getseqscommand.cpp

   1 /*
   2  *  getseqscommand.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 "getseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15
  16 GetSeqsCommand::GetSeqsCommand(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 = "";  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                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport or listfile."); m->mothurOutEndLine(); abort = true; }
 123
 124                         int okay = 2;
 125                         if (outputDir != "") { okay++; }
 126
 127                         if (parameters.size() > okay) { m->mothurOut("You may only enter one of the following: fasta, name, group, alignreport or listfile."); m->mothurOutEndLine(); abort = true;  }
 128                 }
 129
 130         }
 131         catch(exception& e) {
 132                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 133                 exit(1);
 134         }
 135 }
 136 //**********************************************************************************************************************
 137
 138 void GetSeqsCommand::help(){
 139         try {
 140                 m->mothurOut("The get.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
 141                 m->mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
 142                 m->mothurOut("The get.seqs command parameters are accnos, fasta, name, group, list and alignreport.  You must provide accnos and one of the other parameters.\n");
 143                 m->mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 144                 m->mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 145                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 146         }
 147         catch(exception& e) {
 148                 m->errorOut(e, "GetSeqsCommand", "help");
 149                 exit(1);
 150         }
 151 }
 152
 153 //**********************************************************************************************************************
 154
 155 int GetSeqsCommand::execute(){
 156         try {
 157
 158                 if (abort == true) { return 0; }
 159
 160                 //get names you want to keep
 161                 readAccnos();
 162
 163                 if (m->control_pressed) { return 0; }
 164
 165                 //read through the correct file and output lines you want to keep
 166                 if (fastafile != "")            {               readFasta();    }
 167                 else if (namefile != "")        {               readName();             }
 168                 else if (groupfile != "")       {               readGroup();    }
 169                 else if (alignfile != "")       {               readAlign();    }
 170                 else if (listfile != "")        {               readList();             }
 171
 172                 if (m->control_pressed) { return 0; }
 173
 174                 if (outputNames.size() != 0) {
 175                         m->mothurOutEndLine();
 176                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 177                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 178                         m->mothurOutEndLine();
 179                 }
 180
 181                 return 0;
 182         }
 183
 184         catch(exception& e) {
 185                 m->errorOut(e, "GetSeqsCommand", "execute");
 186                 exit(1);
 187         }
 188 }
 189
 190 //**********************************************************************************************************************
 191 int GetSeqsCommand::readFasta(){
 192         try {
 193                 if (outputDir == "") { outputDir += hasPath(fastafile); }
 194                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" +  getExtension(fastafile);
 195                 ofstream out;
 196                 openOutputFile(outputFileName, out);
 197
 198
 199                 ifstream in;
 200                 openInputFile(fastafile, in);
 201                 string name;
 202
 203                 bool wroteSomething = false;
 204
 205                 while(!in.eof()){
 206
 207                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 208
 209                         Sequence currSeq(in);
 210                         name = currSeq.getName();
 211
 212                         if (name != "") {
 213                                 //if this name is in the accnos file
 214                                 if (names.count(name) == 1) {
 215                                         wroteSomething = true;
 216
 217                                         currSeq.printSequence(out);
 218
 219                                         names.erase(name);
 220                                 }
 221                         }
 222                         gobble(in);
 223                 }
 224                 in.close();
 225                 out.close();
 226
 227                 if (wroteSomething == false) {
 228                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 229                         remove(outputFileName.c_str());
 230                 }else {  outputNames.push_back(outputFileName); }
 231
 232                 return 0;
 233
 234         }
 235         catch(exception& e) {
 236                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 237                 exit(1);
 238         }
 239 }
 240 //**********************************************************************************************************************
 241 int GetSeqsCommand::readList(){
 242         try {
 243                 if (outputDir == "") { outputDir += hasPath(listfile); }
 244                 string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" +  getExtension(listfile);
 245                 ofstream out;
 246                 openOutputFile(outputFileName, out);
 247
 248                 ifstream in;
 249                 openInputFile(listfile, in);
 250
 251                 bool wroteSomething = false;
 252
 253                 while(!in.eof()){
 254
