getlineagecommand.cpp

   1 /*
   2  *  getlineagecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/24/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getlineagecommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15
  16 GetLineageCommand::GetLineageCommand(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", "taxon", "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 = m->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 = m->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("list");
  64                                 //user has given a template file
  65                                 if(it != parameters.end()){
  66                                         path = m->hasPath(it->second);
  67                                         //if the user has not given a path then, add inputdir. else leave path alone.
  68                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
  69                                 }
  70
  71                                 it = parameters.find("name");
  72                                 //user has given a template file
  73                                 if(it != parameters.end()){
  74                                         path = m->hasPath(it->second);
  75                                         //if the user has not given a path then, add inputdir. else leave path alone.
  76                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
  77                                 }
  78
  79                                 it = parameters.find("group");
  80                                 //user has given a template file
  81                                 if(it != parameters.end()){
  82                                         path = m->hasPath(it->second);
  83                                         //if the user has not given a path then, add inputdir. else leave path alone.
  84                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
  85                                 }
  86
  87                                 it = parameters.find("taxonomy");
  88                                 //user has given a template file
  89                                 if(it != parameters.end()){
  90                                         path = m->hasPath(it->second);
  91                                         //if the user has not given a path then, add inputdir. else leave path alone.
  92                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
  93                                 }
  94                         }
  95
  96
  97                         //check for required parameters
  98                         fastafile = validParameter.validFile(parameters, "fasta", true);
  99                         if (fastafile == "not open") { abort = true; }
 100                         else if (fastafile == "not found") {  fastafile = "";  }
 101
 102                         namefile = validParameter.validFile(parameters, "name", true);
 103                         if (namefile == "not open") { abort = true; }
 104                         else if (namefile == "not found") {  namefile = "";  }
 105
 106                         groupfile = validParameter.validFile(parameters, "group", true);
 107                         if (groupfile == "not open") { abort = true; }
 108                         else if (groupfile == "not found") {  groupfile = "";  }
 109
 110                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 111                         if (alignfile == "not open") { abort = true; }
 112                         else if (alignfile == "not found") {  alignfile = "";  }
 113
 114                         listfile = validParameter.validFile(parameters, "list", true);
 115                         if (listfile == "not open") { abort = true; }
 116                         else if (listfile == "not found") {  listfile = "";  }
 117
 118                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 119                         if (taxfile == "not open") { abort = true; }
 120                         else if (taxfile == "not found") {  taxfile = ""; m->mothurOut("The taxonomy parameter is required for the get.lineage command."); m->mothurOutEndLine();  abort = true; }
 121
 122                         string usedDups = "true";
 123                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 124                         dups = m->isTrue(temp);
 125
 126                         taxons = validParameter.validFile(parameters, "taxon", false);
 127                         if (taxons == "not found") { taxons = "";  m->mothurOut("No taxons given, please correct."); m->mothurOutEndLine();  abort = true;  }
 128                         else {
 129                                 //rip off quotes
 130                                 if (taxons[0] == '\'') {  taxons = taxons.substr(1); }
 131                                 if (taxons[(taxons.length()-1)] == '\'') {  taxons = taxons.substr(0, (taxons.length()-1)); }
 132                         }
 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                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 138
 139                 }
 140
 141         }
 142         catch(exception& e) {
 143                 m->errorOut(e, "GetLineageCommand", "GetLineageCommand");
 144                 exit(1);
 145         }
 146 }
 147 //**********************************************************************************************************************
 148
 149 void GetLineageCommand::help(){
 150         try {
 151                 m->mothurOut("The get.lineage command reads a taxonomy file and any of the following file types: fasta, name, group, list or alignreport file.\n");
 152                 m->mothurOut("It outputs a file containing only the sequences from the taxonomy file that are from the taxon requested.\n");
 153                 m->mothurOut("The get.lineage command parameters are taxon, fasta, name, group, list, taxonomy, alignreport and dups.  You must provide taxonomy and taxon.\n");
 154                 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");
 155                 m->mothurOut("The taxon parameter allows you to select the taxons you would like to get.\n");
 156                 m->mothurOut("The get.lineage command should be in the following format: get.lineage(taxonomy=yourTaxonomyFile, taxon=yourTaxons).\n");
 157                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon=Bacteria;Firmicutes;Bacilli;Lactobacillales;).\n");
 158                 m->mothurOut("Note: If you are running mothur in script mode you must wrap the taxon in ' characters so mothur will ignore the ; in the taxon.\n");
 159                 m->mothurOut("Example get.lineage(taxonomy=amazon.silva.taxonomy, taxon='Bacteria;Firmicutes;Bacilli;Lactobacillales;').\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, "GetLineageCommand", "help");
 164                 exit(1);
 165         }
 166 }
 167
 168 //**********************************************************************************************************************
 169
 170 int GetLineageCommand::execute(){
 171         try {
 172
 173                 if (abort == true) { return 0; }
 174
 175                 if (m->control_pressed) { return 0; }
 176
 177                 //read through the correct file and output lines you want to keep
 178                 if (taxfile != "")                      {               readTax();              }  //fills the set of names to get
 179                 if (namefile != "")                     {               readName();             }
 180                 if (fastafile != "")            {               readFasta();    }
 181                 if (groupfile != "")            {               readGroup();    }
 182                 if (alignfile != "")            {               readAlign();    }
