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" };
  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
  33                         //check to make sure all parameters are valid for command
  34                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //check for required parameters
  39                         accnosfile = validParameter.validFile(parameters, "accnos", true);
  40                         if (accnosfile == "not open") { abort = true; }
  41                         else if (accnosfile == "not found") {  accnosfile = "";  mothurOut("You must provide an accnos file."); mothurOutEndLine(); abort = true; }
  42
  43                         fastafile = validParameter.validFile(parameters, "fasta", true);
  44                         if (fastafile == "not open") { abort = true; }
  45                         else if (fastafile == "not found") {  fastafile = "";  }
  46
  47                         namefile = validParameter.validFile(parameters, "name", true);
  48                         if (namefile == "not open") { abort = true; }
  49                         else if (namefile == "not found") {  namefile = "";  }
  50
  51                         groupfile = validParameter.validFile(parameters, "group", true);
  52                         if (groupfile == "not open") { abort = true; }
  53                         else if (groupfile == "not found") {  groupfile = "";  }
  54
  55                         alignfile = validParameter.validFile(parameters, "alignreport", true);
  56                         if (alignfile == "not open") { abort = true; }
  57                         else if (alignfile == "not found") {  alignfile = "";  }
  58
  59                         listfile = validParameter.validFile(parameters, "list", true);
  60                         if (listfile == "not open") { abort = true; }
  61                         else if (listfile == "not found") {  listfile = "";  }
  62
  63                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == ""))  { mothurOut("You must provide one of the following: fasta, name, group, alignreport or list."); mothurOutEndLine(); abort = true; }
  64
  65                         if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport, or list."); mothurOutEndLine(); abort = true;  }
  66                 }
  67
  68         }
  69         catch(exception& e) {
  70                 errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75
  76 void RemoveSeqsCommand::help(){
  77         try {
  78                 mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
  79                 mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
  80                 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");
  81                 mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
  82                 mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
  83                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  84         }
  85         catch(exception& e) {
  86                 errorOut(e, "RemoveSeqsCommand", "help");
  87                 exit(1);
  88         }
  89 }
  90
  91 //**********************************************************************************************************************
  92
  93 int RemoveSeqsCommand::execute(){
  94         try {
  95
  96                 if (abort == true) { return 0; }
  97
  98                 //get names you want to keep
  99                 readAccnos();
 100
 101                 //read through the correct file and output lines you want to keep
 102                 if (fastafile != "")            {               readFasta();    }
 103                 else if (namefile != "")        {               readName();             }
 104                 else if (groupfile != "")       {               readGroup();    }
 105                 else if (alignfile != "")       {               readAlign();    }
 106                 else if (listfile != "")        {               readList();             }
 107
 108                 return 0;
 109         }
 110
 111         catch(exception& e) {
 112                 errorOut(e, "RemoveSeqsCommand", "execute");
 113                 exit(1);
 114         }
 115 }
 116
 117 //**********************************************************************************************************************
 118 void RemoveSeqsCommand::readFasta(){
 119         try {
 120                 string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
 121                 ofstream out;
 122                 openOutputFile(outputFileName, out);
 123
 124                 ifstream in;
 125                 openInputFile(fastafile, in);
 126                 string name;
 127
 128                 bool wroteSomething = false;
 129
 130                 while(!in.eof()){
 131                         Sequence currSeq(in);
 132                         name = currSeq.getName();
 133
 134                         if (name != "") {
 135                                 //if this name is in the accnos file
 136                                 if (names.count(name) == 0) {
 137                                         wroteSomething = true;
 138
 139                                         currSeq.printSequence(out);
 140                                 }else {         names.erase(name);              }
 141                         }
 142                         gobble(in);
 143                 }
 144                 in.close();
 145                 out.close();
 146
 147                 if (wroteSomething == false) {
 148                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 149                         remove(outputFileName.c_str());
 150                 }
 151
 152         }
 153         catch(exception& e) {
 154                 errorOut(e, "RemoveSeqsCommand", "readFasta");
 155                 exit(1);
