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