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 = "";  mothurOut("You must provide an accnos file."); 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 == ""))  { mothurOut("You must provide one of the following: fasta, name, group, alignreport or listfile."); mothurOutEndLine(); abort = true; }
 123
 124                         int okay = 2;
 125                         if (outputDir != "") { okay++; }
 126
 127                         if (parameters.size() > okay) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport or listfile."); mothurOutEndLine(); abort = true;  }
 128                 }
 129
 130         }
 131         catch(exception& e) {
 132                 errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 133                 exit(1);
 134         }
 135 }
 136 //**********************************************************************************************************************
 137
 138 void GetSeqsCommand::help(){
 139         try {
 140                 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                 mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
 142                 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                 mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 144                 mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 145                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 146         }
 147         catch(exception& e) {
 148                 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                 return 0;
 171         }
 172
 173         catch(exception& e) {
 174                 errorOut(e, "GetSeqsCommand", "execute");
 175                 exit(1);
 176         }
 177 }
 178
 179 //**********************************************************************************************************************
 180 void GetSeqsCommand::readFasta(){
 181         try {
 182                 if (outputDir == "") { outputDir += hasPath(fastafile); }
 183                 string outputFileName = outputDir + getRootName(getSimpleName(fastafile)) + "pick" +  getExtension(fastafile);
 184                 ofstream out;
 185                 openOutputFile(outputFileName, out);
 186
 187                 ifstream in;
 188                 openInputFile(fastafile, in);
 189                 string name;
 190
 191                 bool wroteSomething = false;
 192
 193                 while(!in.eof()){
 194                         Sequence currSeq(in);
 195                         name = currSeq.getName();
 196
 197                         if (name != "") {
 198                                 //if this name is in the accnos file
 199                                 if (names.count(name) == 1) {
 200                                         wroteSomething = true;
 201
 202                                         currSeq.printSequence(out);
 203
 204                                         names.erase(name);
 205                                 }
 206                         }
 207                         gobble(in);
 208                 }
 209                 in.close();
 210                 out.close();
 211
 212                 if (wroteSomething == false) {
 213                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 214                         remove(outputFileName.c_str());
 215                 }
 216
 217         }
 218         catch(exception& e) {
 219                 errorOut(e, "GetSeqsCommand", "readFasta");
 220                 exit(1);
 221         }
 222 }
 223 //**********************************************************************************************************************
 224 void GetSeqsCommand::readList(){
 225         try {
 226                 if (outputDir == "") { outputDir += hasPath(listfile); }
 227                 string outputFileName = outputDir + getRootName(getSimpleName(listfile)) + "pick" +  getExtension(listfile);
 228                 ofstream out;
 229                 openOutputFile(outputFileName, out);
 230
 231                 ifstream in;
 232                 openInputFile(listfile, in);
 233
 234                 bool wroteSomething = false;
 235
 236                 while(!in.eof()){
 237                         //read in list vector
 238                         ListVector list(in);
 239
 240                         //make a new list vector
 241                         ListVector newList;
 242                         newList.setLabel(list.getLabel());
 243
 244                         //for each bin
 245                         for (int i = 0; i < list.getNumBins(); i++) {
 246
 247                                 //parse out names that are in accnos file
 248                                 string binnames = list.get(i);
 249
 250                                 string newNames = "";
 251                                 while (binnames.find_first_of(',') != -1) {
 252                                         string name = binnames.substr(0,binnames.find_first_of(','));
 253                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 254
 255                                         //if that name is in the .accnos file, add it
 256                                         if (names.count(name) == 1) {  newNames += name + ",";  }
 257                                 }
 258
 259                                 //get last name
 260                                 if (names.count(binnames) == 1) {  newNames += binnames;  }
 261
 262                                 //if there are names in this bin add to new list
 263                                 if (newNames != "") {  newList.push_back(newNames);     }
 264                         }
 265
 266                         //print new listvector
 267                         if (newList.getNumBins() != 0) {
 268                                 wroteSomething = true;
 269                                 newList.print(out);
 270                         }
 271
 272                         gobble(in);
 273                 }
 274                 in.close();
 275                 out.close();
 276
 277                 if (wroteSomething == false) {
 278                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 279                         remove(outputFileName.c_str());
 280                 }
 281
 282         }
 283         catch(exception& e) {
 284                 errorOut(e, "GetSeqsCommand", "readList");
 285                 exit(1);
 286         }
 287 }
 288 //**********************************************************************************************************************
 289 void GetSeqsCommand::readName(){
 290         try {
 291                 if (outputDir == "") { outputDir += hasPath(namefile); }
 292                 string outputFileName = outputDir + getRootName(getSimpleName(namefile)) + "pick" +  getExtension(namefile);
 293                 ofstream out;
 294                 openOutputFile(outputFileName, out);
 295
 296                 ifstream in;
 297                 openInputFile(namefile, in);
 298                 string name, firstCol, secondCol;
 299
 300                 bool wroteSomething = false;
 301
 302
 303                 while(!in.eof()){
 304
 305                         in >> firstCol;
 306                         in >> secondCol;
 307
 308                         vector<string> parsedNames;
 309                         //parse second column saving each name
 310                         while (secondCol.find_first_of(',') != -1) {
 311                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 312                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 313                                 parsedNames.push_back(name);
 314                         }
 315
 316                         //get name after last ,
 317                         parsedNames.push_back(secondCol);