 183                 if (listfile != "")                     {               readList();             }
 184
 185
 186                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        remove(outputNames[i].c_str());  } return 0; }
 187
 188                 if (outputNames.size() != 0) {
 189                         m->mothurOutEndLine();
 190                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 191                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 192                         m->mothurOutEndLine();
 193                 }
 194
 195                 return 0;
 196         }
 197
 198         catch(exception& e) {
 199                 m->errorOut(e, "GetLineageCommand", "execute");
 200                 exit(1);
 201         }
 202 }
 203
 204 //**********************************************************************************************************************
 205 int GetLineageCommand::readFasta(){
 206         try {
 207                 string thisOutputDir = outputDir;
 208                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 209                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 210                 ofstream out;
 211                 m->openOutputFile(outputFileName, out);
 212
 213
 214                 ifstream in;
 215                 m->openInputFile(fastafile, in);
 216                 string name;
 217
 218                 bool wroteSomething = false;
 219
 220                 while(!in.eof()){
 221
 222                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 223
 224                         Sequence currSeq(in);
 225                         name = currSeq.getName();
 226
 227                         if (name != "") {
 228                                 //if this name is in the accnos file
 229                                 if (names.count(name) != 0) {
 230                                         wroteSomething = true;
 231
 232                                         currSeq.printSequence(out);
 233                                 }
 234                         }
 235                         m->gobble(in);
 236                 }
 237                 in.close();
 238                 out.close();
 239
 240                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 241                 outputNames.push_back(outputFileName);
 242
 243                 return 0;
 244
 245         }
 246         catch(exception& e) {
 247                 m->errorOut(e, "GetLineageCommand", "readFasta");
 248                 exit(1);
 249         }
 250 }
 251 //**********************************************************************************************************************
 252 int GetLineageCommand::readList(){
 253         try {
 254                 string thisOutputDir = outputDir;
 255                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 256                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 257                 ofstream out;
 258                 m->openOutputFile(outputFileName, out);
 259
 260                 ifstream in;
 261                 m->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                         m->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, "GetLineageCommand", "readList");
 320                 exit(1);
 321         }
 322 }
 323 //**********************************************************************************************************************
 324 int GetLineageCommand::readName(){
 325         try {
 326                 string thisOutputDir = outputDir;
 327                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 328                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 329                 ofstream out;
 330                 m->openOutputFile(outputFileName, out);
 331
 332
 333                 ifstream in;
 334                 m->openInputFile(namefile, in);
 335                 string name, firstCol, secondCol;
 336
 337                 bool wroteSomething = false;
 338
 339
 340                 while(!in.eof()){
 341
 342                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 343
 344                         in >> firstCol;
 345                         in >> secondCol;
 346
 347                         string hold = "";
 348                         if (dups) { hold = secondCol; }
 349
 350                         vector<string> parsedNames;
 351                         //parse second column saving each name
 352                         while (secondCol.find_first_of(',') != -1) {
 353                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 354                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 355                                 parsedNames.push_back(name);
 356                         }
 357
 358                         //get name after last ,
 359                         parsedNames.push_back(secondCol);
 360
 361                         vector<string> validSecond;
 362                         for (int i = 0; i < parsedNames.size(); i++) {
 363                                 if (names.count(parsedNames[i]) != 0) {
 364                                         validSecond.push_back(parsedNames[i]);
 365                                 }
 366                         }
 367
 368                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 369                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 370                                 out << firstCol << '\t' << hold << endl;
 371                                 wroteSomething = true;
 372                         }else {
 373                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 374                                 if (names.count(firstCol) != 0) {
 375
 376                                         wroteSomething = true;
 377
 378                                         out << firstCol << '\t';
 379
 380                                         //you know you have at least one valid second since first column is valid
 381                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 382                                         out << validSecond[validSecond.size()-1] << endl;
 383
 384
 385                                 //make first name in set you come to first column and then add the remaining names to second column
 386                                 }else {
 387                                         //you want part of this row
 388                                         if (validSecond.size() != 0) {
 389
 390                                                 wroteSomething = true;
 391
 392                                                 out << validSecond[0] << '\t';
 393
 394                                                 //you know you have at least one valid second since first column is valid
 395                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 396                                                 out << validSecond[validSecond.size()-1] << endl;
 397                                         }
 398                                 }
 399                         }
 400                         m->gobble(in);
 401                 }
 402                 in.close();
 403                 out.close();
 404
 405                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 406                 outputNames.push_back(outputFileName);
 407
 408                 return 0;
 409
 410         }
 411         catch(exception& e) {
 412                 m->errorOut(e, "GetLineageCommand", "readName");
 413                 exit(1);
 414         }
 415 }
 416
 417 //**********************************************************************************************************************
 418 int GetLineageCommand::readGroup(){
 419         try {
