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                                         seqmap[name] = sequence;
  30                                         it = data.find(sequence);
  31                                         if (it == data.end()) {         //it's unique.
  32                                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  33                                                 data[sequence].groupnumber = 1;
  34                                                 data[sequence].names = name;
  35                                         }else { // its a duplicate.
  36                                                 data[sequence].names += "," + name;
  37                                                 data[sequence].groupnumber++;
  38                                         }
  39                                         name = (line.substr(1, (line.npos))); //The line you just read is a new name so rip off '>'
  40                                         sequence = "";
  41                                 }
  42                         }
  43                 }
  44
  45                 //store last sequence and name info.
  46                 seqmap[name] = sequence;
  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         catch(exception& e) {
  59                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  60                 exit(1);
  61         }
  62         catch(...) {
  63                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  64                 exit(1);
  65         }
  66 }
  67 /*******************************************************************************/
  68 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
  69         return data[seq].groupname;
  70 }
  71 /*******************************************************************************/
  72 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
  73         return data[seq].names;
  74 }
  75 /*******************************************************************************/
  76 int FastaMap::getGroupNumber(string seq) {      //pass a sequence get the number of identical sequences.
  77         return data[seq].groupnumber;
  78 }
  79 /*******************************************************************************/
  80 string FastaMap::getSequence(string name) {
  81         it2 = seqmap.find(name);
  82         if (it2 == seqmap.end()) {      //it's not found
  83                 return "not found";
  84         }else { // found it
  85                 return it2->second;
  86         }
  87 }
  88 /*******************************************************************************/
  89 void FastaMap::push_back(string name, string seq) {
  90         it = data.find(seq);
  91         if (it == data.end()) {         //it's unique.
  92                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
  93                 data[seq].groupnumber = 1;
  94                 data[seq].names = name;
  95         }else { // its a duplicate.
  96                 data[seq].names += "," + name;
  97                 data[seq].groupnumber++;
  98         }
  99
 100         seqmap[name] = seq;
 101 }
 102 /*******************************************************************************/
 103 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 104         return data.size();
 105 }
 106 /*******************************************************************************/
 107 void FastaMap::printNamesFile(ostream& out){ //prints data
 108         try {
 109                 // two column file created with groupname and them list of identical sequence names
 110                 for (it = data.begin(); it != data.end(); it++) {
 111                         out << it->second.groupname << '\t' << it->second.names << endl;
 112                 }
 113         }
 114         catch(exception& e) {
 115                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 116                 exit(1);
 117         }
 118         catch(...) {
 119                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 120                 exit(1);
 121         }
 122 }
 123 /*******************************************************************************/
 124 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 125         try {
 126                 // two column file created with groupname and them list of identical sequence names
 127                 for (it = data.begin(); it != data.end(); it++) {
 128                         out << ">" << it->second.groupname << endl;
 129                         out << it->first << endl;
 130                 }
 131         }
 132         catch(exception& e) {
 133                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 134                 exit(1);
 135         }
 136         catch(...) {
 137                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 138                 exit(1);
 139         }
 140 }
 141 /*******************************************************************************/
 142