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