sequence.cpp

   1 /*
   2  *  sequence.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 12/15/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10 using namespace std;
  11
  12 #include "sequence.hpp"
  13
  14 /***********************************************************************/
  15
  16 Sequence::Sequence()  {}
  17
  18 /***********************************************************************/
  19
  20 Sequence::Sequence(string newName, string sequence) {
  21         name = newName;
  22         if(sequence.find_first_of('-') != string::npos) {
  23                 setAligned(sequence);
  24         }
  25         setUnaligned(sequence);
  26 }
  27 //********************************************************************************************************************
  28
  29 Sequence::Sequence(ifstream& fastaFile){
  30
  31         string accession;                               //      provided a file handle to a fasta-formatted sequence file, read in the next
  32         fastaFile >> accession;                 //      accession number and sequence we find...
  33         setName(accession);
  34
  35         char letter;
  36         string sequence;
  37
  38         while(fastaFile){
  39                 letter= fastaFile.get();
  40                 if(letter == '>'){
  41                         fastaFile.putback(letter);
  42                         break;
  43                 }
  44                 else if(isprint(letter)){
  45                         letter = toupper(letter);
  46                         if(letter == 'U'){letter = 'T';}
  47                         sequence += letter;
  48                 }
  49
  50         }
  51
  52         if(sequence.find_first_of('-') != string::npos){        //      if there are any gaps in the sequence, assume that it is
  53                 setAligned(sequence);                                                   //      an alignment file
  54         }
  55         setUnaligned(sequence);                                                         //      also set the unaligned sequence file
  56 }
  57
  58 //********************************************************************************************************************
  59
  60 string Sequence::convert2ints() {
  61
  62         if(unaligned == "")     {       /* need to throw an error */    }
  63
  64         string processed;
  65
  66         for(int i=0;i<unaligned.length();i++) {
  67                 if(toupper(unaligned[i]) == 'A')                {       processed += '0';       }
  68                 else if(toupper(unaligned[i]) == 'C')   {       processed += '1';       }
  69                 else if(toupper(unaligned[i]) == 'G')   {       processed += '2';       }
  70                 else if(toupper(unaligned[i]) == 'T')   {       processed += '3';       }
  71                 else if(toupper(unaligned[i]) == 'U')   {       processed += '3';       }
  72                 else                                                                    {       processed += '4';       }
  73         }
  74         return processed;
  75 }
  76
  77 //********************************************************************************************************************
  78
  79 void Sequence::setName(string seqName) {
  80         if(seqName[0] == '>')   {       name = seqName.substr(1);       }
  81         else                                    {       name = seqName;                         }
  82 }
  83
  84 //********************************************************************************************************************
  85
  86 void Sequence::setUnaligned(string sequence){
  87
  88         if(sequence.find_first_of('-') != string::npos) {
  89                 string temp = "";
  90                 for(int j=0;j<sequence.length();j++) {
  91                         if(isalpha(sequence[j]))        {       temp += sequence[j];    }
  92                 }
  93                 unaligned = temp;
  94         }
  95         else {
  96                 unaligned = sequence;
  97         }
  98
  99 }
 100
 101 //********************************************************************************************************************
 102
 103 void Sequence::setAligned(string sequence){
 104
 105         //if the alignment starts or ends with a gap, replace it with a period to indicate missing data
 106
 107         if(sequence[0] == '-'){
 108                 for(int i=0;i<sequence.length();i++){
 109                         if(sequence[i] == '-'){
 110                                 sequence[i] = '.';
 111                         }
 112                         else{
 113                                 break;
 114                         }
 115                 }
 116                 for(int i=sequence.length()-1;i>=0;i--){
 117                         if(sequence[i] == '-'){
 118                                 sequence[i] = '.';
 119                         }
 120                         else{
 121                                 break;
 122                         }
 123                 }
 124         }
 125         aligned = sequence;
 126 }
 127
 128 //********************************************************************************************************************
 129
 130 void Sequence::setPairwise(string sequence){
 131         pairwise = sequence;
 132 }
 133
 134 //********************************************************************************************************************
 135
 136 string Sequence::getName(){
 137         return name;
 138 }
 139
 140 //********************************************************************************************************************
 141
 142 string Sequence::getAligned(){
 143         return aligned;
 144 }
 145
 146 //********************************************************************************************************************
 147
 148 string Sequence::getPairwise(){
 149         return pairwise;
 150 }
 151
 152 //********************************************************************************************************************
 153
 154 string Sequence::getUnaligned(){
 155         return unaligned;
 156 }
 157
 158 //********************************************************************************************************************
 159
 160 int Sequence::getLength(){
 161         if(unaligned.length() > aligned.length())
 162                 return unaligned.length();
 163         return aligned.length();
 164 }
 165
 166 //********************************************************************************************************************
 167
 168 void Sequence::printSequence(ostream& out){
 169         string toPrint = unaligned;
 170         if(aligned.length() > unaligned.length())
 171                 toPrint = aligned;
 172         out << ">" << name << "\n" << toPrint << "\n";
 173 }
 174
 175 //********************************************************************************************************************
 176
 177 int Sequence::getUnalignLength(){
 178         return unaligned.length();
 179 }
 180
 181 //********************************************************************************************************************
 182
 183 int Sequence::getAlignLength(){
 184         return aligned.length();
 185 }
 186
 187 //********************************************************************************************************************
 188
 189
 190