listseqscommand.cpp

   1 /*
   2  *  listseqscommand.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 "listseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "counttable.h"
  14
  15
  16 //**********************************************************************************************************************
  17 vector<string> ListSeqsCommand::setParameters(){
  18         try {
  19         CommandParameter pfastq("fastq", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pfastq);
  20                 CommandParameter pfasta("fasta", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pfasta);
  21                 CommandParameter pname("name", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pname);
  22         CommandParameter pcount("count", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pcount);
  23                 CommandParameter pgroup("group", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(pgroup);
  24                 CommandParameter plist("list", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(plist);
  25                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false,true); parameters.push_back(ptaxonomy);
  26                 CommandParameter palignreport("alignreport", "InputTypes", "", "", "FNGLT", "FNGLT", "none","accnos",false,false); parameters.push_back(palignreport);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "ListSeqsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string ListSeqsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The list.seqs command reads a fasta, name, group, count, list, taxonomy, fastq or alignreport file and outputs a .accnos file containing sequence names.\n";
  44                 helpString += "The list.seqs command parameters are fasta, name, group, count, list, taxonomy, fastq and alignreport.  You must provide one of these parameters.\n";
  45                 helpString += "The list.seqs command should be in the following format: list.seqs(fasta=yourFasta).\n";
  46                 helpString += "Example list.seqs(fasta=amazon.fasta).\n";
  47                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  48                 return helpString;
  49         }
  50         catch(exception& e) {
  51                 m->errorOut(e, "ListSeqsCommand", "getHelpString");
  52                 exit(1);
  53         }
  54 }
  55 //**********************************************************************************************************************
  56 string ListSeqsCommand::getOutputPattern(string type) {
  57     try {
  58         string pattern = "";
  59
  60         if (type == "accnos") {  pattern = "[filename],accnos"; }
  61         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  62
  63         return pattern;
  64     }
  65     catch(exception& e) {
  66         m->errorOut(e, "ListSeqsCommand", "getOutputPattern");
  67         exit(1);
  68     }
  69 }
  70 //**********************************************************************************************************************
  71 ListSeqsCommand::ListSeqsCommand(){
  72         try {
  73                 abort = true; calledHelp = true;
  74                 setParameters();
  75                 vector<string> tempOutNames;
  76                 outputTypes["accnos"] = tempOutNames;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84
  85 ListSeqsCommand::ListSeqsCommand(string option)  {
  86         try {
  87                 abort = false; calledHelp = false;
  88
  89                 //allow user to run help
  90                 if(option == "help") { help(); abort = true; calledHelp = true; }
  91                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  92                 else {
  93                         vector<string> myArray = setParameters();
  94
  95                         OptionParser parser(option);
  96                         map<string,string> parameters = parser.getParameters();
  97
  98                         ValidParameters validParameter;
  99                         map<string,string>::iterator it;
 100
 101                         //check to make sure all parameters are valid for command
 102                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 103                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 104                         }
 105
 106                         //initialize outputTypes
 107                         vector<string> tempOutNames;
 108                         outputTypes["accnos"] = tempOutNames;
 109
 110                         //if the user changes the output directory command factory will send this info to us in the output parameter
 111                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115                         if (inputDir == "not found"){   inputDir = "";          }
 116                         else {
 117                                 string path;
 118                                 it = parameters.find("alignreport");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 124                                 }
 125
 126                                 it = parameters.find("fasta");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 132                                 }
 133
 134                                 it = parameters.find("list");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 140                                 }
 141
 142                                 it = parameters.find("name");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("group");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("taxonomy");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 164                                 }
 165
 166                 it = parameters.find("count");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["count"] = inputDir + it->second;            }
 172                                 }
 173
 174                 it = parameters.find("fastq");
 175                                 //user has given a template file
 176                                 if(it != parameters.end()){
 177                                         path = m->hasPath(it->second);
 178                                         //if the user has not given a path then, add inputdir. else leave path alone.
