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