 156         }
 157 }
 158 //**********************************************************************************************************************
 159 void RemoveSeqsCommand::readList(){
 160         try {
 161                 string outputFileName = getRootName(listfile) + "pick" +  getExtension(listfile);
 162                 ofstream out;
 163                 openOutputFile(outputFileName, out);
 164
 165                 ifstream in;
 166                 openInputFile(listfile, in);
 167
 168                 bool wroteSomething = false;
 169
 170                 while(!in.eof()){
 171                         //read in list vector
 172                         ListVector list(in);
 173
 174                         //make a new list vector
 175                         ListVector newList;
 176                         newList.setLabel(list.getLabel());
 177
 178                         //for each bin
 179                         for (int i = 0; i < list.getNumBins(); i++) {
 180
 181                                 //parse out names that are in accnos file
 182                                 string binnames = list.get(i);
 183
 184                                 string newNames = "";
 185                                 while (binnames.find_first_of(',') != -1) {
 186                                         string name = binnames.substr(0,binnames.find_first_of(','));
 187                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 188
 189                                         //if that name is in the .accnos file, add it
 190                                         if (names.count(name) == 0) {  newNames += name + ",";  }
 191                                 }
 192
 193                                 //get last name
 194                                 if (names.count(binnames) == 0) {  newNames += binnames;  }
 195
 196                                 //if there are names in this bin add to new list
 197                                 if (newNames != "") {  newList.push_back(newNames);     }
 198                         }
 199
 200                         //print new listvector
 201                         if (newList.getNumBins() != 0) {
 202                                 wroteSomething = true;
 203                                 newList.print(out);
 204                         }
 205
 206                         gobble(in);
 207                 }
 208                 in.close();
 209                 out.close();
 210
 211                 if (wroteSomething == false) {
 212                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 213                         remove(outputFileName.c_str());
 214                 }
 215
 216         }
 217         catch(exception& e) {
 218                 errorOut(e, "RemoveSeqsCommand", "readList");
 219                 exit(1);
 220         }
 221 }
 222 //**********************************************************************************************************************
 223 void RemoveSeqsCommand::readName(){
 224         try {
 225
 226                 string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
 227
 228                 ofstream out;
 229                 openOutputFile(outputFileName, out);
 230
 231                 ifstream in;
 232                 openInputFile(namefile, in);
 233                 string name, firstCol, secondCol;
 234
 235                 bool wroteSomething = false;
 236
 237                 while(!in.eof()){
 238
 239                         in >> firstCol;
 240                         in >> secondCol;
 241
 242                         vector<string> parsedNames;
 243                         //parse second column saving each name
 244                         while (secondCol.find_first_of(',') != -1) {
 245                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 246                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 247                                 parsedNames.push_back(name);
 248
 249                         }
 250
 251                         //get name after last ,
 252                         parsedNames.push_back(secondCol);
 253
 254                         vector<string> validSecond;  validSecond.clear();
 255                         for (int i = 0; i < parsedNames.size(); i++) {
 256                                 if (names.count(parsedNames[i]) == 0) {
 257                                         validSecond.push_back(parsedNames[i]);
 258                                 }
 259                         }
 260
 261                         //if the name in the first column is in the set then print it and any other names in second column also in set
 262                         if (names.count(firstCol) == 0) {
 263
 264                                 wroteSomething = true;
 265
 266                                 out << firstCol << '\t';
 267
 268                                 //you know you have at least one valid second since first column is valid
 269                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 270                                 out << validSecond[validSecond.size()-1] << endl;
 271
 272                         //make first name in set you come to first column and then add the remaining names to second column
 273                         }else {
 274
 275                                 //you want part of this row
 276                                 if (validSecond.size() != 0) {
 277
 278                                         wroteSomething = true;
 279
 280                                         out << validSecond[0] << '\t';
 281
 282                                         //you know you have at least one valid second since first column is valid
 283                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 284                                         out << validSecond[validSecond.size()-1] << endl;