 179                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 180                                 }
 181                         }
 182
 183                         //check for required parameters
 184                         fastafile = validParameter.validFile(parameters, "fasta", true);
 185                         if (fastafile == "not open") { abort = true; }
 186                         else if (fastafile == "not found") {  fastafile = "";  }
 187                         else { m->setFastaFile(fastafile); }
 188
 189                         namefile = validParameter.validFile(parameters, "name", true);
 190                         if (namefile == "not open") { abort = true; }
 191                         else if (namefile == "not found") {  namefile = "";  }
 192                         else { m->setNameFile(namefile); }
 193
 194                         groupfile = validParameter.validFile(parameters, "group", true);
 195                         if (groupfile == "not open") { abort = true; }
 196                         else if (groupfile == "not found") {  groupfile = "";  }
 197                         else { m->setGroupFile(groupfile); }
 198
 199                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 200                         if (alignfile == "not open") { abort = true; }
 201                         else if (alignfile == "not found") {  alignfile = "";  }
 202
 203                         listfile = validParameter.validFile(parameters, "list", true);
 204                         if (listfile == "not open") { abort = true; }
 205                         else if (listfile == "not found") {  listfile = "";  }
 206                         else { m->setListFile(listfile); }
 207
 208                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 209                         if (taxfile == "not open") { abort = true; }
 210                         else if (taxfile == "not found") {  taxfile = "";  }
 211                         else { m->setTaxonomyFile(taxfile); }
 212
 213             countfile = validParameter.validFile(parameters, "count", true);
 214                         if (countfile == "not open") { abort = true; }
 215                         else if (countfile == "not found") {  countfile = "";  }
 216                         else { m->setCountTableFile(countfile); }
 217
 218             fastqfile = validParameter.validFile(parameters, "fastq", true);
 219                         if (fastqfile == "not open") { abort = true; }
 220                         else if (fastqfile == "not found") {  fastqfile = "";  }
 221
 222                         if ((fastqfile == "") && (countfile == "") && (fastafile == "") && (namefile == "") && (listfile == "") && (groupfile == "") && (alignfile == "") && (taxfile == ""))  { m->mothurOut("You must provide a file."); m->mothurOutEndLine(); abort = true; }
 223
 224                         int okay = 1;
 225                         if (outputDir != "") { okay++; }
 226                         if (inputDir != "") { okay++; }
 227
 228                         if (parameters.size() > okay) { m->mothurOut("You may only enter one file."); m->mothurOutEndLine(); abort = true;  }
 229                 }
 230
 231         }
 232         catch(exception& e) {
 233                 m->errorOut(e, "ListSeqsCommand", "ListSeqsCommand");
 234                 exit(1);
 235         }
 236 }
 237 //**********************************************************************************************************************
 238
 239 int ListSeqsCommand::execute(){
 240         try {
 241
 242                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 243
 244                 //read functions fill names vector
 245                 if (fastafile != "")            {       inputFileName = fastafile;      readFasta();    }
 246         else if (fastqfile != "")       {       inputFileName = fastqfile;      readFastq();    }
 247                 else if (namefile != "")        {       inputFileName = namefile;       readName();             }
 248                 else if (groupfile != "")       {       inputFileName = groupfile;      readGroup();    }
 249                 else if (alignfile != "")       {       inputFileName = alignfile;      readAlign();    }
 250                 else if (listfile != "")        {       inputFileName = listfile;       readList();             }
 251                 else if (taxfile != "")         {       inputFileName = taxfile;        readTax();              }
 252         else if (countfile != "")       {       inputFileName = countfile;      readCount();    }
 253
 254                 if (m->control_pressed) { outputTypes.clear();  return 0; }
 255
 256                 //sort in alphabetical order
 257                 sort(names.begin(), names.end());
 258
 259                 if (outputDir == "") {  outputDir += m->hasPath(inputFileName);  }
 260
 261         map<string, string> variables;
 262         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(inputFileName));
 263                 string outputFileName = getOutputFileName("accnos", variables);
 264
 265                 ofstream out;
 266                 m->openOutputFile(outputFileName, out);
 267                 outputNames.push_back(outputFileName); outputTypes["accnos"].push_back(outputFileName);
 268
 269                 //output to .accnos file
 270                 for (int i = 0; i < names.size(); i++) {
 271
