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
  13 //**********************************************************************************************************************
  14
  15 GetSeqsCommand::GetSeqsCommand(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", "align", "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, "align", 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, align."); mothurOutEndLine(); abort = true; }
  59
  60                         if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, align."); mothurOutEndLine(); abort = true;  }
  61                 }
  62
  63         }
  64         catch(exception& e) {
  65                 errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  66                 exit(1);
  67         }
  68 }
  69 //**********************************************************************************************************************
  70
  71 void GetSeqsCommand::help(){
  72         try {
  73                 mothurOut("The get.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 only the sequences in the .accnos file.\n");
  75                 mothurOut("The get.seqs command parameters are accnos, fasta, name, group and align.  You must provide accnos and one of the other parameters.\n");
  76                 mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
  77                 mothurOut("Example get.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, "GetSeqsCommand", "help");
  82                 exit(1);
  83         }
  84 }
  85
  86 //**********************************************************************************************************************
  87
  88 int GetSeqsCommand::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, "GetSeqsCommand", "execute");
 107                 exit(1);
 108         }
 109 }
 110
 111 //**********************************************************************************************************************
 112 void GetSeqsCommand::readFasta(){
 113         try {
 114                 string outputFileName = getRootName(fastafile) + "pick";
 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) == 1) {
 130                                 wroteSomething = true;
 131
 132                                 currSeq.printSequence(out);
 133
 134                                 names.erase(name);
 135                         }
 136
 137                         gobble(in);
 138                 }
 139                 in.close();
 140                 out.close();
 141
 142                 if (wroteSomething == false) {
 143                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 144                         remove(outputFileName.c_str());
 145                 }
 146
 147         }
 148         catch(exception& e) {
 149                 errorOut(e, "GetSeqsCommand", "readFasta");
 150                 exit(1);
 151         }
 152 }
 153
 154 //**********************************************************************************************************************
 155 void GetSeqsCommand::readName(){
 156         try {
 157
 158                 string outputFileName = getRootName(namefile) + "pick";
 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
 169                 while(!in.eof()){
 170
 171                         in >> firstCol;
 172                         in >> secondCol;
 173
 174                         vector<string> parsedNames;
 175                         //parse second column saving each name
 176                         while (secondCol.find_first_of(',') != -1) {
 177                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 178                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 179                                 parsedNames.push_back(name);
 180                         }
 181
 182                         //get name after last ,
 183                         parsedNames.push_back(secondCol);
 184
 185                         vector<string> validSecond;
 186                         for (int i = 0; i < parsedNames.size(); i++) {
 187                                 if (names.count(parsedNames[i]) == 1) {
 188                                         validSecond.push_back(parsedNames[i]);
 189                                         names.erase(parsedNames[i]);
 190                                 }
 191                         }
 192
 193
 194                         //if the name in the first column is in the set then print it and any other names in second column also in set
 195                         if (names.count(firstCol) == 1) {
 196
 197                                 wroteSomething = true;
 198
 199                                 out << firstCol << '\t';
 200
 201                                 //you know you have at least one valid second since first column is valid
 202                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 203                                 out << validSecond[validSecond.size()-1] << endl;
 204
 205                                 names.erase(firstCol);
 206
 207                         //make first name in set you come to first column and then add the remaining names to second column
 208                         }else {
 209                                 //you want part of this row
 210                                 if (validSecond.size() != 0) {
 211
 212                                         wroteSomething = true;
 213
 214                                         out << validSecond[0] << '\t';
 215
 216                                         //you know you have at least one valid second since first column is valid
 217                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 218                                         out << validSecond[validSecond.size()-1] << endl;
 219                                 }
 220                         }
 221
 222                         gobble(in);
 223                 }
 224                 in.close();
 225                 out.close();
 226
 227                 if (wroteSomething == false) {
 228                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 229                         remove(outputFileName.c_str());
 230                 }
 231
 232         }
 233         catch(exception& e) {
 234                 errorOut(e, "GetSeqsCommand", "readName");
 235                 exit(1);
 236         }
 237 }
 238
 239 //**********************************************************************************************************************
 240 void GetSeqsCommand::readGroup(){
 241         try {
 242
 243                 string outputFileName = getRootName(groupfile) + "pick";
 244                 ofstream out;
 245                 openOutputFile(outputFileName, out);
 246
 247                 ifstream in;
 248                 openInputFile(groupfile, in);
 249                 string name, group;
 250
 251                 bool wroteSomething = false;
 252
 253                 while(!in.eof()){
 254
 255                         in >> name;                             //read from first column
 256                         in >> group;                    //read from second column
 257
 258                         //if this name is in the accnos file
 259                         if (names.count(name) == 1) {
 260                                 wroteSomething = true;
 261
 262                                 out << name << '\t' << group << endl;
 263
 264                                 names.erase(name);
 265                         }
 266
 267                         gobble(in);
 268                 }
 269                 in.close();
 270                 out.close();
 271
 272                 if (wroteSomething == false) {
 273                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 274                         remove(outputFileName.c_str());
 275                 }
 276
 277         }
 278         catch(exception& e) {
 279                 errorOut(e, "GetSeqsCommand", "readGroup");
 280                 exit(1);
 281         }
 282 }
 283
 284 //**********************************************************************************************************************
 285 //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
 286 void GetSeqsCommand::readAlign(){
 287         try {
 288                 string outputFileName = getRootName(alignfile) + "pick";
 289                 ofstream out;
 290                 openOutputFile(outputFileName, out);
 291
 292                 ifstream in;
 293                 openInputFile(alignfile, in);
 294                 string name, junk;
 295
 296                 bool wroteSomething = false;
 297
 298                 //read column headers
 299                 for (int i = 0; i < 16; i++) {
 300                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 301                         else                    {       break;                  }
 302                 }
 303                 out << endl;
 304
 305                 while(!in.eof()){
 306
 307                         in >> name;                             //read from first column
 308
 309                         //if this name is in the accnos file
 310                         if (names.count(name) == 1) {
 311                                 wroteSomething = true;
 312
 313                                 out << name << '\t';
 314
 315                                 //read rest
 316                                 for (int i = 0; i < 15; i++) {
 317                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 318                                         else                    {       break;                  }
 319                                 }
 320                                 out << endl;
 321
 322                                 names.erase(name);
 323
 324                         }else {//still read just don't do anything with it
 325                                 //read rest
 326                                 for (int i = 0; i < 15; i++) {
 327                                         if (!in.eof())  {       in >> junk;             }
 328                                         else                    {       break;                  }
 329                                 }
 330                         }
 331
 332                         gobble(in);
 333                 }
 334                 in.close();
 335                 out.close();
 336
 337                 if (wroteSomething == false) {
 338                         mothurOut("Your file does not contain any sequence from the .accnos file."); mothurOutEndLine();
 339                         remove(outputFileName.c_str());
 340                 }
 341
 342         }
 343         catch(exception& e) {
 344                 errorOut(e, "GetSeqsCommand", "readAlign");
 345                 exit(1);
 346         }
 347 }
 348 //**********************************************************************************************************************
 349 void GetSeqsCommand::readAccnos(){
 350         try {
 351
 352                 ifstream in;
 353                 openInputFile(accnosfile, in);
 354                 string name;
 355
 356                 while(!in.eof()){
 357                         in >> name;
 358
 359                         names.insert(name);
 360
 361                         gobble(in);
 362                 }
 363                 in.close();
 364
 365         }
 366         catch(exception& e) {
 367                 errorOut(e, "GetSeqsCommand", "readAccnos");
 368                 exit(1);
 369         }
 370 }
 371
 372 //**********************************************************************************************************************
 373