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                 int c;
  18                 string temp;
  19
  20
  21                 //read through file
  22                 while ((c = in.get()) != EOF) {
  23                         name = ""; sequence = "";
  24                         //is this a name
  25                         if (c == '>') {
  26                                 name = readName(in);
  27                                 sequence = readSequence(in);
  28                         }else {  cout << "Error fasta in your file. Please correct." << endl; }
  29
  30                         //store info in map
  31                         //input sequence info into map
  32                         seqmap[name] = sequence;
  33                         it = data.find(sequence);
  34                         if (it == data.end()) {         //it's unique.
  35                                 data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
  36                                 data[sequence].groupnumber = 1;
  37                                 data[sequence].names = name;
  38                         }else { // its a duplicate.
  39                                 data[sequence].names += "," + name;
  40                                 data[sequence].groupnumber++;
  41                         }
  42
  43                         gobble(in);
  44                 }
  45
  46         }
  47         catch(exception& e) {
  48                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  49                 exit(1);
  50         }
  51         catch(...) {
  52                 cout << "An unknown error has occurred in the FastaMap class function readFastaFile. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  53                 exit(1);
  54         }
  55 }
  56 /*******************************************************************************/
  57 string FastaMap::readName(ifstream& in) {
  58         try{
  59                 string name = "";
  60                 int c;
  61                 string temp;
  62
  63                 while ((c = in.get()) != EOF) {
  64                         //if c is not a line return
  65                         if (c != 10) {
  66                                 name += c;
  67                         }else { break;  }
  68                 }
  69
  70                 return name;
  71         }
  72         catch(exception& e) {
  73                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readName. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  74                 exit(1);
  75         }
  76         catch(...) {
  77                 cout << "An unknown error has occurred in the FastaMap class function readName. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
  78                 exit(1);
  79         }
  80 }
  81
  82 /*******************************************************************************/
  83 string FastaMap::readSequence(ifstream& in) {
  84         try{
  85                 string sequence = "";
  86                 string line;
  87                 int pos, c;
  88
  89                 while (!in.eof()) {
  90                         //save position in file in case next line is a new name.
  91                         pos = in.tellg();
  92                         line = "";
  93                         in >> line;
  94                         //if you are at a new name
  95                         if (line[0] == '>') {
  96                                 //put file pointer back since you are now at a new name
  97                                 in.seekg(pos, ios::beg);
  98                                 c = in.get();  //because you put it back to a newline char
  99                                 break;
 100                         }else {  sequence += line;      }
 101                 }
 102
 103                 return sequence;
 104         }
 105         catch(exception& e) {
 106                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function readSequence. 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 readSequence. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 111                 exit(1);
 112         }
 113 }
 114 /*******************************************************************************/
 115 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
 116         return data[seq].groupname;
 117 }
 118 /*******************************************************************************/
 119 string FastaMap::getNames(string seq) { //pass a sequence get the string of names in the group separated by ','s.
 120         return data[seq].names;
 121 }
 122 /*******************************************************************************/
 123 int FastaMap::getGroupNumber(string seq) {      //pass a sequence get the number of identical sequences.
 124         return data[seq].groupnumber;
 125 }
 126 /*******************************************************************************/
 127 string FastaMap::getSequence(string name) {
 128         it2 = seqmap.find(name);
 129         if (it2 == seqmap.end()) {      //it's not found
 130                 return "not found";
 131         }else { // found it
 132                 return it2->second;
 133         }
 134 }
 135 /*******************************************************************************/
 136 void FastaMap::push_back(string name, string seq) {
 137         it = data.find(seq);
 138         if (it == data.end()) {         //it's unique.
 139                 data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
 140                 data[seq].groupnumber = 1;
 141                 data[seq].names = name;
 142         }else { // its a duplicate.
 143                 data[seq].names += "," + name;
 144                 data[seq].groupnumber++;
 145         }
 146
 147         seqmap[name] = seq;
 148 }
 149 /*******************************************************************************/
 150 int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 151         return data.size();
 152 }
 153 /*******************************************************************************/
 154 void FastaMap::printNamesFile(ostream& out){ //prints data
 155         try {
 156                 // two column file created with groupname and them list of identical sequence names
 157                 for (it = data.begin(); it != data.end(); it++) {
 158                         out << it->second.groupname << '\t' << it->second.names << endl;
 159                 }
 160         }
 161         catch(exception& e) {
 162                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 163                 exit(1);
 164         }
 165         catch(...) {
 166                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 167                 exit(1);
 168         }
 169 }
 170 /*******************************************************************************/
 171 void FastaMap::printCondensedFasta(ostream& out){ //prints data
 172         try {
 173                 //creates a fasta file
 174                 for (it = data.begin(); it != data.end(); it++) {
 175                         out << ">" << it->second.groupname << endl;
 176                         out << it->first << endl;
 177                 }
 178         }
 179         catch(exception& e) {
 180                 cout << "Standard Error: " << e.what() << " has occurred in the FastaMap class Function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 181                 exit(1);
 182         }
 183         catch(...) {
 184                 cout << "An unknown error has occurred in the FastaMap class function print. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
 185                 exit(1);
 186         }
 187 }
 188 /*******************************************************************************/
 189