modified sequence class to read fasta files with comments. this required modification...

[mothur.git] / sequence.cpp

diff --git a/sequence.cpp b/sequence.cpp
index 98aa7b01862c494018c710ec607f5132d8ce2db1..752e081e367fc6d572b5622e0b4eb6003a28f9bb 100644 (file)
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -7,8 +7,6 @@
  *
  */
 
-using namespace std;
-
 #include "sequence.hpp"
 
 /***********************************************************************/
@@ -23,42 +21,91 @@ Sequence::Sequence(string newName, string sequence) {
 
        initialize();   
        name = newName;
-       if(sequence.find_first_of('-') != string::npos) {
-               setAligned(sequence);
-       }
+       
+       //setUnaligned removes any gap characters for us
        setUnaligned(sequence);
+       setAligned(sequence);
        
 }
 //********************************************************************************************************************
-
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
 Sequence::Sequence(ifstream& fastaFile){
-       initialize();
-       
-       string accession;                               //      provided a file handle to a fasta-formatted sequence file, read in the next
-       fastaFile >> accession;                 //      accession number and sequence we find...
-       setName(accession);
 
-       char letter;
+       initialize();
+       fastaFile >> name;
+       name = name.substr(1);
        string sequence;
        
-       while(fastaFile){
-               letter= fastaFile.get();
-               if(letter == '>'){
-                       fastaFile.putback(letter);
+       //read comments
+       while ((name[0] == '#') && fastaFile) { 
+           while (!fastaFile.eof())    {       char c = fastaFile.get(); if (c == 10 || c == 13){      break;  }       } // get rest of line if there's any crap there
+               sequence = getCommentString(fastaFile);
+               
+               if (fastaFile) {  
+                       fastaFile >> name;  
+                       name = name.substr(1);  
+               }else { 
+                       name = "";
                        break;
                }
-               else if(isprint(letter)){
-                       letter = toupper(letter);
-                       if(letter == 'U'){letter = 'T';}
-                       sequence += letter;
+       }
+       
+       //read real sequence
+       while (!fastaFile.eof())        {       char c = fastaFile.get(); if (c == 10 || c == 13){      break;  }       } // get rest of line if there's any crap there
+               
+       sequence = getSequenceString(fastaFile);                
+       
+       setAligned(sequence);   
+       //setUnaligned removes any gap characters for us                                                
+       setUnaligned(sequence);                                                         
+}
+//********************************************************************************************************************
+string Sequence::getSequenceString(ifstream& fastaFile) {
+       try {
+               char letter;
+               string sequence = "";   
+               
+               while(fastaFile){
+                       letter= fastaFile.get();
+                       if(letter == '>'){
+                               fastaFile.putback(letter);
+                               break;
+                       }
+                       else if(isprint(letter)){
+                               letter = toupper(letter);
+                               if(letter == 'U'){letter = 'T';}
+                               sequence += letter;
+                       }
                }
                
+               return sequence;
        }
-
-       if(sequence.find_first_of('-') != string::npos){        //      if there are any gaps in the sequence, assume that it is
-               setAligned(sequence);                                                   //      an alignment file
+       catch(exception& e) {
+               errorOut(e, "Sequence", "getSequenceString");
+               exit(1);
+       }
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(ifstream& fastaFile) {
+       try {
+               char letter;
+               string sequence = "";
+               
+               while(fastaFile){
+                       letter=fastaFile.get();
+                       if((letter == '\r') || (letter == '\n')){  
+                               gobble(fastaFile);  //in case its a \r\n situation
+                               break;
+                       }
+               }
+               
+               return sequence;
+       }
+       catch(exception& e) {
+               errorOut(e, "Sequence", "getCommentString");
+               exit(1);
        }
-       setUnaligned(sequence);                                                         //      also set the unaligned sequence file
 }
 
 //********************************************************************************************************************
@@ -91,7 +138,7 @@ void Sequence::setName(string seqName) {
 
 void Sequence::setUnaligned(string sequence){
        
-       if(sequence.find_first_of('-') != string::npos) {
+       if(sequence.find_first_of('.') != string::npos || sequence.find_first_of('-') != string::npos) {
                string temp = "";
                for(int j=0;j<sequence.length();j++) {
                        if(isalpha(sequence[j]))        {       temp += sequence[j];    }
@@ -112,6 +159,7 @@ void Sequence::setAligned(string sequence){
        //if the alignment starts or ends with a gap, replace it with a period to indicate missing data
        aligned = sequence;
        alignmentLength = aligned.length();
+       setUnaligned(sequence); 
 
        if(aligned[0] == '-'){
                for(int i=0;i<alignmentLength;i++){
@@ -294,6 +342,7 @@ void Sequence::reverseComplement(){
                else                                            {       temp += 'N';    }
        }
        unaligned = temp;
+       aligned = temp;
        
 }