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                 if (m->control_pressed) { return 0; }
 172
 173                 //read through the correct file and output lines you want to keep
 174                 if (fastafile != "")            {               readFasta();    }
 175                 else if (namefile != "")        {               readName();             }
 176                 else if (groupfile != "")       {               readGroup();    }
 177                 else if (alignfile != "")       {               readAlign();    }
 178                 else if (listfile != "")        {               readList();             }
 179
 180                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str()); } return 0; }
 181
 182                 if (outputNames.size() != 0) {
 183                         m->mothurOutEndLine();
 184                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 185                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 186                         m->mothurOutEndLine();
 187                 }
 188
 189                 return 0;
 190         }
 191
 192         catch(exception& e) {
 193                 m->errorOut(e, "RemoveSeqsCommand", "execute");
 194                 exit(1);
 195         }
 196 }
 197
 198 //**********************************************************************************************************************
 199 int RemoveSeqsCommand::readFasta(){
 200         try {
 201                 if (outputDir == "") {  outputDir += hasPath(fastafile);  }
 202                 string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
 203                 ofstream out;
 204                 openOutputFile(outputFileName, out);
 205
 206                 ifstream in;
 207                 openInputFile(fastafile, in);
 208                 string name;
 209
 210                 bool wroteSomething = false;
 211
 212                 while(!in.eof()){
 213                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 214
 215                         Sequence currSeq(in);
 216                         name = currSeq.getName();
 217
 218                         if (name != "") {
 219                                 //if this name is in the accnos file
 220                                 if (names.count(name) == 0) {
 221                                         wroteSomething = true;
 222
 223                                         currSeq.printSequence(out);
 224                                 }else {         names.erase(name);              }
 225                         }
 226                         gobble(in);
 227                 }
 228                 in.close();
 229                 out.close();
 230
 231                 if (wroteSomething == false) {
 232                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 233                         remove(outputFileName.c_str());
 234                 }else { outputNames.push_back(outputFileName); }
 235
 236                 return 0;
 237
 238         }
 239         catch(exception& e) {
 240                 m->errorOut(e, "RemoveSeqsCommand", "readFasta");
 241                 exit(1);
 242         }
 243 }
 244 //**********************************************************************************************************************
 245 int RemoveSeqsCommand::readList(){
 246         try {
 247                 if (outputDir == "") {  outputDir += hasPath(listfile);  }
 248                 string outputFileName = getRootName(listfile) + "pick" +  getExtension(listfile);
 249                 ofstream out;
 250                 openOutputFile(outputFileName, out);
 251
 252                 ifstream in;
 253                 openInputFile(listfile, in);
 254
 255                 bool wroteSomething = false;
 256
 257                 while(!in.eof()){
 258                         //read in list vector
 259                         ListVector list(in);
 260
 261                         //make a new list vector
 262                         ListVector newList;
 263                         newList.setLabel(list.getLabel());
 264
 265                         //for each bin
 266                         for (int i = 0; i < list.getNumBins(); i++) {
 267                                 if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 268
 269                                 //parse out names that are in accnos file
 270                                 string binnames = list.get(i);
 271
 272                                 string newNames = "";
 273                                 while (binnames.find_first_of(',') != -1) {
 274                                         string name = binnames.substr(0,binnames.find_first_of(','));
 275                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 276
 277                                         //if that name is in the .accnos file, add it
 278                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 279                                 }
 280
 281                                 //get last name
 282                                 if (names.count(binnames) == 0) {  newNames += binnames;  }
 283
 284                                 //if there are names in this bin add to new list
 285                                 if (newNames != "") {  newList.push_back(newNames);     }
 286                         }
 287
 288                         //print new listvector
 289                         if (newList.getNumBins() != 0) {
 290                                 wroteSomething = true;
 291                                 newList.print(out);
 292                         }
 293
