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
  12 /*******************************************************************************/
  13 void FastaMap::readFastaFile(ifstream& in) {
  14         try {
  15                 string name, sequence, line;
  16                 sequence = "";
  17
  18                 in >> line;
  19                 name = line.substr(1, line.length());  //rips off '>'
  20
  21                 //read through file
  22                 while (!in.eof()) {
  23                         in >> line;
  24
  25                         if (line[0] != '>') {  //if it's a sequence line
  26                                 sequence += line;
  27                         }
  28                         else{
  29                         //input sequence info into map
  30                                 seqmap[name] = sequence;
  31
  32                                 it = data.find(sequence);
  33                                 if (it == data.end()) {         //it's unique.
  34                                         data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  35                                         data[sequence].groupnumber = 1;
  36                                         data[sequence].names = name;
  37                                 }else { // its a duplicate.
  38                                         data[sequence].names += "," + name;
  39                                         data[sequence].groupnumber++;
  40                                 }
  41                                 name = (line.substr(1, (line.npos))); //The line you just read is a new name so rip off '>'
  42                                 sequence = "";
  43                         }
  44
  45                         gobble(in);
  46                 }
  47                 it = data.find(sequence);
  48                 if (it == data.end()) {         //it's unique.
  49                         data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  50                         data[sequence].groupnumber = 1;
  51                         data[sequence].names = name;
  52                 }else { // its a duplicate.
  53                         data[sequence].names += "," + name;
  54                         data[sequence].groupnumber++;
  55                 }
  56
  57
  58         }
  59         catch(exception& e) {
  60                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  61                 exit(1);
  62         }
  63         catch(...) {
  64                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  65                 exit(1);
  66         }
  67 }
  68 /*******************************************************************************/
  69 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
  70         return data[seq].groupname;
  71 }
  72 /*******************************************************************************/
  73 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
  74         return data[seq].names;
  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 string FastaMap::getSequence(string name) {
  82         it2 = seqmap.find(name);
  83         if (it2 == seqmap.end()) {      //it's not found
  84                 return "not found";
  85         }else { // found it
  86                 return it2->second;
  87         }
  88 }
  89 /*******************************************************************************/
  90 void FastaMap::push_back(string name, string seq) {
  91         it = data.find(seq);
  92         if (it == data.end()) {         //it's unique.
  93                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
  94                 data[seq].groupnumber = 1;
  95                 data[seq].names = name;
  96         }else { // its a duplicate.
  97                 data[seq].names += "," + name;
  98                 data[seq].groupnumber++;
  99         }
 100
 101         seqmap[name] = seq;
 102 }
 103 /*******************************************************************************/
 104 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 105         return data.size();
 106 }
 107 /*******************************************************************************/
 108 void FastaMap::printNamesFile(ostream& out){ //prints data
 109         try {
 110                 // two column file created with groupname and them list of identical sequence names
 111                 for (it = data.begin(); it != data.end(); it++) {
 112                         out << it->second.groupname << '\t' << it->second.names << endl;
 113                 }
 114         }
 115         catch(exception& e) {
 116                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 117                 exit(1);
 118         }
 119         catch(...) {
 120                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 121                 exit(1);
 122         }
 123 }
 124 /*******************************************************************************/
 125 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 126         try {
 127                 // two column file created with groupname and them list of identical sequence names
 128                 for (it = data.begin(); it != data.end(); it++) {
 129                         out << ">" << it->second.groupname << endl;
 130                         out << it->first << endl;
 131                 }
 132         }
 133         catch(exception& e) {
 134                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 135                 exit(1);
 136         }
 137         catch(...) {
 138                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 139                 exit(1);
 140         }
 141 }
 142 /*******************************************************************************/
 143