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(ifstream& in) {
  16         try {
  17                 string name, sequence, line;
  18                 sequence = "";
  19 //              int c;
  20                 string temp;
  21
  22
  23                 //read through file
  24 //              while ((c = in.get()) != EOF) {
  25 //                      name = ""; sequence = "";
  26 //                      //is this a name
  27 //                      if (c == '>') {
  28 //                              name = readName(in);
  29 //                              sequence = readSequence(in);
  30 //                      }else {  cout << "Error fasta in your file. Please correct." << endl; }
  31
  32                         //store info in map
  33                         //input sequence info into map
  34                 while(!in.eof()){
  35                         Sequence currSeq(in);
  36                         name = currSeq.getName();
  37                         sequence = currSeq.getUnaligned();
  38                         seqmap[name] = sequence;
  39                         it = data.find(sequence);
  40                         if (it == data.end()) {         //it's unique.
  41                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  42                                 data[sequence].groupnumber = 1;
  43                                 data[sequence].names = name;
  44                         }else { // its a duplicate.
  45                                 data[sequence].names += "," + name;
  46                                 data[sequence].groupnumber++;
  47                         }
  48
  49                         gobble(in);
  50                 }
  51
  52         }
  53         catch(exception& e) {
  54                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  55                 exit(1);
  56         }
  57         catch(...) {
  58                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  59                 exit(1);
  60         }
  61 }
  62
  63 /*******************************************************************************/
  64
  65 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
  66         return data[seq].groupname;
  67 }
  68
  69 /*******************************************************************************/
  70
  71 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
  72         return data[seq].names;
  73 }
  74
  75 /*******************************************************************************/
  76
  77 int FastaMap::getGroupNumber(string seq) {      //pass a sequence get the number of identical sequences.
  78         return data[seq].groupnumber;
  79 }
  80
  81 /*******************************************************************************/
  82
  83 string FastaMap::getSequence(string name) {
  84         it2 = seqmap.find(name);
  85         if (it2 == seqmap.end()) {      //it's not found
  86                 return "not found";
  87         }else { // found it
  88                 return it2->second;
  89         }
  90 }
  91
  92 /*******************************************************************************/
  93
  94 void FastaMap::push_back(string name, string seq) {
  95         it = data.find(seq);
  96         if (it == data.end()) {         //it's unique.
  97                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
  98                 data[seq].groupnumber = 1;
  99                 data[seq].names = name;
 100         }else { // its a duplicate.
 101                 data[seq].names += "," + name;
 102                 data[seq].groupnumber++;
 103         }
 104
 105         seqmap[name] = seq;
 106 }
 107
 108 /*******************************************************************************/
 109
 110 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 111         return data.size();
 112 }
 113
 114 /*******************************************************************************/
 115
 116 void FastaMap::printNamesFile(ostream& out){ //prints data
 117         try {
 118                 // two column file created with groupname and them list of identical sequence names
 119                 for (it = data.begin(); it != data.end(); it++) {
 120                         out << it->second.groupname << '\t' << it->second.names << endl;
 121                 }
 122         }
 123         catch(exception& e) {
 124                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 125                 exit(1);
 126         }
 127         catch(...) {
 128                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 129                 exit(1);
 130         }
 131 }
 132
 133 /*******************************************************************************/
 134
 135 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 136         try {
 137                 //creates a fasta file
 138                 for (it = data.begin(); it != data.end(); it++) {
 139                         out << ">" << it->second.groupname << endl;
 140                         out << it->first << endl;
 141                 }
 142         }
 143         catch(exception& e) {
 144                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 145                 exit(1);
 146         }
 147         catch(...) {
 148                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 149                 exit(1);
 150         }
 151 }
 152
 153 /*******************************************************************************/
 154