 214                                 if (names.count(name) == 0) {
 215                                         wroteSomething = true;
 216
 217                                         currSeq.printSequence(out);
 218                                 }else {         names.erase(name);              }
 219                         }
 220                         gobble(in);
 221                 }
 222                 in.close();
 223                 out.close();
 224
 225                 if (wroteSomething == false) {
 226                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 227                         remove(outputFileName.c_str());
 228                 }else { outputNames.push_back(outputFileName); }
 229
 230         }
 231         catch(exception& e) {
 232                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 233                 exit(1);
 234         }
 235 }
 236 //**********************************************************************************************************************
 237 void RemoveSeqsCommand::readList(){
 238         try {
 239                 if (outputDir == "") {  outputDir += hasPath(listfile);  }
 240                 string outputFileName = getRootName(listfile) + "pick" +  getExtension(listfile);
 241                 ofstream out;
 242                 openOutputFile(outputFileName, out);
 243
 244                 ifstream in;
 245                 openInputFile(listfile, in);
 246
 247                 bool wroteSomething = false;
 248
 249                 while(!in.eof()){
 250                         //read in list vector
 251                         ListVector list(in);
 252
 253                         //make a new list vector
 254                         ListVector newList;
 255                         newList.setLabel(list.getLabel());
 256
 257                         //for each bin
 258                         for (int i = 0; i < list.getNumBins(); i++) {
 259
 260                                 //parse out names that are in accnos file
 261                                 string binnames = list.get(i);
 262
 263                                 string newNames = "";
 264                                 while (binnames.find_first_of(',') != -1) {
 265                                         string name = binnames.substr(0,binnames.find_first_of(','));
 266                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 267
 268                                         //if that name is in the .accnos file, add it
 269                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 270                                 }
 271
 272                                 //get last name
 273                                 if (names.count(binnames) == 0) {  newNames += binnames;  }
 274
 275                                 //if there are names in this bin add to new list
 276                                 if (newNames != "") {  newList.push_back(newNames);     }
 277                         }
 278
 279                         //print new listvector
 280                         if (newList.getNumBins() != 0) {
 281                                 wroteSomething = true;
 282                                 newList.print(out);
 283                         }
 284
 285                         gobble(in);
 286                 }
 287                 in.close();
 288                 out.close();
 289
 290                 if (wroteSomething == false) {
 291                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 292                         remove(outputFileName.c_str());
 293                 }else { outputNames.push_back(outputFileName); }
 294
 295         }
 296         catch(exception& e) {
 297                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 298                 exit(1);
 299         }
 300 }
 301 //**********************************************************************************************************************
 302 void RemoveSeqsCommand::readName(){
 303         try {
 304                 if (outputDir == "") {  outputDir += hasPath(namefile);  }
 305                 string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
 306                 string outputFileName2 = getRootName(namefile) + "dups.accnos";
 307
 308                 ofstream out2;
 309                 if (dups) {      openOutputFile(outputFileName2, out2); }
 310                 bool wroteDups = false;
 311
 312                 ofstream out;
 313                 openOutputFile(outputFileName, out);
 314
 315                 ifstream in;
 316                 openInputFile(namefile, in);
 317                 string name, firstCol, secondCol;
 318
 319                 bool wroteSomething = false;
 320
 321                 while(!in.eof()){
 322
 323                         in >> firstCol;
 324                         in >> secondCol;
 325
 326                         vector<string> parsedNames;
 327                         //parse second column saving each name
 328                         while (secondCol.find_first_of(',') != -1) {
 329                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 330                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 331                                 parsedNames.push_back(name);
 332
 333                         }
 334
 335                         //get name after last ,
 336                         parsedNames.push_back(secondCol);
 337
 338                         vector<string> validSecond;  validSecond.clear();
 339                         for (int i = 0; i < parsedNames.size(); i++) {
 340                                 if (names.count(parsedNames[i]) == 0) {
 341                                         validSecond.push_back(parsedNames[i]);
 342                                 }
 343                         }
 344
 345                         if ((dups) && (validSecond.size() != parsedNames.size())) {
 346                                 wroteDups = true;
 347                                 for (int i = 0; i < parsedNames.size(); i++) {  out2 << parsedNames[i] << endl;  }
 348                         }else {
 349                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 350                                 if (names.count(firstCol) == 0) {
 351
 352                                         wroteSomething = true;
 353
 354                                         out << firstCol << '\t';
 355
 356                                         //you know you have at least one valid second since first column is valid
 357                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 358                                         out << validSecond[validSecond.size()-1] << endl;
 359
 360                                         //make first name in set you come to first column and then add the remaining names to second column
 361                                 }else {
 362
 363                                         //you want part of this row