 420                 string thisOutputDir = outputDir;
 421                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 422                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 423                 ofstream out;
 424                 m->openOutputFile(outputFileName, out);
 425
 426
 427                 ifstream in;
 428                 m->openInputFile(groupfile, in);
 429                 string name, group;
 430
 431                 bool wroteSomething = false;
 432
 433                 while(!in.eof()){
 434
 435                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 436
 437
 438                         in >> name;                             //read from first column
 439                         in >> group;                    //read from second column
 440
 441                         //if this name is in the accnos file
 442                         if (names.count(name) != 0) {
 443                                 wroteSomething = true;
 444
 445                                 out << name << '\t' << group << endl;
 446                         }
 447
 448                         m->gobble(in);
 449                 }
 450                 in.close();
 451                 out.close();
 452
 453                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 454                 outputNames.push_back(outputFileName);
 455
 456                 return 0;
 457
 458         }
 459         catch(exception& e) {
 460                 m->errorOut(e, "GetLineageCommand", "readGroup");
 461                 exit(1);
 462         }
 463 }
 464 //**********************************************************************************************************************
 465 int GetLineageCommand::readTax(){
 466         try {
 467                 string thisOutputDir = outputDir;
 468                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 469                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 470                 ofstream out;
 471                 m->openOutputFile(outputFileName, out);
 472
 473                 ifstream in;
 474                 m->openInputFile(taxfile, in);
 475                 string name, tax;
 476
 477                 bool wroteSomething = false;
 478
 479                 while(!in.eof()){
 480
 481                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 482
 483                         in >> name;                             //read from first column
 484                         in >> tax;                      //read from second column
 485
 486                         string newtax = tax;
 487
 488                         //if the users file contains confidence scores we want to ignore them when searching for the taxons
 489                         int hasConfidences = tax.find_first_of('(');
 490                         if (hasConfidences != string::npos) {
 491                                 newtax = removeConfidences(tax);
 492                         }
 493
 494                         int pos = newtax.find(taxons);
 495
 496                         if (pos != string::npos) { //this sequence contains the taxon the user wants
 497                                 names.insert(name);
 498                                 out << name << '\t' << tax << endl;
 499                         }
 500
 501                         m->gobble(in);
 502                 }
 503                 in.close();
 504                 out.close();
 505
 506                 if (names.size() == 0) { m->mothurOut("Your taxonomy file does not contain any sequences from " + taxons + "."); m->mothurOutEndLine();  }
 507                 outputNames.push_back(outputFileName);
 508
 509                 return 0;
 510
 511         }
 512         catch(exception& e) {
 513                 m->errorOut(e, "GetLineageCommand", "readTax");
 514                 exit(1);
 515         }
 516 }
 517 /**************************************************************************************************/
 518 string GetLineageCommand::removeConfidences(string tax) {
 519         try {
 520
 521                 string taxon = "";
 522                 int taxLength = tax.length();
 523                 for(int i=0;i<taxLength;i++){
 524                         if(tax[i] == ';'){
 525                                 taxon = taxon.substr(0, taxon.find_first_of('(')); //rip off confidence
 526                                 taxon += ";";
 527                         }
 528                         else{
 529                                 taxon += tax[i];
 530                         }
 531                 }
 532
 533                 return taxon;
 534         }
 535         catch(exception& e) {
 536                 m->errorOut(e, "GetLineageCommand", "removeConfidences");
 537                 exit(1);
 538         }
 539 }
 540 //**********************************************************************************************************************
 541 //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
 542 int GetLineageCommand::readAlign(){
 543         try {
 544                 string thisOutputDir = outputDir;
 545                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 546                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 547                 ofstream out;
 548                 m->openOutputFile(outputFileName, out);
 549
 550
 551                 ifstream in;
 552                 m->openInputFile(alignfile, in);
 553                 string name, junk;
 554
 555                 bool wroteSomething = false;
 556
 557                 //read column headers
 558                 for (int i = 0; i < 16; i++) {
 559                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 560                         else                    {       break;                  }
 561                 }
 562                 out << endl;
 563
 564                 while(!in.eof()){
 565
 566                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 567
 568
 569                         in >> name;                             //read from first column
 570
 571                         //if this name is in the accnos file
 572                         if (names.count(name) != 0) {
 573                                 wroteSomething = true;
 574
 575                                 out << name << '\t';
 576
 577                                 //read rest
 578                                 for (int i = 0; i < 15; i++) {
 579                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 580                                         else                    {       break;                  }
 581                                 }
 582                                 out << endl;
 583
 584                         }else {//still read just don't do anything with it
 585                                 //read rest
 586                                 for (int i = 0; i < 15; i++) {
 587                                         if (!in.eof())  {       in >> junk;             }
 588                                         else                    {       break;                  }
 589                                 }
 590                         }
 591
 592                         m->gobble(in);
 593                 }
 594                 in.close();
 595                 out.close();
 596
 597                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 598                 outputNames.push_back(outputFileName);
 599
 600                 return 0;
 601
 602         }
 603         catch(exception& e) {
 604                 m->errorOut(e, "GetLineageCommand", "readAlign");
 605                 exit(1);
 606         }
 607 }
 608 //**********************************************************************************************************************
 609
 610
 611
 612