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"};
  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
  33                         //check to make sure all parameters are valid for command
  34                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  35                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  36                         }
  37
  38                         //check for required parameters
  39                         accnosfile = validParameter.validFile(parameters, "accnos", true);
  40                         if (accnosfile == "not open") { abort = true; }
  41                         else if (accnosfile == "not found") {  accnosfile = "";  mothurOut("You must provide an accnos file."); mothurOutEndLine(); abort = true; }
  42
  43                         fastafile = validParameter.validFile(parameters, "fasta", true);
  44                         if (fastafile == "not open") { abort = true; }
  45                         else if (fastafile == "not found") {  fastafile = "";  }
  46
  47                         namefile = validParameter.validFile(parameters, "name", true);
  48                         if (namefile == "not open") { abort = true; }
  49                         else if (namefile == "not found") {  namefile = "";  }
  50
  51                         groupfile = validParameter.validFile(parameters, "group", true);
  52                         if (groupfile == "not open") { abort = true; }
  53                         else if (groupfile == "not found") {  groupfile = "";  }
  54
  55                         alignfile = validParameter.validFile(parameters, "alignreport", true);
  56                         if (alignfile == "not open") { abort = true; }
  57                         else if (alignfile == "not found") {  alignfile = "";  }
  58
  59                         listfile = validParameter.validFile(parameters, "list", true);
  60                         if (listfile == "not open") { abort = true; }
  61                         else if (listfile == "not found") {  listfile = "";  }
  62
  63                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == ""))  { mothurOut("You must provide one of the following: fasta, name, group, alignreport or listfile."); mothurOutEndLine(); abort = true; }
  64
  65                         if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport or listfile."); mothurOutEndLine(); abort = true;  }
  66                 }
  67
  68         }
  69         catch(exception& e) {
  70                 errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75
  76 void GetSeqsCommand::help(){
  77         try {
  78                 mothurOut("The get.seqs command reads an .accnos file and one of the following file types: fasta, name, group, list or alignreport file.\n");
  79                 mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
  80                 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");
  81                 mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
  82                 mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
  83                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  84         }
  85         catch(exception& e) {
  86                 errorOut(e, "GetSeqsCommand", "help");
  87                 exit(1);
  88         }
  89 }
  90
  91 //**********************************************************************************************************************
  92
  93 int GetSeqsCommand::execute(){
  94         try {
  95
  96                 if (abort == true) { return 0; }
  97
  98                 //get names you want to keep
  99                 readAccnos();
 100
 101                 //read through the correct file and output lines you want to keep
 102                 if (fastafile != "")            {               readFasta();    }
 103                 else if (namefile != "")        {               readName();             }
 104                 else if (groupfile != "")       {               readGroup();    }
 105                 else if (alignfile != "")       {               readAlign();    }
 106                 else if (listfile != "")        {               readList();             }
 107
 108                 return 0;
 109         }
 110
 111         catch(exception& e) {
 112                 errorOut(e, "GetSeqsCommand", "execute");
 113                 exit(1);
 114         }
 115 }
 116
 117 //**********************************************************************************************************************
 118 void GetSeqsCommand::readFasta(){
 119         try {
 120                 string outputFileName = getRootName(fastafile) + "pick" +  getExtension(fastafile);
 121                 ofstream out;
 122                 openOutputFile(outputFileName, out);
 123
 124                 ifstream in;
 125                 openInputFile(fastafile, in);
 126                 string name;
 127
 128                 bool wroteSomething = false;
 129
 130                 while(!in.eof()){
 131                         Sequence currSeq(in);
 132                         name = currSeq.getName();
 133
 134                         if (name != "") {
 135                                 //if this name is in the accnos file
 136                                 if (names.count(name) == 1) {
 137                                         wroteSomething = true;
 138
 139                                         currSeq.printSequence(out);
 140
 141                                         names.erase(name);
 142                                 }
 143                         }
 144                         gobble(in);
 145                 }
 146                 in.close();
 147                 out.close();
 148
 149                 if (wroteSomething == false) {
 150                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 151                         remove(outputFileName.c_str());
