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", "dups", "list","taxonomy","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                                 it = parameters.find("taxonomy");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 101                                 }
 102                         }
 103
 104
 105                         //check for required parameters
 106                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 107                         if (accnosfile == "not open") { abort = true; }
 108                         else if (accnosfile == "not found") {  accnosfile = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 109
 110                         fastafile = validParameter.validFile(parameters, "fasta", true);
 111                         if (fastafile == "not open") { abort = true; }
 112                         else if (fastafile == "not found") {  fastafile = "";  }
 113
 114                         namefile = validParameter.validFile(parameters, "name", true);
 115                         if (namefile == "not open") { abort = true; }
 116                         else if (namefile == "not found") {  namefile = "";  }
 117
 118                         groupfile = validParameter.validFile(parameters, "group", true);
 119                         if (groupfile == "not open") { abort = true; }
 120                         else if (groupfile == "not found") {  groupfile = "";  }
 121
 122                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 123                         if (alignfile == "not open") { abort = true; }
 124                         else if (alignfile == "not found") {  alignfile = "";  }
 125
 126                         listfile = validParameter.validFile(parameters, "list", true);
 127                         if (listfile == "not open") { abort = true; }
 128                         else if (listfile == "not found") {  listfile = "";  }
 129
 130                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 131                         if (taxfile == "not open") { abort = true; }
 132                         else if (taxfile == "not found") {  taxfile = "";  }
 133
 134                         string usedDups = "true";
 135                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 136                         dups = isTrue(temp);
 137
 138                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy or listfile."); m->mothurOutEndLine(); abort = true; }
 139
 140                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 141
 142                 }
 143
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 147                 exit(1);
 148         }
 149 }
 150 //**********************************************************************************************************************
 151
 152 void GetSeqsCommand::help(){
 153         try {
 154                 m->mothurOut("The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, list, taxonomy or alignreport file.\n");
 155                 m->mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
 156                 m->mothurOut("The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide accnos and at least one of the other parameters.\n");
 157                 m->mothurOut("The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n");
 158                 m->mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 159                 m->mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 160                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 161         }
 162         catch(exception& e) {
 163                 m->errorOut(e, "GetSeqsCommand", "help");
 164                 exit(1);
 165         }
 166 }
 167
 168 //**********************************************************************************************************************
 169
 170 int GetSeqsCommand::execute(){
 171         try {
 172
 173                 if (abort == true) { return 0; }
 174
 175                 //get names you want to keep
 176                 readAccnos();
 177
 178                 if (m->control_pressed) { return 0; }
 179
 180                 //read through the correct file and output lines you want to keep
 181                 if (namefile != "")                     {               readName();             }
 182                 if (fastafile != "")            {               readFasta();    }
 183                 if (groupfile != "")            {               readGroup();    }
 184                 if (alignfile != "")            {               readAlign();    }
 185                 if (listfile != "")                     {               readList();             }
 186                 if (taxfile != "")                      {               readTax();              }
 187
 188                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 189
 190                 if (outputNames.size() != 0) {
 191                         m->mothurOutEndLine();
 192                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 193                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 194                         m->mothurOutEndLine();
 195                 }
 196
 197                 return 0;
 198         }
 199
 200         catch(exception& e) {
 201                 m->errorOut(e, "GetSeqsCommand", "execute");
 202                 exit(1);
 203         }
 204 }
 205
 206 //**********************************************************************************************************************
 207 int GetSeqsCommand::readFasta(){
 208         try {
 209                 if (outputDir == "") { outputDir += hasPath(fastafile); }
 210                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" +  getExtension(fastafile);
 211                 ofstream out;
 212                 openOutputFile(outputFileName, out);
 213
 214
 215                 ifstream in;
 216                 openInputFile(fastafile, in);
 217                 string name;
 218
 219                 bool wroteSomething = false;
 220
 221                 while(!in.eof()){
 222
 223                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 224
 225                         Sequence currSeq(in);
 226                         name = currSeq.getName();
 227
 228                         if (name != "") {
 229                                 //if this name is in the accnos file
 230                                 if (names.count(name) != 0) {
 231                                         wroteSomething = true;
 232
 233                                         currSeq.printSequence(out);
 234                                 }
 235                         }
 236                         gobble(in);
 237                 }
 238                 in.close();
 239                 out.close();
 240
 241                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 242                 outputNames.push_back(outputFileName);
 243
 244                 return 0;
 245
 246         }
 247         catch(exception& e) {
 248                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 249                 exit(1);
 250         }
 251 }