 364                                         if (validSecond.size() != 0) {
 365
 366                                                 wroteSomething = true;
 367
 368                                                 out << validSecond[0] << '\t';
 369
 370                                                 //you know you have at least one valid second since first column is valid
 371                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 372                                                 out << validSecond[validSecond.size()-1] << endl;
 373                                         }
 374                                 }
 375                         }
 376                         gobble(in);
 377                 }
 378                 in.close();
 379                 out.close();
 380
 381                 if (dups) { out2.close();  }
 382                 if (wroteDups == false) {
 383                         remove(outputFileName2.c_str());
 384                 }else { outputNames.push_back(outputFileName2); }
 385
 386                 if (wroteSomething == false) {
 387                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 388                         remove(outputFileName.c_str());
 389                 }else { outputNames.push_back(outputFileName); }
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 394                 exit(1);
 395         }
 396 }
 397
 398 //**********************************************************************************************************************
 399 void RemoveSeqsCommand::readGroup(){
 400         try {
 401                 if (outputDir == "") {  outputDir += hasPath(groupfile);  }
 402                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 403                 ofstream out;
 404                 openOutputFile(outputFileName, out);
 405
 406                 ifstream in;
 407                 openInputFile(groupfile, in);
 408                 string name, group;
 409
 410                 bool wroteSomething = false;
 411
 412                 while(!in.eof()){
 413
 414                         in >> name;                             //read from first column
 415                         in >> group;                    //read from second column
 416
 417                         //if this name is in the accnos file
 418                         if (names.count(name) == 0) {
 419                                 wroteSomething = true;
 420                                 out << name << '\t' << group << endl;
 421                         }else {         names.erase(name);              }
 422
 423                         gobble(in);
 424                 }
 425                 in.close();
 426                 out.close();
 427
 428                 if (wroteSomething == false) {
 429                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 430                         remove(outputFileName.c_str());
 431                 }else { outputNames.push_back(outputFileName); }
 432
 433         }
 434         catch(exception& e) {
 435                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 436                 exit(1);
 437         }
 438 }
 439
 440 //**********************************************************************************************************************
 441 //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
 442 void RemoveSeqsCommand::readAlign(){
 443         try {
 444                 if (outputDir == "") {  outputDir += hasPath(alignfile);  }
 445                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 446                 ofstream out;
 447                 openOutputFile(outputFileName, out);
 448
 449                 ifstream in;
 450                 openInputFile(alignfile, in);
 451                 string name, junk;
 452
 453                 bool wroteSomething = false;
 454
 455                 //read column headers
 456                 for (int i = 0; i < 16; i++) {
 457                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 458                         else                    {       break;                  }
 459                 }
 460                 out << endl;
 461
 462                 while(!in.eof()){
 463
 464                         in >> name;                             //read from first column
 465
 466                         //if this name is in the accnos file
 467                         if (names.count(name) == 0) {
 468                                 wroteSomething = true;
 469
 470                                 out << name << '\t';
 471
 472                                 //read rest
 473                                 for (int i = 0; i < 15; i++) {
 474                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 475                                         else                    {       break;                  }
 476                                 }
 477                                 out << endl;
 478
 479                         }else {//still read just don't do anything with it
 480                                 names.erase(name);
 481
 482                                 //read rest
 483                                 for (int i = 0; i < 15; i++) {
 484                                         if (!in.eof())  {       in >> junk;             }
 485                                         else                    {       break;                  }
 486                                 }
 487                         }
 488
 489                         gobble(in);
 490                 }
 491                 in.close();
 492                 out.close();
 493
 494                 if (wroteSomething == false) {
 495                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 496                         remove(outputFileName.c_str());
 497                 }else { outputNames.push_back(outputFileName); }
 498
 499         }
 500         catch(exception& e) {
 501                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 502                 exit(1);
 503         }
 504 }
 505 //**********************************************************************************************************************
 506 void RemoveSeqsCommand::readAccnos(){
 507         try {
 508
 509                 ifstream in;
 510                 openInputFile(accnosfile, in);
 511                 string name;
 512
 513                 while(!in.eof()){
 514                         in >> name;
 515
 516                         names.insert(name);
 517
 518                         gobble(in);
 519                 }
 520                 in.close();
 521
 522         }
 523         catch(exception& e) {
 524                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 525                 exit(1);
 526         }
 527 }
 528
 529 //**********************************************************************************************************************
 530
 531