 272                         if (m->control_pressed) { outputTypes.clear(); out.close(); m->mothurRemove(outputFileName); return 0; }
 273
 274                         out << names[i] << endl;
 275                 }
 276                 out.close();
 277
 278                 if (m->control_pressed) { outputTypes.clear();  m->mothurRemove(outputFileName); return 0; }
 279
 280                 m->setAccnosFile(outputFileName);
 281
 282                 m->mothurOutEndLine();
 283                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 284                 m->mothurOut(outputFileName); m->mothurOutEndLine();
 285                 m->mothurOutEndLine();
 286
 287                 //set accnos file as new current accnosfile
 288                 string current = "";
 289                 itTypes = outputTypes.find("accnos");
 290                 if (itTypes != outputTypes.end()) {
 291                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setAccnosFile(current); }
 292                 }
 293
 294                 return 0;
 295         }
 296
 297         catch(exception& e) {
 298                 m->errorOut(e, "ListSeqsCommand", "execute");
 299                 exit(1);
 300         }
 301 }
 302 //**********************************************************************************************************************
 303 int ListSeqsCommand::readFastq(){
 304         try {
 305
 306                 ifstream in;
 307                 m->openInputFile(fastqfile, in);
 308                 string name;
 309
 310                 //ofstream out;
 311                 //string newFastaName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "numsAdded.fasta";
 312                 //m->openOutputFile(newFastaName, out);
 313                 int count = 1;
 314                 //string lastName = "";
 315
 316                 while(!in.eof()){
 317
 318                         if (m->control_pressed) { in.close(); return 0; }
 319
 320                         //read sequence name
 321                         string name = m->getline(in); m->gobble(in);
 322
 323                         if (name[0] == '@') {
 324                 vector<string> splits = m->splitWhiteSpace(name);
 325                 name = splits[0];
 326                 name = name.substr(1);
 327                 m->checkName(name);
 328                 names.push_back(name);
 329                 //get rest of lines
 330                 name = m->getline(in); m->gobble(in);
 331                 name = m->getline(in); m->gobble(in);
 332                 name = m->getline(in); m->gobble(in);
 333             }
 334
 335                         m->gobble(in);
 336                         if (m->debug) { count++; cout << "[DEBUG]: count = " + toString(count) + ", name = " + name + "\n"; }
 337                 }
 338                 in.close();
 339                 //out.close();
 340
 341                 return 0;
 342
 343         }
 344         catch(exception& e) {
 345                 m->errorOut(e, "ListSeqsCommand", "readFastq");
 346                 exit(1);
 347         }
 348 }
 349
 350 //**********************************************************************************************************************
 351 int ListSeqsCommand::readFasta(){
 352         try {
 353
 354                 ifstream in;
 355                 m->openInputFile(fastafile, in);
 356                 string name;
 357
 358                 //ofstream out;
 359                 //string newFastaName = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "numsAdded.fasta";
 360                 //m->openOutputFile(newFastaName, out);
 361                 int count = 1;
 362                 //string lastName = "";
 363
 364                 while(!in.eof()){
 365
 366                         if (m->control_pressed) { in.close(); return 0; }
 367
 368                         Sequence currSeq(in);
 369                         name = currSeq.getName();
 370
 371                         if (name != "") {  names.push_back(name);  }
 372
 373                         m->gobble(in);
 374                         if (m->debug) { count++; cout << "[DEBUG]: count = " + toString(count) + ", name = " + currSeq.getName() + "\n"; }
 375                 }
 376                 in.close();
 377                 //out.close();
 378
 379                 return 0;
 380
 381         }
 382         catch(exception& e) {
 383                 m->errorOut(e, "ListSeqsCommand", "readFasta");
 384                 exit(1);
 385         }
 386 }
 387 //**********************************************************************************************************************
 388 int ListSeqsCommand::readList(){
 389         try {
 390                 ifstream in;
 391                 m->openInputFile(listfile, in);
 392
 393                 if(!in.eof()){
 394                         //read in list vector
 395                         ListVector list(in);
 396
 397                         //for each bin
 398                         for (int i = 0; i < list.getNumBins(); i++) {
 399                                 string binnames = list.get(i);
 400
 401                                 if (m->control_pressed) { in.close(); return 0; }
 402
 403                                 m->splitAtComma(binnames, names);
 404                         }
 405                 }
 406                 in.close();
 407
 408                 return 0;
 409
 410         }
 411         catch(exception& e) {
