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 != "") {
 284                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 285                                         newList.push_back(newNames);
 286                                 }
 287                         }
 288
 289                         //print new listvector
 290                         if (newList.getNumBins() != 0) {
 291                                 wroteSomething = true;
 292                                 newList.print(out);
 293                         }
 294
 295                         gobble(in);
 296                 }
 297                 in.close();
 298                 out.close();
 299
 300                 if (wroteSomething == false) {
 301                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 302                         remove(outputFileName.c_str());
 303                 }else {  outputNames.push_back(outputFileName); }
 304
 305                 return 0;
 306
 307         }
 308         catch(exception& e) {
 309                 m->errorOut(e, "GetSeqsCommand", "readList");
 310                 exit(1);
 311         }
 312 }
 313 //**********************************************************************************************************************
 314 int GetSeqsCommand::readName(){
 315         try {
 316                 if (outputDir == "") { outputDir += hasPath(namefile); }
 317                 string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" +  getExtension(namefile);
 318                 ofstream out;
 319                 openOutputFile(outputFileName, out);
 320
 321
 322                 ifstream in;
 323                 openInputFile(namefile, in);
 324                 string name, firstCol, secondCol;
 325
 326                 bool wroteSomething = false;
 327
 328
 329                 while(!in.eof()){
 330
 331                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 332
 333                         in >> firstCol;
 334                         in >> secondCol;
 335
 336                         vector<string> parsedNames;
 337                         //parse second column saving each name
 338                         while (secondCol.find_first_of(',') != -1) {
 339                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 340                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 341                                 parsedNames.push_back(name);
 342                         }
 343
 344                         //get name after last ,
 345                         parsedNames.push_back(secondCol);
 346
 347                         vector<string> validSecond;
 348                         for (int i = 0; i < parsedNames.size(); i++) {
 349                                 if (names.count(parsedNames[i]) == 1) {
 350                                         validSecond.push_back(parsedNames[i]);
 351                                 }
 352                         }
 353
 354
 355                         //if the name in the first column is in the set then print it and any other names in second column also in set
 356                         if (names.count(firstCol) == 1) {
 357
 358                                 wroteSomething = true;
 359
 360                                 out << firstCol << '\t';
 361
 362                                 //you know you have at least one valid second since first column is valid
 363                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 364                                 out << validSecond[validSecond.size()-1] << endl;
 365
 366
 367                         //make first name in set you come to first column and then add the remaining names to second column
 368                         }else {
 369                                 //you want part of this row
 370                                 if (validSecond.size() != 0) {
 371
 372                                         wroteSomething = true;
 373
 374                                         out << validSecond[0] << '\t';
 375
 376                                         //you know you have at least one valid second since first column is valid
 377                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 378                                         out << validSecond[validSecond.size()-1] << endl;
 379                                 }
 380                         }
 381
 382                         gobble(in);
 383                 }
 384                 in.close();
 385                 out.close();
 386
 387                 if (wroteSomething == false) {
 388                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 389                         remove(outputFileName.c_str());
 390                 }else {  outputNames.push_back(outputFileName); }
 391
 392                 return 0;
 393
 394         }
 395         catch(exception& e) {
 396                 m->errorOut(e, "GetSeqsCommand", "readName");
 397                 exit(1);
 398         }
 399 }
 400
 401 //**********************************************************************************************************************
 402 int GetSeqsCommand::readGroup(){
 403         try {
 404                 if (outputDir == "") { outputDir += hasPath(groupfile); }
 405                 string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
 406                 ofstream out;
 407                 openOutputFile(outputFileName, out);
 408
 409
 410                 ifstream in;
 411                 openInputFile(groupfile, in);
 412                 string name, group;
 413
 414                 bool wroteSomething = false;
 415
 416                 while(!in.eof()){
 417
 418                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 419
 420
 421                         in >> name;                             //read from first column
 422                         in >> group;                    //read from second column
 423
 424                         //if this name is in the accnos file
 425                         if (names.count(name) == 1) {
 426                                 wroteSomething = true;
 427
 428                                 out << name << '\t' << group << endl;
 429
 430                                 names.erase(name);
 431                         }
 432
 433                         gobble(in);
 434                 }
 435                 in.close();
 436                 out.close();
 437
 438                 if (wroteSomething == false) {
 439                         m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();
 440                         remove(outputFileName.c_str());
 441                 }else {  outputNames.push_back(outputFileName); }
 442
 443                 return 0;
 444
 445         }
 446         catch(exception& e) {
 447                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 448                 exit(1);
 449         }
 450 }
 451
 452 //**********************************************************************************************************************
 453 //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
 454 int GetSeqsCommand::readAlign(){
 455         try {
 456                 if (outputDir == "") { outputDir += hasPath(alignfile); }
 457                 string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
 458                 ofstream out;
 459                 openOutputFile(outputFileName, out);
 460
 461
 462                 ifstream in;
 463                 openInputFile(alignfile, in);
 464                 string name, junk;
 465
 466                 bool wroteSomething = false;
 467
 468                 //read column headers
 469                 for (int i = 0; i < 16; i++) {
 470                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 471                         else                    {       break;                  }
 472                 }
 473                 out << endl;
 474
 475                 while(!in.eof()){
 476
 477                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 478
 479
 480                         in >> name;                             //read from first column
 481
 482                         //if this name is in the accnos file
 483                         if (names.count(name) == 1) {
 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                                 names.erase(name);
 496
 497                         }else {//still read just don't do anything with it
 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 does not contain any sequence 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, "GetSeqsCommand", "readAlign");
 520                 exit(1);
 521         }
 522 }
 523 //**********************************************************************************************************************
 524
 525 int GetSeqsCommand::readAccnos(){
 526         try {
 527
 528                 ifstream in;
 529                 openInputFile(accnosfile, in);
 530                 string name;
 531
 532                 while(!in.eof()){
 533                         in >> name;
 534
 535                         names.insert(name);
 536
 537                         gobble(in);
 538                 }
 539                 in.close();
 540
 541                 return 0;
 542
 543         }
 544         catch(exception& e) {
 545                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 546                 exit(1);
 547         }
 548 }
 549
 550 //**********************************************************************************************************************
 551