 285                                 }
 286                         }
 287
 288                         gobble(in);
 289                 }
 290                 in.close();
 291                 out.close();
 292
 293                 if (wroteSomething == false) {
 294                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 295                         remove(outputFileName.c_str());
 296                 }
 297
 298         }
 299         catch(exception& e) {
 300                 errorOut(e, "RemoveSeqsCommand", "readName");
 301                 exit(1);
 302         }
 303 }
 304
 305 //**********************************************************************************************************************
 306 void RemoveSeqsCommand::readGroup(){
 307         try {
 308
 309                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 310                 ofstream out;
 311                 openOutputFile(outputFileName, out);
 312
 313                 ifstream in;
 314                 openInputFile(groupfile, in);
 315                 string name, group;
 316
 317                 bool wroteSomething = false;
 318
 319                 while(!in.eof()){
 320
 321                         in >> name;                             //read from first column
 322                         in >> group;                    //read from second column
 323
 324                         //if this name is in the accnos file
 325                         if (names.count(name) == 0) {
 326                                 wroteSomething = true;
 327                                 out << name << '\t' << group << endl;
 328                         }else {         names.erase(name);              }
 329
 330                         gobble(in);
 331                 }
 332                 in.close();
 333                 out.close();
 334
 335                 if (wroteSomething == false) {
 336                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 337                         remove(outputFileName.c_str());
 338                 }
 339
 340         }
 341         catch(exception& e) {
 342                 errorOut(e, "RemoveSeqsCommand", "readGroup");
 343                 exit(1);
 344         }
 345 }
 346
 347 //**********************************************************************************************************************
 348 //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
 349 void RemoveSeqsCommand::readAlign(){
 350         try {
 351                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 352                 ofstream out;
 353                 openOutputFile(outputFileName, out);
 354
 355                 ifstream in;
 356                 openInputFile(alignfile, in);
 357                 string name, junk;
 358
 359                 bool wroteSomething = false;
 360
 361                 //read column headers
 362                 for (int i = 0; i < 16; i++) {
 363                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 364                         else                    {       break;                  }
 365                 }
 366                 out << endl;
 367
 368                 while(!in.eof()){
 369
 370                         in >> name;                             //read from first column
 371
 372                         //if this name is in the accnos file
 373                         if (names.count(name) == 0) {
 374                                 wroteSomething = true;
 375
 376                                 out << name << '\t';
 377
 378                                 //read rest
 379                                 for (int i = 0; i < 15; i++) {
 380                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 381                                         else                    {       break;                  }
 382                                 }
 383                                 out << endl;
 384
 385                         }else {//still read just don't do anything with it
 386                                 names.erase(name);
 387
 388                                 //read rest
 389                                 for (int i = 0; i < 15; i++) {
 390                                         if (!in.eof())  {       in >> junk;             }
 391                                         else                    {       break;                  }
 392                                 }
 393                         }
 394
 395                         gobble(in);
 396                 }
 397                 in.close();
 398                 out.close();
 399
 400                 if (wroteSomething == false) {
 401                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 402                         remove(outputFileName.c_str());
 403                 }
 404
 405         }
 406         catch(exception& e) {
 407                 errorOut(e, "RemoveSeqsCommand", "readAlign");
 408                 exit(1);
 409         }
 410 }
 411 //**********************************************************************************************************************
 412 void RemoveSeqsCommand::readAccnos(){
 413         try {
 414
 415                 ifstream in;
 416                 openInputFile(accnosfile, in);
 417                 string name;
 418
 419                 while(!in.eof()){
 420                         in >> name;
 421
 422                         names.insert(name);
 423
 424                         gobble(in);
 425                 }
 426                 in.close();
 427
 428         }
 429         catch(exception& e) {
 430                 errorOut(e, "RemoveSeqsCommand", "readAccnos");
 431                 exit(1);
 432         }
 433 }
 434
 435 //**********************************************************************************************************************
 436
 437