 294                         gobble(in);
 295                 }
 296                 in.close();
 297                 out.close();
 298
 299                 if (wroteSomething == false) {
 300                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 301                         remove(outputFileName.c_str());
 302                 }else { outputNames.push_back(outputFileName); }
 303
 304                 return 0;
 305
 306         }
 307         catch(exception& e) {
 308                 m->errorOut(e, "RemoveSeqsCommand", "readList");
 309                 exit(1);
 310         }
 311 }
 312 //**********************************************************************************************************************
 313 int RemoveSeqsCommand::readName(){
 314         try {
 315                 if (outputDir == "") {  outputDir += hasPath(namefile);  }
 316                 string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
 317                 string outputFileName2 = getRootName(namefile) + "dups.accnos";
 318
 319                 ofstream out2;
 320                 if (dups) {      openOutputFile(outputFileName2, out2); }
 321                 bool wroteDups = false;
 322
 323                 ofstream out;
 324                 openOutputFile(outputFileName, out);
 325
 326                 ifstream in;
 327                 openInputFile(namefile, in);
 328                 string name, firstCol, secondCol;
 329
 330                 bool wroteSomething = false;
 331
 332                 while(!in.eof()){
 333                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str()); if (dups) { out2.close(); remove(outputFileName2.c_str()); } return 0; }
 334
 335                         in >> firstCol;
 336                         in >> secondCol;
 337
 338                         vector<string> parsedNames;
 339                         //parse second column saving each name
 340                         while (secondCol.find_first_of(',') != -1) {
 341                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 342                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 343                                 parsedNames.push_back(name);
 344
 345                         }
 346
 347                         //get name after last ,
 348                         parsedNames.push_back(secondCol);
 349
 350                         vector<string> validSecond;  validSecond.clear();
 351                         for (int i = 0; i < parsedNames.size(); i++) {
 352                                 if (names.count(parsedNames[i]) == 0) {
 353                                         validSecond.push_back(parsedNames[i]);
 354                                 }
 355                         }
 356
 357                         if ((dups) && (validSecond.size() != parsedNames.size())) {
 358                                 wroteDups = true;
 359                                 for (int i = 0; i < parsedNames.size(); i++) {  out2 << parsedNames[i] << endl;  }
 360                         }else {
 361                                         //if the name in the first column is in the set then print it and any other names in second column also in set
 362                                 if (names.count(firstCol) == 0) {
 363
 364                                         wroteSomething = true;
 365
 366                                         out << firstCol << '\t';
 367
 368                                         //you know you have at least one valid second since first column is valid
 369                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 370                                         out << validSecond[validSecond.size()-1] << endl;
 371
 372                                         //make first name in set you come to first column and then add the remaining names to second column
 373                                 }else {
 374
 375                                         //you want part of this row
 376                                         if (validSecond.size() != 0) {
 377
 378                                                 wroteSomething = true;
 379
 380                                                 out << validSecond[0] << '\t';
 381
 382                                                 //you know you have at least one valid second since first column is valid
 383                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 384                                                 out << validSecond[validSecond.size()-1] << endl;
 385                                         }
 386                                 }
 387                         }
 388                         gobble(in);
 389                 }
 390                 in.close();
 391                 out.close();
 392
 393                 if (dups) { out2.close();  }
 394                 if (wroteDups == false) {
 395                         remove(outputFileName2.c_str());
 396                 }else { outputNames.push_back(outputFileName2); }
 397
 398                 if (wroteSomething == false) {
 399                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 400                         remove(outputFileName.c_str());
 401                 }else { outputNames.push_back(outputFileName); }
 402
 403                 return 0;
 404         }
 405         catch(exception& e) {
 406                 m->errorOut(e, "RemoveSeqsCommand", "readName");
 407                 exit(1);
 408         }
 409 }
 410
 411 //**********************************************************************************************************************
 412 int RemoveSeqsCommand::readGroup(){
