X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=752e081e367fc6d572b5622e0b4eb6003a28f9bb;hb=96b36196d49a3d1f6bc49a26a9d2aa2da7ff876e;hp=e9487c519e0d3e0378266eb537d15e28979546dc;hpb=a6c698b20eda3671d22466ab6b98b36331a30804;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index e9487c5..752e081 100644
--- 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);
-		isAligned = 1;
-	}
+	
+	//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){
-	
-	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++){
@@ -131,7 +179,7 @@ void Sequence::setAligned(string sequence){
 			}
 		}
 	}
-	
+	isAligned = 1;	
 }
 
 //********************************************************************************************************************
@@ -249,11 +297,13 @@ int Sequence::getStartPos(){
 	if(endPos == -1){
 		for(int j = 0; j < alignmentLength; j++) {
 			if(aligned[j] != '.'){
-				startPos = j;
+				startPos = j + 1;
 				break;
 			}
 		}
-	}	
+	}
+	if(isAligned == 0){	startPos = 1;	}
+
 	return startPos;
 }
 
@@ -263,11 +313,13 @@ int Sequence::getEndPos(){
 	if(endPos == -1){
 		for(int j=alignmentLength-1;j>=0;j--){
 			if(aligned[j] != '.'){
-				endPos = j;
+				endPos = j + 1;
 				break;
 			}
 		}
 	}
+	if(isAligned == 0){	endPos = numBases;	}
+	
 	return endPos;
 }
 
@@ -278,3 +330,20 @@ bool Sequence::getIsAligned(){
 }
 
 //********************************************************************************************************************
+
+void Sequence::reverseComplement(){
+
+	string temp;
+	for(int i=numBases-1;i>=0;i--){
+		if(unaligned[i] == 'A')		{	temp += 'T';	}
+		else if(unaligned[i] == 'T'){	temp += 'A';	}
+		else if(unaligned[i] == 'G'){	temp += 'C';	}
+		else if(unaligned[i] == 'C'){	temp += 'G';	}
+		else						{	temp += 'N';	}
+	}
+	unaligned = temp;
+	aligned = temp;
+	
+}
+
+//********************************************************************************************************************