removeseqscommand.cpp

   1 /*
   2  *  removeseqscommand.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 "removeseqscommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14
  15 RemoveSeqsCommand::RemoveSeqsCommand(string option){
  16         try {
  17                 abort = false;
  18
  19                 //allow user to run help
  20                 if(option == "help") { help(); abort = true; }
  21
  22                 else {
  23                         //valid paramters for this command
  24                         string Array[] =  {"fasta","name", "group", "alignreport", "accnos" };
  25                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  26
  27                         OptionParser parser(option);
  28                         map<string,string> parameters = parser.getParameters();
  29
  30                         ValidParameters validParameter;
  31
  32                         //check to make sure all parameters are valid for command
  33                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  34                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  35                         }
  36
  37                         //check for required parameters
  38                         accnosfile = validParameter.validFile(parameters, "accnos", true);
  39                         if (accnosfile == "not open") { abort = true; }
  40                         else if (accnosfile == "not found") {  accnosfile = "";  mothurOut("You must provide an accnos file."); mothurOutEndLine(); abort = true; }
  41
  42                         fastafile = validParameter.validFile(parameters, "fasta", true);
  43                         if (fastafile == "not open") { abort = true; }
  44                         else if (fastafile == "not found") {  fastafile = "";  }
  45
  46                         namefile = validParameter.validFile(parameters, "name", true);
  47                         if (namefile == "not open") { abort = true; }
  48                         else if (namefile == "not found") {  namefile = "";  }
  49
  50                         groupfile = validParameter.validFile(parameters, "group", true);
  51                         if (groupfile == "not open") { abort = true; }
  52                         else if (groupfile == "not found") {  groupfile = "";  }
  53
  54                         alignfile = validParameter.validFile(parameters, "alignreport", true);
  55                         if (alignfile == "not open") { abort = true; }
  56                         else if (alignfile == "not found") {  alignfile = "";  }
  57
  58                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == ""))  { mothurOut("You must provide one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true; }
  59
  60                         if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true;  }
  61                 }
  62
  63         }
  64         catch(exception& e) {
  65                 errorOut(e, "RemoveSeqsCommand", "RemoveSeqsCommand");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70
  71 void RemoveSeqsCommand::help(){
  72         try {
  73                 mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group or alignreport file.\n");
  74                 mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
  75                 mothurOut("The remove.seqs command parameters are accnos, fasta, name, group and alignreport.  You must provide accnos and one of the other parameters.\n");
  76                 mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
  77                 mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
  78                 mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
  79         }
  80         catch(exception& e) {
  81                 errorOut(e, "RemoveSeqsCommand", "help");
  82                 exit(1);
  83         }
  84 }
  85
  86 //**********************************************************************************************************************
  87
  88 int RemoveSeqsCommand::execute(){
  89         try {
  90
  91                 if (abort == true) { return 0; }
  92
  93                 //get names you want to keep
  94                 readAccnos();
  95
  96                 //read through the correct file and output lines you want to keep
  97                 if (fastafile != "")            {               readFasta();    }
  98                 else if (namefile != "")        {               readName();             }
  99                 else if (groupfile != "")       {               readGroup();    }
 100                 else if (alignfile != "")       {               readAlign();    }
 101
 102                 return 0;
 103         }
 104
 105         catch(exception& e) {
 106                 errorOut(e, "RemoveSeqsCommand", "execute");
 107                 exit(1);
 108         }
 109 }
 110
 111 //**********************************************************************************************************************
 112 void RemoveSeqsCommand::readFasta(){
 113         try {
 114                 string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
 115                 ofstream out;
 116                 openOutputFile(outputFileName, out);
 117
 118                 ifstream in;
 119                 openInputFile(fastafile, in);
 120                 string name;
 121
 122                 bool wroteSomething = false;
 123
 124                 while(!in.eof()){
 125                         Sequence currSeq(in);
 126                         name = currSeq.getName();
 127
 128                         //if this name is in the accnos file
 129                         if (names.count(name) == 0) {
 130                                 wroteSomething = true;
 131
 132                                 currSeq.printSequence(out);
 133                         }else {         names.erase(name);              }
 134
 135                         gobble(in);
 136                 }
 137                 in.close();
 138                 out.close();
 139
 140                 if (wroteSomething == false) {
 141                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 142                         remove(outputFileName.c_str());
 143                 }
 144
 145         }
 146         catch(exception& e) {
 147                 errorOut(e, "RemoveSeqsCommand", "readFasta");
 148                 exit(1);
 149         }
 150 }
 151
 152 //**********************************************************************************************************************
 153 void RemoveSeqsCommand::readName(){
 154         try {
 155
 156                 string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
 157
 158                 ofstream out;
 159                 openOutputFile(outputFileName, out);
 160
 161                 ifstream in;
 162                 openInputFile(namefile, in);
 163                 string name, firstCol, secondCol;
 164
 165                 bool wroteSomething = false;
 166
 167                 while(!in.eof()){
 168
 169                         in >> firstCol;
 170                         in >> secondCol;
 171
 172                         vector<string> parsedNames;
 173                         //parse second column saving each name
 174                         while (secondCol.find_first_of(',') != -1) {
 175                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 176                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 177                                 parsedNames.push_back(name);
 178
 179                         }
 180
 181                         //get name after last ,
