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