 413         try {
 414                 if (outputDir == "") {  outputDir += hasPath(groupfile);  }
 415                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 416                 ofstream out;
 417                 openOutputFile(outputFileName, out);
 418
 419                 ifstream in;
 420                 openInputFile(groupfile, in);
 421                 string name, group;
 422
 423                 bool wroteSomething = false;
 424
 425                 while(!in.eof()){
 426                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 427
 428                         in >> name;                             //read from first column
 429                         in >> group;                    //read from second column
 430
 431                         //if this name is in the accnos file
 432                         if (names.count(name) == 0) {
 433                                 wroteSomething = true;
 434                                 out << name << '\t' << group << endl;
 435                         }else {         names.erase(name);              }
 436
 437                         gobble(in);
 438                 }
 439                 in.close();
 440                 out.close();
 441
 442                 if (wroteSomething == false) {
 443                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 444                         remove(outputFileName.c_str());
 445                 }else { outputNames.push_back(outputFileName); }
 446
 447                 return 0;
 448         }
 449         catch(exception& e) {
 450                 m->errorOut(e, "RemoveSeqsCommand", "readGroup");
 451                 exit(1);
 452         }
 453 }
 454
 455 //**********************************************************************************************************************
 456 //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
 457 int RemoveSeqsCommand::readAlign(){
 458         try {
 459                 if (outputDir == "") {  outputDir += hasPath(alignfile);  }
 460                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 461                 ofstream out;
 462                 openOutputFile(outputFileName, out);
 463
 464                 ifstream in;
 465                 openInputFile(alignfile, in);
 466                 string name, junk;
 467
 468                 bool wroteSomething = false;
 469
 470                 //read column headers
 471                 for (int i = 0; i < 16; i++) {
 472                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 473                         else                    {       break;                  }
 474                 }
 475                 out << endl;
 476
 477                 while(!in.eof()){
 478                         if (m->control_pressed) { in.close();  out.close();  remove(outputFileName.c_str());  return 0; }
 479
 480                         in >> name;                             //read from first column
 481
 482                         //if this name is in the accnos file
 483                         if (names.count(name) == 0) {
 484                                 wroteSomething = true;
 485
 486                                 out << name << '\t';
 487
 488                                 //read rest
 489                                 for (int i = 0; i < 15; i++) {
 490                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 491                                         else                    {       break;                  }
 492                                 }
 493                                 out << endl;
 494
 495                         }else {//still read just don't do anything with it
 496                                 names.erase(name);
 497
 498                                 //read rest
 499                                 for (int i = 0; i < 15; i++) {
 500                                         if (!in.eof())  {       in >> junk;             }
 501                                         else                    {       break;                  }
 502                                 }
 503                         }
 504
 505                         gobble(in);
 506                 }
 507                 in.close();
 508                 out.close();
 509
 510                 if (wroteSomething == false) {
 511                         m->mothurOut("Your file contains only sequences from the .accnos file."); m->mothurOutEndLine();
 512                         remove(outputFileName.c_str());
 513                 }else { outputNames.push_back(outputFileName); }
 514
 515                 return 0;
 516
 517         }
 518         catch(exception& e) {
 519                 m->errorOut(e, "RemoveSeqsCommand", "readAlign");
 520                 exit(1);
 521         }
 522 }
 523 //**********************************************************************************************************************
 524 void RemoveSeqsCommand::readAccnos(){
 525         try {
 526
 527                 ifstream in;
 528                 openInputFile(accnosfile, in);
 529                 string name;
 530
 531                 while(!in.eof()){
 532                         in >> name;
 533
 534                         names.insert(name);
 535
 536                         gobble(in);
 537                 }
 538                 in.close();
 539
 540         }
 541         catch(exception& e) {
 542                 m->errorOut(e, "RemoveSeqsCommand", "readAccnos");
 543                 exit(1);
 544         }
 545 }
 546
 547 //**********************************************************************************************************************
 548
 549