 152                 }
 153
 154         }
 155         catch(exception& e) {
 156                 errorOut(e, "GetSeqsCommand", "readFasta");
 157                 exit(1);
 158         }
 159 }
 160 //**********************************************************************************************************************
 161 void GetSeqsCommand::readList(){
 162         try {
 163                 string outputFileName = getRootName(listfile) + "pick" +  getExtension(listfile);
 164                 ofstream out;
 165                 openOutputFile(outputFileName, out);
 166
 167                 ifstream in;
 168                 openInputFile(listfile, in);
 169
 170                 bool wroteSomething = false;
 171
 172                 while(!in.eof()){
 173                         //read in list vector
 174                         ListVector list(in);
 175
 176                         //make a new list vector
 177                         ListVector newList;
 178                         newList.setLabel(list.getLabel());
 179
 180                         //for each bin
 181                         for (int i = 0; i < list.getNumBins(); i++) {
 182
 183                                 //parse out names that are in accnos file
 184                                 string binnames = list.get(i);
 185
 186                                 string newNames = "";
 187                                 while (binnames.find_first_of(',') != -1) {
 188                                         string name = binnames.substr(0,binnames.find_first_of(','));
 189                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 190
 191                                         //if that name is in the .accnos file, add it
 192                                         if (names.count(name) == 1) {  newNames += name + ",";  }
 193                                 }
 194
 195                                 //get last name
 196                                 if (names.count(binnames) == 1) {  newNames += binnames;  }
 197
 198                                 //if there are names in this bin add to new list
 199                                 if (newNames != "") {  newList.push_back(newNames);     }
 200                         }
 201
 202                         //print new listvector
 203                         if (newList.getNumBins() != 0) {
 204                                 wroteSomething = true;
 205                                 newList.print(out);
 206                         }
 207
 208                         gobble(in);
 209                 }
 210                 in.close();
 211                 out.close();
 212
 213                 if (wroteSomething == false) {
 214                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 215                         remove(outputFileName.c_str());
 216                 }
 217
 218         }
 219         catch(exception& e) {
 220                 errorOut(e, "GetSeqsCommand", "readList");
 221                 exit(1);
 222         }
 223 }
 224 //**********************************************************************************************************************
 225 void GetSeqsCommand::readName(){
 226         try {
 227
 228                 string outputFileName = getRootName(namefile) + "pick" +  getExtension(namefile);
 229                 ofstream out;
 230                 openOutputFile(outputFileName, out);
 231
 232                 ifstream in;
 233                 openInputFile(namefile, in);
 234                 string name, firstCol, secondCol;
 235
 236                 bool wroteSomething = false;
 237
 238
 239                 while(!in.eof()){
 240
 241                         in >> firstCol;
 242                         in >> secondCol;
 243
 244                         vector<string> parsedNames;
 245                         //parse second column saving each name
 246                         while (secondCol.find_first_of(',') != -1) {
 247                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 248                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 249                                 parsedNames.push_back(name);
 250                         }
 251
 252                         //get name after last ,
 253                         parsedNames.push_back(secondCol);
 254
 255                         vector<string> validSecond;
 256                         for (int i = 0; i < parsedNames.size(); i++) {
 257                                 if (names.count(parsedNames[i]) == 1) {
 258                                         validSecond.push_back(parsedNames[i]);
 259                                 }
 260                         }
 261
 262
 263                         //if the name in the first column is in the set then print it and any other names in second column also in set
 264                         if (names.count(firstCol) == 1) {
 265
 266                                 wroteSomething = true;
 267
 268                                 out << firstCol << '\t';
 269
 270                                 //you know you have at least one valid second since first column is valid
 271                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 272                                 out << validSecond[validSecond.size()-1] << endl;
 273
 274
 275                         //make first name in set you come to first column and then add the remaining names to second column
 276                         }else {
 277                                 //you want part of this row
 278                                 if (validSecond.size() != 0) {
 279
 280                                         wroteSomething = true;
 281
 282                                         out << validSecond[0] << '\t';
 283
 284                                         //you know you have at least one valid second since first column is valid