 412                 m->errorOut(e, "ListSeqsCommand", "readList");
 413                 exit(1);
 414         }
 415 }
 416
 417 //**********************************************************************************************************************
 418 int ListSeqsCommand::readName(){
 419         try {
 420
 421                 ifstream in;
 422                 m->openInputFile(namefile, in);
 423                 string name, firstCol, secondCol;
 424
 425                 while(!in.eof()){
 426
 427                         if (m->control_pressed) { in.close(); return 0; }
 428
 429                         in >> firstCol;
 430                         in >> secondCol;
 431
 432                         //parse second column saving each name
 433                         m->splitAtComma(secondCol, names);
 434
 435                         m->gobble(in);
 436                 }
 437                 in.close();
 438                 return 0;
 439
 440         }
 441         catch(exception& e) {
 442                 m->errorOut(e, "ListSeqsCommand", "readName");
 443                 exit(1);
 444         }
 445 }
 446
 447 //**********************************************************************************************************************
 448 int ListSeqsCommand::readGroup(){
 449         try {
 450
 451                 ifstream in;
 452                 m->openInputFile(groupfile, in);
 453                 string name, group;
 454
 455                 while(!in.eof()){
 456
 457                         if (m->control_pressed) { in.close(); return 0; }
 458
 459                         in >> name;     m->gobble(in);                  //read from first column
 460                         in >> group;                    //read from second column
 461
 462                         names.push_back(name);
 463
 464                         m->gobble(in);
 465                 }
 466                 in.close();
 467                 return 0;
 468
 469         }
 470         catch(exception& e) {
 471                 m->errorOut(e, "ListSeqsCommand", "readGroup");
 472                 exit(1);
 473         }
 474 }
 475 //**********************************************************************************************************************
 476 int ListSeqsCommand::readCount(){
 477         try {
 478                 CountTable ct;
 479                 ct.readTable(countfile, false, false);
 480
 481         if (m->control_pressed) { return 0; }
 482
 483         names = ct.getNamesOfSeqs();
 484
 485         return 0;
 486
 487         }
 488         catch(exception& e) {
 489                 m->errorOut(e, "ListSeqsCommand", "readCount");
 490                 exit(1);
 491         }
 492 }
 493 //**********************************************************************************************************************
 494 //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
 495 int ListSeqsCommand::readAlign(){
 496         try {
 497
 498                 ifstream in;
 499                 m->openInputFile(alignfile, in);
 500                 string name, junk;
 501
 502                 //read column headers
 503                 for (int i = 0; i < 16; i++) {
 504                         if (!in.eof())  {       in >> junk;             }
 505                         else                    {       break;                  }
 506                 }
 507                 //m->getline(in);
 508
 509                 while(!in.eof()){
 510
 511                         if (m->control_pressed) { in.close(); return 0; }
 512
 513                         in >> name;                             //read from first column
 514                         //m->getline(in);
 515                         //read rest
 516                         for (int i = 0; i < 15; i++) {
 517                                 if (!in.eof())  {       in >> junk;             }
 518                                 else                    {       break;                  }
 519                         }
 520
 521                         names.push_back(name);
 522
 523                         m->gobble(in);
 524                 }
 525                 in.close();
 526
 527                 return 0;
 528
 529
 530         }
 531         catch(exception& e) {
 532                 m->errorOut(e, "ListSeqsCommand", "readAlign");
 533                 exit(1);
 534         }
 535 }
 536 //**********************************************************************************************************************
 537 int ListSeqsCommand::readTax(){
 538         try {
 539
 540                 ifstream in;
 541                 m->openInputFile(taxfile, in);
 542                 string name, firstCol, secondCol;
 543
 544                 while(!in.eof()){
 545
 546                         if (m->control_pressed) { in.close(); return 0; }
 547
 548                         in >> firstCol;
 549                         in >> secondCol;
 550
 551                         names.push_back(firstCol);
 552
 553                         m->gobble(in);
 554
 555                 }
 556                 in.close();
 557
 558                 return 0;
 559
 560         }
 561         catch(exception& e) {
 562                 m->errorOut(e, "ListSeqsCommand", "readTax");
 563                 exit(1);
 564         }
 565 }
 566 //**********************************************************************************************************************