 252 //**********************************************************************************************************************
 253 int GetSeqsCommand::readList(){
 254         try {
 255                 if (outputDir == "") { outputDir += hasPath(listfile); }
 256                 string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" +  getExtension(listfile);
 257                 ofstream out;
 258                 openOutputFile(outputFileName, out);
 259
 260                 ifstream in;
 261                 openInputFile(listfile, in);
 262
 263                 bool wroteSomething = false;
 264
 265                 while(!in.eof()){
 266
 267                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 268
 269                         //read in list vector
 270                         ListVector list(in);
 271
 272                         //make a new list vector
 273                         ListVector newList;
 274                         newList.setLabel(list.getLabel());
 275
 276                         //for each bin
 277                         for (int i = 0; i < list.getNumBins(); i++) {
 278
 279                                 //parse out names that are in accnos file
 280                                 string binnames = list.get(i);
 281
 282                                 string newNames = "";
 283                                 while (binnames.find_first_of(',') != -1) {
 284                                         string name = binnames.substr(0,binnames.find_first_of(','));
 285                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 286
 287                                         //if that name is in the .accnos file, add it
 288                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 289                                 }
 290
 291                                 //get last name
 292                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 293
 294                                 //if there are names in this bin add to new list
 295                                 if (newNames != "") {
 296                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 297                                         newList.push_back(newNames);
 298                                 }
 299                         }
 300
 301                         //print new listvector
 302                         if (newList.getNumBins() != 0) {
 303                                 wroteSomething = true;
 304                                 newList.print(out);
 305                         }
 306
 307                         gobble(in);
 308                 }
 309                 in.close();
 310                 out.close();
 311
 312                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 313                 outputNames.push_back(outputFileName);
 314
 315                 return 0;
 316
 317         }
 318         catch(exception& e) {
 319                 m->errorOut(e, "GetSeqsCommand", "readList");
 320                 exit(1);
 321         }
 322 }
 323 //**********************************************************************************************************************
 324 int GetSeqsCommand::readName(){
 325         try {
 326                 if (outputDir == "") { outputDir += hasPath(namefile); }
 327                 string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" +  getExtension(namefile);
 328                 ofstream out;
 329                 openOutputFile(outputFileName, out);
 330
 331
 332                 ifstream in;
 333                 openInputFile(namefile, in);
 334                 string name, firstCol, secondCol;
 335
 336                 bool wroteSomething = false;
 337
 338
 339                 while(!in.eof()){
 340
 341                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 342
 343                         in >> firstCol;
 344                         in >> secondCol;
 345
 346                         string hold = "";
 347                         if (dups) { hold = secondCol; }
 348
 349                         vector<string> parsedNames;
 350                         //parse second column saving each name
 351                         while (secondCol.find_first_of(',') != -1) {
 352                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 353                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 354                                 parsedNames.push_back(name);
 355                         }
 356
 357                         //get name after last ,
 358                         parsedNames.push_back(secondCol);
 359
 360                         vector<string> validSecond;
 361                         for (int i = 0; i < parsedNames.size(); i++) {
 362                                 if (names.count(parsedNames[i]) != 0) {
 363                                         validSecond.push_back(parsedNames[i]);
 364                                 }
 365                         }
 366
 367                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 368                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 369                                 out << firstCol << '\t' << hold << endl;
 370                                 wroteSomething = true;
 371                         }else {
 372                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 373                                 if (names.count(firstCol) != 0) {
 374
 375                                         wroteSomething = true;
 376
 377                                         out << firstCol << '\t';
 378
 379                                         //you know you have at least one valid second since first column is valid
 380                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 381                                         out << validSecond[validSecond.size()-1] << endl;
 382
 383
 384                                 //make first name in set you come to first column and then add the remaining names to second column
 385                                 }else {
 386                                         //you want part of this row
 387                                         if (validSecond.size() != 0) {
 388
 389                                                 wroteSomething = true;
 390
 391                                                 out << validSecond[0] << '\t';
 392
 393                                                 //you know you have at least one valid second since first column is valid
 394                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 395                                                 out << validSecond[validSecond.size()-1] << endl;
 396                                         }
 397                                 }
 398                         }
 399                         gobble(in);
 400                 }
 401                 in.close();
 402                 out.close();
 403
 404                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 405                 outputNames.push_back(outputFileName);
 406
 407                 return 0;
 408
 409         }
 410         catch(exception& e) {
 411                 m->errorOut(e, "GetSeqsCommand", "readName");
 412                 exit(1);
