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 (name != "") {
 129                                 //if this name is in the accnos file
 130                                 if (names.count(name) == 0) {
 131                                         wroteSomething = true;
 132
 133                                         currSeq.printSequence(out);
 134                                 }else {         names.erase(name);              }
 135                         }
 136                         gobble(in);
 137                 }
 138                 in.close();
 139                 out.close();
 140
 141                 if (wroteSomething == false) {
 142                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 143                         remove(outputFileName.c_str());
 144                 }
 145
 146         }
 147         catch(exception& e) {
 148                 errorOut(e, "RemoveSeqsCommand", "readFasta");
 149                 exit(1);
 150         }
 151 }
 152
 153 //**********************************************************************************************************************
 154 void RemoveSeqsCommand::readName(){
 155         try {
 156
 157                 string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
 158
 159                 ofstream out;
 160                 openOutputFile(outputFileName, out);
 161
 162                 ifstream in;
 163                 openInputFile(namefile, in);
 164                 string name, firstCol, secondCol;
 165
 166                 bool wroteSomething = false;
 167
 168                 while(!in.eof()){
 169
 170                         in >> firstCol;
 171                         in >> secondCol;
 172
 173                         vector<string> parsedNames;
 174                         //parse second column saving each name
 175                         while (secondCol.find_first_of(',') != -1) {
 176                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 177                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 178                                 parsedNames.push_back(name);
 179
 180                         }
 181
 182                         //get name after last ,
 183                         parsedNames.push_back(secondCol);
 184
 185                         vector<string> validSecond;  validSecond.clear();
 186                         for (int i = 0; i < parsedNames.size(); i++) {
 187                                 if (names.count(parsedNames[i]) == 0) {
 188                                         validSecond.push_back(parsedNames[i]);
 189                                 }
 190                         }
 191
 192                         //if the name in the first column is in the set then print it and any other names in second column also in set
 193                         if (names.count(firstCol) == 0) {
 194
 195                                 wroteSomething = true;
 196
 197                                 out << firstCol << '\t';
 198
 199                                 //you know you have at least one valid second since first column is valid
 200                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 201                                 out << validSecond[validSecond.size()-1] << endl;
 202
 203                         //make first name in set you come to first column and then add the remaining names to second column
 204                         }else {
 205
 206                                 //you want part of this row
 207                                 if (validSecond.size() != 0) {
 208
 209                                         wroteSomething = true;
 210
 211                                         out << validSecond[0] << '\t';
 212
 213                                         //you know you have at least one valid second since first column is valid
 214                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 215                                         out << validSecond[validSecond.size()-1] << endl;
 216                                 }
 217                         }
 218
 219                         gobble(in);
 220                 }
 221                 in.close();
 222                 out.close();
 223
 224                 if (wroteSomething == false) {
 225                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 226                         remove(outputFileName.c_str());
 227                 }
 228
 229         }
 230         catch(exception& e) {
 231                 errorOut(e, "RemoveSeqsCommand", "readName");
 232                 exit(1);
 233         }
 234 }
 235
 236 //**********************************************************************************************************************
 237 void RemoveSeqsCommand::readGroup(){
 238         try {
 239
 240                 string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
 241                 ofstream out;
 242                 openOutputFile(outputFileName, out);
 243
 244                 ifstream in;
 245                 openInputFile(groupfile, in);
 246                 string name, group;
 247
 248                 bool wroteSomething = false;
 249
 250                 while(!in.eof()){
 251
 252                         in >> name;                             //read from first column
 253                         in >> group;                    //read from second column
 254
 255                         //if this name is in the accnos file
 256                         if (names.count(name) == 0) {
 257                                 wroteSomething = true;
 258                                 out << name << '\t' << group << endl;
 259                         }else {         names.erase(name);              }
 260
 261                         gobble(in);
 262                 }
 263                 in.close();
 264                 out.close();
 265
 266                 if (wroteSomething == false) {
 267                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 268                         remove(outputFileName.c_str());
 269                 }
 270
 271         }
 272         catch(exception& e) {
 273                 errorOut(e, "RemoveSeqsCommand", "readGroup");
 274                 exit(1);
 275         }
 276 }
 277
 278 //**********************************************************************************************************************
 279 //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
 280 void RemoveSeqsCommand::readAlign(){
 281         try {
 282                 string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
 283                 ofstream out;
 284                 openOutputFile(outputFileName, out);
 285
 286                 ifstream in;
 287                 openInputFile(alignfile, in);
 288                 string name, junk;
 289
 290                 bool wroteSomething = false;
 291
 292                 //read column headers
 293                 for (int i = 0; i < 16; i++) {
 294                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 295                         else                    {       break;                  }
 296                 }
 297                 out << endl;
 298
 299                 while(!in.eof()){
 300
 301                         in >> name;                             //read from first column
 302
 303                         //if this name is in the accnos file
 304                         if (names.count(name) == 0) {
 305                                 wroteSomething = true;
 306
 307                                 out << name << '\t';
 308
 309                                 //read rest
 310                                 for (int i = 0; i < 15; i++) {
 311                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 312                                         else                    {       break;                  }
 313                                 }
 314                                 out << endl;
 315
 316                         }else {//still read just don't do anything with it
 317                                 names.erase(name);
 318
 319                                 //read rest
 320                                 for (int i = 0; i < 15; i++) {
 321                                         if (!in.eof())  {       in >> junk;             }
 322                                         else                    {       break;                  }
 323                                 }
 324                         }
 325
 326                         gobble(in);
 327                 }
 328                 in.close();
 329                 out.close();
 330
 331                 if (wroteSomething == false) {
 332                         mothurOut("Your file contains only sequences from the .accnos file."); mothurOutEndLine();
 333                         remove(outputFileName.c_str());
 334                 }
 335
 336         }
 337         catch(exception& e) {
 338                 errorOut(e, "RemoveSeqsCommand", "readAlign");
 339                 exit(1);
 340         }
 341 }
 342 //**********************************************************************************************************************
 343 void RemoveSeqsCommand::readAccnos(){
 344         try {
 345
 346                 ifstream in;
 347                 openInputFile(accnosfile, in);
 348                 string name;
 349
 350                 while(!in.eof()){
 351                         in >> name;
 352
 353                         names.insert(name);
 354
 355                         gobble(in);
 356                 }
 357                 in.close();
 358
 359         }
 360         catch(exception& e) {
 361                 errorOut(e, "RemoveSeqsCommand", "readAccnos");
 362                 exit(1);
 363         }
 364 }
 365
 366 //**********************************************************************************************************************
 367
 368