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