 413         }
 414 }
 415
 416 //**********************************************************************************************************************
 417 int GetSeqsCommand::readGroup(){
 418         try {
 419                 if (outputDir == "") { outputDir += hasPath(groupfile); }
 420                 string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
 421                 ofstream out;
 422                 openOutputFile(outputFileName, out);
 423
 424
 425                 ifstream in;
 426                 openInputFile(groupfile, in);
 427                 string name, group;
 428
 429                 bool wroteSomething = false;
 430
 431                 while(!in.eof()){
 432
 433                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 434
 435
 436                         in >> name;                             //read from first column
 437                         in >> group;                    //read from second column
 438
 439                         //if this name is in the accnos file
 440                         if (names.count(name) != 0) {
 441                                 wroteSomething = true;
 442
 443                                 out << name << '\t' << group << endl;
 444                         }
 445
 446                         gobble(in);
 447                 }
 448                 in.close();
 449                 out.close();
 450
 451                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 452                 outputNames.push_back(outputFileName);
 453
 454                 return 0;
 455
 456         }
 457         catch(exception& e) {
 458                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 459                 exit(1);
 460         }
 461 }
 462 //**********************************************************************************************************************
 463 int GetSeqsCommand::readTax(){
 464         try {
 465                 if (outputDir == "") { outputDir += hasPath(taxfile); }
 466                 string outputFileName = outputDir + getRootName(getSimpleName(taxfile)) + "pick" + getExtension(taxfile);
 467                 ofstream out;
 468                 openOutputFile(outputFileName, out);
 469
 470                 ifstream in;
 471                 openInputFile(taxfile, in);
 472                 string name, tax;
 473
 474                 bool wroteSomething = false;
 475
 476                 while(!in.eof()){
 477
 478                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 479
 480                         in >> name;                             //read from first column
 481                         in >> tax;                      //read from second column
 482
 483                         //if this name is in the accnos file
 484                         if (names.count(name) != 0) {
 485                                 wroteSomething = true;
 486
 487                                 out << name << '\t' << tax << endl;
 488                         }
 489
 490                         gobble(in);
 491                 }
 492                 in.close();
 493                 out.close();
 494
 495                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 496                 outputNames.push_back(outputFileName);
 497
 498                 return 0;
 499
 500         }
 501         catch(exception& e) {
 502                 m->errorOut(e, "GetSeqsCommand", "readTax");
 503                 exit(1);
 504         }
 505 }
 506 //**********************************************************************************************************************
 507 //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
 508 int GetSeqsCommand::readAlign(){
 509         try {
 510                 if (outputDir == "") { outputDir += hasPath(alignfile); }
 511                 string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
 512                 ofstream out;
 513                 openOutputFile(outputFileName, out);
 514
 515
 516                 ifstream in;
 517                 openInputFile(alignfile, in);
 518                 string name, junk;
 519
 520                 bool wroteSomething = false;
 521
 522                 //read column headers
 523                 for (int i = 0; i < 16; i++) {
 524                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 525                         else                    {       break;                  }
 526                 }
 527                 out << endl;
 528
 529                 while(!in.eof()){
 530
 531                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 532
 533
 534                         in >> name;                             //read from first column
 535
 536                         //if this name is in the accnos file
 537                         if (names.count(name) != 0) {
 538                                 wroteSomething = true;
 539
 540                                 out << name << '\t';
 541
 542                                 //read rest
 543                                 for (int i = 0; i < 15; i++) {
 544                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 545                                         else                    {       break;                  }
 546                                 }
 547                                 out << endl;
 548
 549                         }else {//still read just don't do anything with it
 550                                 //read rest
 551                                 for (int i = 0; i < 15; i++) {
 552                                         if (!in.eof())  {       in >> junk;             }
 553                                         else                    {       break;                  }
 554                                 }
 555                         }
 556
 557                         gobble(in);
 558                 }
 559                 in.close();
 560                 out.close();
 561
 562                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 563                 outputNames.push_back(outputFileName);
 564
 565                 return 0;
 566
 567         }
 568         catch(exception& e) {
 569                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 570                 exit(1);
 571         }
 572 }
 573 //**********************************************************************************************************************
 574
 575 int GetSeqsCommand::readAccnos(){
 576         try {
 577
 578                 ifstream in;
 579                 openInputFile(accnosfile, in);
 580                 string name;
 581
 582                 while(!in.eof()){
 583                         in >> name;
 584
 585                         names.insert(name);
 586
 587                         gobble(in);
 588                 }
 589                 in.close();
 590
 591                 return 0;
 592
 593         }
 594         catch(exception& e) {
 595                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 596                 exit(1);
 597         }
 598 }
 599
 600 //**********************************************************************************************************************
 601