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                         if (line != "") {
  25                                 if (isalnum(line.at(0))) {  //if it's a sequence line
  26                                         sequence += line;
  27                                 }
  28                                 else{
  29                                 //input sequence info into map
  30                                         seqmap[name] = sequence;
  31                                         it = data.find(sequence);
  32                                         if (it == data.end()) {         //it's unique.
  33                                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  34                                                 data[sequence].groupnumber = 1;
  35                                                 data[sequence].names = name;
  36                                         }else { // its a duplicate.
  37                                                 data[sequence].names += "," + name;
  38                                                 data[sequence].groupnumber++;
  39                                         }
  40                                         name = (line.substr(1, (line.npos))); //The line you just read is a new name so rip off '>'
  41                                         sequence = "";
  42                                 }
  43                         }
  44                         gobble(in);
  45                 }
  46
  47
  48         }
  49         catch(exception& e) {
  50                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  51                 exit(1);
  52         }
  53         catch(...) {
  54                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  55                 exit(1);
  56         }
  57 }
  58 /*******************************************************************************/
  59 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
  60         return data[seq].groupname;
  61 }
  62 /*******************************************************************************/
  63 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
  64         return data[seq].names;
  65 }
  66 /*******************************************************************************/
  67 int FastaMap::getGroupNumber(string seq) {      //pass a sequence get the number of identical sequences.
  68         return data[seq].groupnumber;
  69 }
  70 /*******************************************************************************/
  71 string FastaMap::getSequence(string name) {
  72         it2 = seqmap.find(name);
  73         if (it2 == seqmap.end()) {      //it's not found
  74                 return "not found";
  75         }else { // found it
  76                 return it2->second;
  77         }
  78 }
  79 /*******************************************************************************/
  80 void FastaMap::push_back(string name, string seq) {
  81         it = data.find(seq);
  82         if (it == data.end()) {         //it's unique.
  83                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
  84                 data[seq].groupnumber = 1;
  85                 data[seq].names = name;
  86         }else { // its a duplicate.
  87                 data[seq].names += "," + name;
  88                 data[seq].groupnumber++;
  89         }
  90
  91         seqmap[name] = seq;
  92 }
  93 /*******************************************************************************/
  94 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
  95         return data.size();
  96 }
  97 /*******************************************************************************/
  98 void FastaMap::printNamesFile(ostream& out){ //prints data
  99         try {
 100                 // two column file created with groupname and them list of identical sequence names
 101                 for (it = data.begin(); it != data.end(); it++) {
 102                         out << it->second.groupname << '\t' << it->second.names << endl;
 103                 }
 104         }
 105         catch(exception& e) {
 106                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 107                 exit(1);
 108         }
 109         catch(...) {
 110                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 111                 exit(1);
 112         }
 113 }
 114 /*******************************************************************************/
 115 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 116         try {
 117                 // two column file created with groupname and them list of identical sequence names
 118                 for (it = data.begin(); it != data.end(); it++) {
 119                         out << ">" << it->second.groupname << endl;
 120                         out << it->first << 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