 285                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 286                                         out << validSecond[validSecond.size()-1] << endl;
 287                                 }
 288                         }
 289
 290                         gobble(in);
 291                 }
 292                 in.close();
 293                 out.close();
 294
 295                 if (wroteSomething == false) {
 296                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 297                         remove(outputFileName.c_str());
 298                 }
 299
 300         }
 301         catch(exception& e) {
 302                 errorOut(e, "GetSeqsCommand", "readName");
 303                 exit(1);
 304         }
 305 }
 306
 307 //**********************************************************************************************************************
 308 void GetSeqsCommand::readGroup(){
 309         try {
 310
 311                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 312                 ofstream out;
 313                 openOutputFile(outputFileName, out);
 314
 315                 ifstream in;
 316                 openInputFile(groupfile, in);
 317                 string name, group;
 318
 319                 bool wroteSomething = false;
 320
 321                 while(!in.eof()){
 322
 323                         in >> name;                             //read from first column
 324                         in >> group;                    //read from second column
 325
 326                         //if this name is in the accnos file
 327                         if (names.count(name) == 1) {
 328                                 wroteSomething = true;
 329
 330                                 out << name << '\t' << group << endl;
 331
 332                                 names.erase(name);
 333                         }
 334
 335                         gobble(in);
 336                 }
 337                 in.close();
 338                 out.close();
 339
 340                 if (wroteSomething == false) {
 341                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 342                         remove(outputFileName.c_str());
 343                 }
 344
 345         }
 346         catch(exception& e) {
 347                 errorOut(e, "GetSeqsCommand", "readGroup");
 348                 exit(1);
 349         }
 350 }
 351
 352 //**********************************************************************************************************************
 353 //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
 354 void GetSeqsCommand::readAlign(){
 355         try {
 356                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 357                 ofstream out;
 358                 openOutputFile(outputFileName, out);
 359
 360                 ifstream in;
 361                 openInputFile(alignfile, in);
 362                 string name, junk;
 363
 364                 bool wroteSomething = false;
 365
 366                 //read column headers
 367                 for (int i = 0; i < 16; i++) {
 368                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 369                         else                    {       break;                  }
 370                 }
 371                 out << endl;
 372
 373                 while(!in.eof()){
 374
 375                         in >> name;                             //read from first column
 376
 377                         //if this name is in the accnos file
 378                         if (names.count(name) == 1) {
 379                                 wroteSomething = true;
 380
 381                                 out << name << '\t';
 382
 383                                 //read rest
 384                                 for (int i = 0; i < 15; i++) {
 385                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 386                                         else                    {       break;                  }
 387                                 }
 388                                 out << endl;
 389
 390                                 names.erase(name);
 391
 392                         }else {//still read just don't do anything with it
 393                                 //read rest
 394                                 for (int i = 0; i < 15; i++) {
 395                                         if (!in.eof())  {       in >> junk;             }
 396                                         else                    {       break;                  }
 397                                 }
 398                         }
 399
 400                         gobble(in);
 401                 }
 402                 in.close();
 403                 out.close();
 404
 405                 if (wroteSomething == false) {
 406                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 407                         remove(outputFileName.c_str());
 408                 }
 409
 410         }
 411         catch(exception& e) {
 412                 errorOut(e, "GetSeqsCommand", "readAlign");
 413                 exit(1);
 414         }
 415 }
 416 //**********************************************************************************************************************
 417
 418 void GetSeqsCommand::readAccnos(){
 419         try {
 420
 421                 ifstream in;
 422                 openInputFile(accnosfile, in);
 423                 string name;
 424
 425                 while(!in.eof()){
 426                         in >> name;
 427
 428                         names.insert(name);
 429
 430                         gobble(in);
 431                 }
 432                 in.close();
 433
 434         }
 435         catch(exception& e) {
 436                 errorOut(e, "GetSeqsCommand", "readAccnos");
 437                 exit(1);
 438         }
 439 }
 440
 441 //**********************************************************************************************************************
 442