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