 255                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 256
 257                         //read in list vector
 258                         ListVector list(in);
 259
 260                         //make a new list vector
 261                         ListVector newList;
 262                         newList.setLabel(list.getLabel());
 263
 264                         //for each bin
 265                         for (int i = 0; i < list.getNumBins(); i++) {
 266
 267                                 //parse out names that are in accnos file
 268                                 string binnames = list.get(i);
 269
 270                                 string newNames = "";
 271                                 while (binnames.find_first_of(',') != -1) {
 272                                         string name = binnames.substr(0,binnames.find_first_of(','));
 273                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 274
 275                                         //if that name is in the .accnos file, add it
 276                                         if (names.count(name) == 1) {  newNames += name + ",";  }
 277                                 }
 278
 279                                 //get last name
 280                                 if (names.count(binnames) == 1) {  newNames += binnames;  }
 281
 282                                 //if there are names in this bin add to new list
 283                                 if (newNames != "") {  newList.push_back(newNames);     }
 284                         }
 285
 286                         //print new listvector
 287                         if (newList.getNumBins() != 0) {
 288                                 wroteSomething = true;
 289                                 newList.print(out);
 290                         }
 291
 292                         gobble(in);
 293                 }
 294                 in.close();
 295                 out.close();
 296
 297                 if (wroteSomething == false) {
 298                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 299                         remove(outputFileName.c_str());
 300                 }else {  outputNames.push_back(outputFileName); }
 301
 302                 return 0;
 303
 304         }
 305         catch(exception& e) {
 306                 m->errorOut(e, "GetSeqsCommand", "readList");
 307                 exit(1);
 308         }
 309 }
 310 //**********************************************************************************************************************
 311 int GetSeqsCommand::readName(){
 312         try {
 313                 if (outputDir == "") { outputDir += hasPath(namefile); }
 314                 string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" +  getExtension(namefile);
 315                 ofstream out;
 316                 openOutputFile(outputFileName, out);
 317
 318
 319                 ifstream in;
 320                 openInputFile(namefile, in);
 321                 string name, firstCol, secondCol;
 322
 323                 bool wroteSomething = false;
 324
 325
 326                 while(!in.eof()){
 327
 328                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 329
 330                         in >> firstCol;
 331                         in >> secondCol;
 332
 333                         vector<string> parsedNames;
 334                         //parse second column saving each name
 335                         while (secondCol.find_first_of(',') != -1) {
 336                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 337                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 338                                 parsedNames.push_back(name);
 339                         }
 340
 341                         //get name after last ,
 342                         parsedNames.push_back(secondCol);
 343
 344                         vector<string> validSecond;
 345                         for (int i = 0; i < parsedNames.size(); i++) {
 346                                 if (names.count(parsedNames[i]) == 1) {
 347                                         validSecond.push_back(parsedNames[i]);
 348                                 }
 349                         }
 350
 351
 352                         //if the name in the first column is in the set then print it and any other names in second column also in set
 353                         if (names.count(firstCol) == 1) {
 354
 355                                 wroteSomething = true;
 356
 357                                 out << firstCol << '\t';
 358
 359                                 //you know you have at least one valid second since first column is valid
 360                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 361                                 out << validSecond[validSecond.size()-1] << endl;
 362
 363
 364                         //make first name in set you come to first column and then add the remaining names to second column
 365                         }else {
 366                                 //you want part of this row
 367                                 if (validSecond.size() != 0) {
 368
 369                                         wroteSomething = true;
 370
 371                                         out << validSecond[0] << '\t';
 372
 373                                         //you know you have at least one valid second since first column is valid
 374                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 375                                         out << validSecond[validSecond.size()-1] << endl;
 376                                 }
 377                         }
 378
 379                         gobble(in);
 380                 }
 381                 in.close();
 382                 out.close();
 383
 384                 if (wroteSomething == false) {
 385                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 386                         remove(outputFileName.c_str());
 387                 }else {  outputNames.push_back(outputFileName); }
