fastamap.cpp

   1 /*
   2  *  fastamap.cpp
   3  *  mothur
   4  *
   5  *  Created by Sarah Westcott on 1/16/09.
   6  *  Copyright 2009 Schloss Lab UMASS AMherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "fastamap.h"
  11 #include "sequence.hpp"
  12
  13 /*******************************************************************************/
  14
  15 void FastaMap::readFastaFile(string inFileName) {
  16         try {
  17                 ifstream in;
  18                 openInputFile(inFileName, in);
  19                 string name, sequence, line;
  20                 sequence = "";
  21                 string temp;
  22
  23                 while(!in.eof()){
  24                         Sequence currSeq(in);
  25                         name = currSeq.getName();
  26
  27                         if(currSeq.getIsAligned())      {       sequence = currSeq.getAligned();        }
  28                         else                                            {       sequence = currSeq.getUnaligned();      }
  29
  30                         seqmap[name] = sequence;
  31                         map<string,group>::iterator it = data.find(sequence);
  32                         if (it == data.end()) {         //it's unique.
  33                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  34 //                              data[sequence].groupnumber = 1;
  35                                 data[sequence].names = name;
  36                         }else { // its a duplicate.
  37                                 data[sequence].names += "," + name;
  38 //                              data[sequence].groupnumber++;
  39                         }
  40
  41                         gobble(in);
  42                 }
  43                 in.close();
  44         }
  45         catch(exception& e) {
  46                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  47                 exit(1);
  48         }
  49         catch(...) {
  50                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  51                 exit(1);
  52         }
  53 }
  54
  55 /*******************************************************************************/
  56
  57 void FastaMap::readFastaFile(string inFastaFile, string oldNameFileName){ //prints data
  58
  59         ifstream oldNameFile;
  60         openInputFile(oldNameFileName, oldNameFile);
  61
  62         map<string,string> oldNameMap;
  63         string name, list;
  64         while(!oldNameFile.eof()){
  65                 oldNameFile >> name >> list;
  66                 oldNameMap[name] = list;
  67                 gobble(oldNameFile);
  68         }
  69         oldNameFile.close();
  70
  71         ifstream inFASTA;
  72         openInputFile(inFastaFile, inFASTA);
  73         string sequence;
  74         while(!inFASTA.eof()){
  75                 Sequence currSeq(inFASTA);
  76                 name = currSeq.getName();
  77
  78                 if(currSeq.getIsAligned())      {       sequence = currSeq.getAligned();        }
  79                 else                                            {       sequence = currSeq.getUnaligned();      }
  80
  81                 seqmap[name] = sequence;
  82                 map<string,group>::iterator it = data.find(sequence);
  83                 if (it == data.end()) {         //it's unique.
  84                         data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  85 //                      data[sequence].groupnumber = 1;
  86                         data[sequence].names = oldNameMap[name];
  87                 }else { // its a duplicate.
  88                         data[sequence].names += "," + oldNameMap[name];
  89 //                      data[sequence].groupnumber++;
  90                 }
  91
  92                 gobble(inFASTA);
  93         }
  94
  95
  96         inFASTA.close();
  97 }
  98
  99 /*******************************************************************************/
 100
 101 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
 102         return data[seq].groupname;
 103 }
 104
 105 /*******************************************************************************/
 106
 107 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
 108         return data[seq].names;
 109 }
 110
 111 /*******************************************************************************/
 112
 113 string FastaMap::getSequence(string name) {
 114
 115         map<string,string>::iterator it = seqmap.find(name);
 116         if (it == seqmap.end()) {       return "not found";             }
 117         else                                    {       return it->second;              }
 118
 119 }
 120
 121 /*******************************************************************************/
 122
 123 void FastaMap::push_back(string name, string seq) {
 124
 125         map<string,group>::iterator it = data.find(seq);
 126         if (it == data.end()) {         //it's unique.
 127                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
 128                 data[seq].names = name;
 129         }else { // its a duplicate.
 130                 data[seq].names += "," + name;
 131         }
 132         seqmap[name] = seq;
 133 }
 134
 135 /*******************************************************************************/
 136
 137 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 138         return data.size();
 139 }
 140
 141 /*******************************************************************************/
 142
 143 void FastaMap::printNamesFile(string outFileName){ //prints data
 144         try {
 145                 ofstream outFile;
 146                 openOutputFile(outFileName, outFile);
 147
 148                 // two column file created with groupname and them list of identical sequence names
 149                 for (map<string,group>::iterator it = data.begin(); it != data.end(); it++) {
 150                         outFile << it->second.groupname << '\t' << it->second.names << endl;
 151                 }
 152                 outFile.close();
 153         }
 154         catch(exception& e) {
 155                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 156                 exit(1);
 157         }
 158         catch(...) {
 159                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 160                 exit(1);
 161         }
 162 }
 163
 164 /*******************************************************************************/
 165
 166 void FastaMap::printCondensedFasta(string outFileName){ //prints data
 167         try {
 168                 ofstream out;
 169                 openOutputFile(outFileName, out);
 170                 //creates a fasta file
 171                 for (map<string,group>::iterator it = data.begin(); it != data.end(); it++) {
 172                         out << ">" << it->second.groupname << endl;
 173                         out << it->first << endl;
 174                 }
 175                 out.close();
 176         }
 177         catch(exception& e) {
 178                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 179                 exit(1);
 180         }
 181         catch(...) {
 182                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 183                 exit(1);
 184         }
 185 }
 186
 187 /*******************************************************************************/
 188