 182                         parsedNames.push_back(secondCol);
 183
 184                         vector<string> validSecond;  validSecond.clear();
 185                         for (int i = 0; i < parsedNames.size(); i++) {
 186                                 if (names.count(parsedNames[i]) == 0) {
 187                                         validSecond.push_back(parsedNames[i]);
 188                                 }
 189                         }
 190
 191                         //if the name in the first column is in the set then print it and any other names in second column also in set
 192                         if (names.count(firstCol) == 0) {
 193
 194                                 wroteSomething = true;
 195
 196                                 out << firstCol << '\t';
 197
 198                                 //you know you have at least one valid second since first column is valid
 199                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 200                                 out << validSecond[validSecond.size()-1] << endl;
 201
 202                         //make first name in set you come to first column and then add the remaining names to second column
 203                         }else {
 204
 205                                 //you want part of this row
 206                                 if (validSecond.size() != 0) {
 207
 208                                         wroteSomething = true;
 209
 210                                         out << validSecond[0] << '\t';
 211
 212                                         //you know you have at least one valid second since first column is valid
 213                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 214                                         out << validSecond[validSecond.size()-1] << endl;
 215                                 }
 216                         }
 217
 218                         gobble(in);
 219                 }
 220                 in.close();
 221                 out.close();
 222
 223                 if (wroteSomething == false) {
 224                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 225                         remove(outputFileName.c_str());
 226                 }
 227
 228         }
 229         catch(exception& e) {
 230                 errorOut(e, "RemoveSeqsCommand", "readName");
 231                 exit(1);
 232         }
 233 }
 234
 235 //**********************************************************************************************************************
 236 void RemoveSeqsCommand::readGroup(){
 237         try {
 238
 239                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 240                 ofstream out;
 241                 openOutputFile(outputFileName, out);
 242
 243                 ifstream in;
 244                 openInputFile(groupfile, in);
 245                 string name, group;
 246
 247                 bool wroteSomething = false;
 248
 249                 while(!in.eof()){
 250
 251                         in >> name;                             //read from first column
 252                         in >> group;                    //read from second column
 253
 254                         //if this name is in the accnos file
 255                         if (names.count(name) == 0) {
 256                                 wroteSomething = true;
 257                                 out << name << '\t' << group << endl;
 258                         }else {         names.erase(name);              }
 259
 260                         gobble(in);
 261                 }
 262                 in.close();
 263                 out.close();
 264
 265                 if (wroteSomething == false) {
 266                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 267                         remove(outputFileName.c_str());
 268                 }
 269
 270         }
 271         catch(exception& e) {
 272                 errorOut(e, "RemoveSeqsCommand", "readGroup");
 273                 exit(1);
 274         }
 275 }
 276
 277 //**********************************************************************************************************************
 278 //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
 279 void RemoveSeqsCommand::readAlign(){
 280         try {
 281                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 282                 ofstream out;
 283                 openOutputFile(outputFileName, out);
 284
 285                 ifstream in;
 286                 openInputFile(alignfile, in);
 287                 string name, junk;
 288
 289                 bool wroteSomething = false;
 290
 291                 //read column headers
 292                 for (int i = 0; i < 16; i++) {
 293                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 294                         else                    {       break;                  }
 295                 }
 296                 out << endl;
 297
 298                 while(!in.eof()){
 299
 300                         in >> name;                             //read from first column
 301
 302                         //if this name is in the accnos file
 303                         if (names.count(name) == 0) {
 304                                 wroteSomething = true;
 305
 306                                 out << name << '\t';
 307
 308                                 //read rest
 309                                 for (int i = 0; i < 15; i++) {
 310                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 311                                         else                    {       break;                  }
 312                                 }
 313                                 out << endl;
 314
 315                         }else {//still read just don't do anything with it
 316                                 names.erase(name);
 317
 318                                 //read rest
 319                                 for (int i = 0; i < 15; i++) {
 320                                         if (!in.eof())  {       in >> junk;             }
 321                                         else                    {       break;                  }
 322                                 }
 323                         }
 324
 325                         gobble(in);
 326                 }
 327                 in.close();
 328                 out.close();
 329
 330                 if (wroteSomething == false) {
 331                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 332                         remove(outputFileName.c_str());
 333                 }
 334
 335         }
 336         catch(exception& e) {
 337                 errorOut(e, "RemoveSeqsCommand", "readAlign");
 338                 exit(1);
 339         }
 340 }
 341 //**********************************************************************************************************************
 342 void RemoveSeqsCommand::readAccnos(){
 343         try {
 344
 345                 ifstream in;
 346                 openInputFile(accnosfile, in);
 347                 string name;
 348
 349                 while(!in.eof()){
 350                         in >> name;
 351
 352                         names.insert(name);
 353
 354                         gobble(in);
 355                 }
 356                 in.close();
 357
 358         }
 359         catch(exception& e) {
 360                 errorOut(e, "RemoveSeqsCommand", "readAccnos");
 361                 exit(1);
 362         }
 363 }
 364
 365 //**********************************************************************************************************************
 366
 367