 388
 389                 return 0;
 390
 391         }
 392         catch(exception& e) {
 393                 m->errorOut(e, "GetSeqsCommand", "readName");
 394                 exit(1);
 395         }
 396 }
 397
 398 //**********************************************************************************************************************
 399 int GetSeqsCommand::readGroup(){
 400         try {
 401                 if (outputDir == "") { outputDir += hasPath(groupfile); }
 402                 string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
 403                 ofstream out;
 404                 openOutputFile(outputFileName, out);
 405
 406
 407                 ifstream in;
 408                 openInputFile(groupfile, in);
 409                 string name, group;
 410
 411                 bool wroteSomething = false;
 412
 413                 while(!in.eof()){
 414
 415                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 416
 417
 418                         in >> name;                             //read from first column
 419                         in >> group;                    //read from second column
 420
 421                         //if this name is in the accnos file
 422                         if (names.count(name) == 1) {
 423                                 wroteSomething = true;
 424
 425                                 out << name << '\t' << group << endl;
 426
 427                                 names.erase(name);
 428                         }
 429
 430                         gobble(in);
 431                 }
 432                 in.close();
 433                 out.close();
 434
 435                 if (wroteSomething == false) {
 436                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 437                         remove(outputFileName.c_str());
 438                 }else {  outputNames.push_back(outputFileName); }
 439
 440                 return 0;
 441
 442         }
 443         catch(exception& e) {
 444                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 445                 exit(1);
 446         }
 447 }
 448
 449 //**********************************************************************************************************************
 450 //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
 451 int GetSeqsCommand::readAlign(){
 452         try {
 453                 if (outputDir == "") { outputDir += hasPath(alignfile); }
 454                 string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
 455                 ofstream out;
 456                 openOutputFile(outputFileName, out);
 457
 458
 459                 ifstream in;
 460                 openInputFile(alignfile, in);
 461                 string name, junk;
 462
 463                 bool wroteSomething = false;
 464
 465                 //read column headers
 466                 for (int i = 0; i < 16; i++) {
 467                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 468                         else                    {       break;                  }
 469                 }
 470                 out << endl;
 471
 472                 while(!in.eof()){
 473
 474                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 475
 476
 477                         in >> name;                             //read from first column
 478
 479                         //if this name is in the accnos file
 480                         if (names.count(name) == 1) {
 481                                 wroteSomething = true;
 482
 483                                 out << name << '\t';
 484
 485                                 //read rest
 486                                 for (int i = 0; i < 15; i++) {
 487                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 488                                         else                    {       break;                  }
 489                                 }
 490                                 out << endl;
 491
 492                                 names.erase(name);
 493
 494                         }else {//still read just don't do anything with it
 495                                 //read rest
 496                                 for (int i = 0; i < 15; i++) {
 497                                         if (!in.eof())  {       in >> junk;             }
 498                                         else                    {       break;                  }
 499                                 }
 500                         }
 501
 502                         gobble(in);
 503                 }
 504                 in.close();
 505                 out.close();
 506
 507                 if (wroteSomething == false) {
 508                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 509                         remove(outputFileName.c_str());
 510                 }else {  outputNames.push_back(outputFileName); }
 511
 512                 return 0;
 513
 514         }
 515         catch(exception& e) {
 516                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 517                 exit(1);
 518         }
 519 }
 520 //**********************************************************************************************************************
 521
 522 int GetSeqsCommand::readAccnos(){
 523         try {
 524
 525                 ifstream in;
 526                 openInputFile(accnosfile, in);
 527                 string name;
 528
 529                 while(!in.eof()){
 530                         in >> name;
 531
 532                         names.insert(name);
 533
 534                         gobble(in);
 535                 }
 536                 in.close();
 537
 538                 return 0;
 539
 540         }
 541         catch(exception& e) {
 542                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 543                 exit(1);
 544         }
 545 }
 546
 547 //**********************************************************************************************************************
 548