 318
 319                         vector<string> validSecond;
 320                         for (int i = 0; i < parsedNames.size(); i++) {
 321                                 if (names.count(parsedNames[i]) == 1) {
 322                                         validSecond.push_back(parsedNames[i]);
 323                                 }
 324                         }
 325
 326
 327                         //if the name in the first column is in the set then print it and any other names in second column also in set
 328                         if (names.count(firstCol) == 1) {
 329
 330                                 wroteSomething = true;
 331
 332                                 out << firstCol << '\t';
 333
 334                                 //you know you have at least one valid second since first column is valid
 335                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 336                                 out << validSecond[validSecond.size()-1] << endl;
 337
 338
 339                         //make first name in set you come to first column and then add the remaining names to second column
 340                         }else {
 341                                 //you want part of this row
 342                                 if (validSecond.size() != 0) {
 343
 344                                         wroteSomething = true;
 345
 346                                         out << validSecond[0] << '\t';
 347
 348                                         //you know you have at least one valid second since first column is valid
 349                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 350                                         out << validSecond[validSecond.size()-1] << endl;
 351                                 }
 352                         }
 353
 354                         gobble(in);
 355                 }
 356                 in.close();
 357                 out.close();
 358
 359                 if (wroteSomething == false) {
 360                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 361                         remove(outputFileName.c_str());
 362                 }
 363
 364         }
 365         catch(exception& e) {
 366                 errorOut(e, "GetSeqsCommand", "readName");
 367                 exit(1);
 368         }
 369 }
 370
 371 //**********************************************************************************************************************
 372 void GetSeqsCommand::readGroup(){
 373         try {
 374                 if (outputDir == "") { outputDir += hasPath(groupfile); }
 375                 string outputFileName = outputDir + getRootName(getSimpleName(groupfile)) + "pick" + getExtension(groupfile);
 376                 ofstream out;
 377                 openOutputFile(outputFileName, out);
 378
 379                 ifstream in;
 380                 openInputFile(groupfile, in);
 381                 string name, group;
 382
 383                 bool wroteSomething = false;
 384
 385                 while(!in.eof()){
 386
 387                         in >> name;                             //read from first column
 388                         in >> group;                    //read from second column
 389
 390                         //if this name is in the accnos file
 391                         if (names.count(name) == 1) {
 392                                 wroteSomething = true;
 393
 394                                 out << name << '\t' << group << endl;
 395
 396                                 names.erase(name);
 397                         }
 398
 399                         gobble(in);
 400                 }
 401                 in.close();
 402                 out.close();
 403
 404                 if (wroteSomething == false) {
 405                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 406                         remove(outputFileName.c_str());
 407                 }
 408
 409         }
 410         catch(exception& e) {
 411                 errorOut(e, "GetSeqsCommand", "readGroup");
 412                 exit(1);
 413         }
 414 }
 415
 416 //**********************************************************************************************************************
 417 //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
 418 void GetSeqsCommand::readAlign(){
 419         try {
 420                 if (outputDir == "") { outputDir += hasPath(alignfile); }
 421                 string outputFileName = outputDir + getRootName(getSimpleName(alignfile)) + "pick.align.report";
 422                 ofstream out;
 423                 openOutputFile(outputFileName, out);
 424
 425                 ifstream in;
 426                 openInputFile(alignfile, in);
 427                 string name, junk;
 428
 429                 bool wroteSomething = false;
 430
 431                 //read column headers
 432                 for (int i = 0; i < 16; i++) {
 433                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 434                         else                    {       break;                  }
 435                 }
 436                 out << endl;
 437
 438                 while(!in.eof()){
 439
 440                         in >> name;                             //read from first column
 441
 442                         //if this name is in the accnos file
 443                         if (names.count(name) == 1) {
 444                                 wroteSomething = true;
 445
 446                                 out << name << '\t';
 447
 448                                 //read rest
 449                                 for (int i = 0; i < 15; i++) {
 450                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 451                                         else                    {       break;                  }
 452                                 }
 453                                 out << endl;
 454
 455                                 names.erase(name);
 456
 457                         }else {//still read just don't do anything with it
 458                                 //read rest
 459                                 for (int i = 0; i < 15; i++) {
 460                                         if (!in.eof())  {       in >> junk;             }
 461                                         else                    {       break;                  }
 462                                 }
 463                         }
 464
 465                         gobble(in);
 466                 }
 467                 in.close();
 468                 out.close();
 469
 470                 if (wroteSomething == false) {
 471                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 472                         remove(outputFileName.c_str());
 473                 }
 474
 475         }
 476         catch(exception& e) {
 477                 errorOut(e, "GetSeqsCommand", "readAlign");
 478                 exit(1);
 479         }
 480 }
 481 //**********************************************************************************************************************
 482
 483 void GetSeqsCommand::readAccnos(){
 484         try {
 485
 486                 ifstream in;
 487                 openInputFile(accnosfile, in);
 488                 string name;
 489
 490                 while(!in.eof()){
 491                         in >> name;
 492
 493                         names.insert(name);
 494
 495                         gobble(in);
 496                 }
 497                 in.close();
 498
 499         }
 500         catch(exception& e) {
 501                 errorOut(e, "GetSeqsCommand", "readAccnos");
 502                 exit(1);
 503         }
 504 }
 505
 506 //**********************************************************************************************************************
 507