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