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                 while(!in.eof()){
 123                         Sequence currSeq(in);
 124                         name = currSeq.getName();
 125
 126                         //if this name is in the accnos file
 127                         if (names.count(name) == 1) {
 128                                 currSeq.printSequence(out);
 129                         }
 130
 131                         gobble(in);
 132                 }
 133                 in.close();
 134                 out.close();
 135
 136         }
 137         catch(exception& e) {
 138                 errorOut(e, "GetSeqsCommand", "readFasta");
 139                 exit(1);
 140         }
 141 }
 142
 143 //**********************************************************************************************************************
 144 void GetSeqsCommand::readName(){
 145         try {
 146
 147                 string outputFileName = getRootName(namefile) + "pick";
 148                 ofstream out;
 149                 openOutputFile(outputFileName, out);
 150
 151                 ifstream in;
 152                 openInputFile(namefile, in);
 153                 string name, firstCol, secondCol;
 154
 155
 156                 while(!in.eof()){
 157
 158                         in >> firstCol;
 159                         in >> secondCol;
 160
 161                         vector<string> parsedNames;
 162                         //parse second column saving each name
 163                         while (secondCol.find_first_of(',') != -1) {
 164                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 165                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 166                                 parsedNames.push_back(name);
 167                         }
 168
 169                         //get name after last ,
 170                         parsedNames.push_back(secondCol);
 171
 172                         vector<string> validSecond;
 173                         for (int i = 0; i < parsedNames.size(); i++) {
 174                                 if (names.count(parsedNames[i]) == 1) {
 175                                         validSecond.push_back(parsedNames[i]);
 176                                 }
 177                         }
 178
 179
 180                         //if the name in the first column is in the set then print it and any other names in second column also in set
 181                         if (names.count(firstCol) == 1) {
 182                                 out << firstCol << '\t';
 183
 184                                 //you know you have at least one valid second since first column is valid
 185                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 186                                 out << validSecond[validSecond.size()-1] << endl;
 187
 188                         //make first name in set you come to first column and then add the remaining names to second column
 189                         }else {
 190                                 //you want part of this row
 191                                 if (validSecond.size() != 0) {
 192                                         out << validSecond[0] << '\t';
 193
 194                                         //you know you have at least one valid second since first column is valid
 195                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 196                                         out << validSecond[validSecond.size()-1] << endl;
 197                                 }
 198                         }
 199
 200                         gobble(in);
 201                 }
 202                 in.close();
 203                 out.close();
 204
 205         }
 206         catch(exception& e) {
 207                 errorOut(e, "GetSeqsCommand", "readName");
 208                 exit(1);
 209         }
 210 }
 211
 212 //**********************************************************************************************************************
 213 void GetSeqsCommand::readGroup(){
 214         try {
 215
 216                 string outputFileName = getRootName(groupfile) + "pick";
 217                 ofstream out;
 218                 openOutputFile(outputFileName, out);
 219
 220                 ifstream in;
 221                 openInputFile(groupfile, in);
 222                 string name, group;
 223
 224                 while(!in.eof()){
 225
 226                         in >> name;                             //read from first column
 227                         in >> group;                    //read from second column
 228
 229                         //if this name is in the accnos file
 230                         if (names.count(name) == 1) {
 231                                 out << name << '\t' << group << endl;
 232                         }
 233
 234                         gobble(in);
 235                 }
 236                 in.close();
 237                 out.close();
 238
 239         }
 240         catch(exception& e) {
 241                 errorOut(e, "GetSeqsCommand", "readGroup");
 242                 exit(1);
 243         }
 244 }
 245
 246 //**********************************************************************************************************************
 247 //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
 248 void GetSeqsCommand::readAlign(){
 249         try {
 250                 string outputFileName = getRootName(alignfile) + "pick";
 251                 ofstream out;
 252                 openOutputFile(outputFileName, out);
 253
 254                 ifstream in;
 255                 openInputFile(alignfile, in);
 256                 string name, junk;
 257
 258                 //read column headers
 259                 for (int i = 0; i < 16; i++) {
 260                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 261                         else                    {       break;                  }
 262                 }
 263                 out << endl;
 264
 265                 while(!in.eof()){
 266
 267                         in >> name;                             //read from first column
 268
 269                         //if this name is in the accnos file
 270                         if (names.count(name) == 1) {
 271
 272                                 out << name << '\t';
 273
 274                                 //read rest
 275                                 for (int i = 0; i < 15; i++) {
 276                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 277                                         else                    {       break;                  }
 278                                 }
 279                                 out << endl;
 280
 281                         }else {//still read just don't do anything with it
 282                                 //read rest
 283                                 for (int i = 0; i < 15; i++) {
 284                                         if (!in.eof())  {       in >> junk;             }
 285                                         else                    {       break;                  }
 286                                 }
 287                         }
 288
 289                         gobble(in);
 290                 }
 291                 in.close();
 292                 out.close();
 293
 294
 295         }
 296         catch(exception& e) {
 297                 errorOut(e, "GetSeqsCommand", "readAlign");
 298                 exit(1);
 299         }
 300 }
 301 //**********************************************************************************************************************
 302 void GetSeqsCommand::readAccnos(){
 303         try {
 304
 305                 ifstream in;
 306                 openInputFile(accnosfile, in);
 307                 string name;
 308
 309                 while(!in.eof()){
 310                         in >> name;
 311
 312                         names.insert(name);
 313
 314                         gobble(in);
 315                 }
 316                 in.close();
 317
 318         }
 319         catch(exception& e) {
 320                 errorOut(e, "GetSeqsCommand", "readAccnos");
 321                 exit(1);
 322         }
 323 }
 324
 325 //**********************************************************************************************************************
 326