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                 getline(in, line);
  19                 name = line.substr(1, line.length());  //rips off '>'
  20
  21                 //read through file
  22                 while (getline(in, line)) {
  23                         if (line != "") {
  24                                 if (isalnum(line.at(0))) {  //if it's a sequence line
  25                                         sequence += line;
  26                                 }
  27                                 else{
  28                                 //input sequence info into map
  29                                         it = data.find(sequence);
  30                                         if (it == data.end()) {         //it's unique.
  31                                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  32                                                 data[sequence].groupnumber = 1;
  33                                                 data[sequence].names = name;
  34                                         }else { // its a duplicate.
  35                                                 data[sequence].names += "," + name;
  36                                                 data[sequence].groupnumber++;
  37                                         }
  38                                         name = (line.substr(1, (line.npos))); //The line you just read is a new name so rip off '>'
  39                                         sequence = "";
  40                                 }
  41                         }
  42                 }
  43
  44                 //store last sequence and name info.
  45                 it = data.find(sequence);
  46                 if (it == data.end()) {         //it's unique.
  47                         data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  48                         data[sequence].groupnumber = 1;
  49                         data[sequence].names = name;
  50                 }else { // its a duplicate.
  51                         data[sequence].names += "," + name;
  52                         data[sequence].groupnumber++;
  53                 }
  54         }
  55         catch(exception& e) {
  56                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  57                 exit(1);
  58         }
  59         catch(...) {
  60                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  61                 exit(1);
  62         }
  63 }
  64 /*******************************************************************************/
  65 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
  66         return data[seq].groupname;
  67 }
  68 /*******************************************************************************/
  69 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
  70         return data[seq].names;
  71 }
  72 /*******************************************************************************/
  73 int FastaMap::getGroupNumber(string seq) {      //pass a sequence get the number of identical sequences.
  74         return data[seq].groupnumber;
  75 }
  76 /*******************************************************************************/
  77 void FastaMap::push_back(string seq, string Name) {//sequencename, name
  78         data[seq].groupname = Name;
  79         data[seq].names = Name;
  80 }
  81 /*******************************************************************************/
  82 void FastaMap::set(string seq, string groupName, string Names) {
  83         data[seq].groupname = groupName;
  84         data[seq].names = Names;
  85 }
  86 /*******************************************************************************/
  87 void FastaMap::clear() { //clears out data
  88         data.clear();
  89 }
  90 /*******************************************************************************/
  91 int FastaMap::size(){ //returns datas size which is the number of unique sequences
  92         return data.size();
  93 }
  94 /*******************************************************************************/
  95 void FastaMap::print(ostream& out){ //prints data
  96         try {
  97                 // two column file created with groupname and them list of identical sequence names
  98                 for (it = data.begin(); it != data.end(); it++) {
  99                         out << it->second.groupname << '\t' << it->second.names << endl;
 100                 }
 101         }
 102         catch(exception& e) {
 103                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 104                 exit(1);
 105         }
 106         catch(...) {
 107                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 108                 exit(1);
 109         }
 110 }
 111 /*******************************************************************************/
 112 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 113         try {
 114                 // two column file created with groupname and them list of identical sequence names
 115                 for (it = data.begin(); it != data.end(); it++) {
 116                         out << ">" << it->second.groupname << endl;
 117                         out << it->first << endl;
 118                 }
 119         }
 120         catch(exception& e) {
 121                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 122                 exit(1);
 123         }
 124         catch(...) {
 125                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 126                 exit(1);
 127         }
 128 }
 129 /*******************************************************************************/
 130