X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=98aa7b01862c494018c710ec607f5132d8ce2db1;hb=544469443afe44920bdf279aefd26d29534cabaf;hp=b59363e13cc92f969145434d1ba8c0fa64573778;hpb=d5d2761f88b41f1006d0b700e0ab51e2ce48b875;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index b59363e..98aa7b0 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -13,39 +13,75 @@ using namespace std;
 
 /***********************************************************************/
 
-Sequence::Sequence()  {}
+Sequence::Sequence(){
+	initialize();
+}
 
 /***********************************************************************/
 
 Sequence::Sequence(string newName, string sequence) {
+
+	initialize();	
 	name = newName;
 	if(sequence.find_first_of('-') != string::npos) {
 		setAligned(sequence);
 	}
 	setUnaligned(sequence);
+	
 }
-
 //********************************************************************************************************************
 
-string Sequence::convert2ints() {
-	
-	if(unaligned == "")	{	/* need to throw an error */	}
+Sequence::Sequence(ifstream& fastaFile){
+	initialize();
 	
-	string processed;
+	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;
+	string sequence;
 	
-	for(int i=0;i<unaligned.length();i++) {
-		if(toupper(unaligned[i]) == 'A')			{	processed += '0';	}
-		else if(toupper(unaligned[i]) == 'C')	{	processed += '1';	}
-		else if(toupper(unaligned[i]) == 'G')	{	processed += '2';	}
-		else if(toupper(unaligned[i]) == 'T')	{	processed += '3';	}
-		else if(toupper(unaligned[i]) == 'U')	{	processed += '3';	}
-		else									{	processed += '4';	}
+	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 processed;
+
+	if(sequence.find_first_of('-') != string::npos){	//	if there are any gaps in the sequence, assume that it is
+		setAligned(sequence);							//	an alignment file
+	}
+	setUnaligned(sequence);								//	also set the unaligned sequence file
 }
 
 //********************************************************************************************************************
 
+void Sequence::initialize(){
+	
+	name = "";
+	unaligned = "";
+	aligned = "";
+	pairwise = "";
+	
+	numBases = 0;
+	alignmentLength = 0;
+	isAligned = 0;
+	startPos = -1;
+	endPos = -1;
+	longHomoPolymer = -1;
+	ambigBases = -1;
+	
+}	
+
+//********************************************************************************************************************
+
 void Sequence::setName(string seqName) {
 	if(seqName[0] == '>')	{	name = seqName.substr(1);	}
 	else					{	name = seqName;				}
@@ -65,13 +101,37 @@ void Sequence::setUnaligned(string sequence){
 	else {
 		unaligned = sequence;
 	}
+	numBases = unaligned.length();
 	
 }
 
 //********************************************************************************************************************
 
 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();
+
+	if(aligned[0] == '-'){
+		for(int i=0;i<alignmentLength;i++){
+			if(aligned[i] == '-'){
+				aligned[i] = '.';
+			}
+			else{
+				break;
+			}
+		}
+		for(int i=alignmentLength-1;i>=0;i--){
+			if(aligned[i] == '-'){
+				aligned[i] = '.';
+			}
+			else{
+				break;
+			}
+		}
+	}
+	isAligned = 1;	
 }
 
 //********************************************************************************************************************
@@ -82,6 +142,25 @@ void Sequence::setPairwise(string sequence){
 
 //********************************************************************************************************************
 
+string Sequence::convert2ints() {
+	
+	if(unaligned == "")	{	/* need to throw an error */	}
+	
+	string processed;
+	
+	for(int i=0;i<unaligned.length();i++) {
+		if(toupper(unaligned[i]) == 'A')		{	processed += '0';	}
+		else if(toupper(unaligned[i]) == 'C')	{	processed += '1';	}
+		else if(toupper(unaligned[i]) == 'G')	{	processed += '2';	}
+		else if(toupper(unaligned[i]) == 'T')	{	processed += '3';	}
+		else if(toupper(unaligned[i]) == 'U')	{	processed += '3';	}
+		else									{	processed += '4';	}
+	}
+	return processed;
+}
+
+//********************************************************************************************************************
+
 string Sequence::getName(){
 	return name;
 }
@@ -106,19 +185,116 @@ string Sequence::getUnaligned(){
 
 //********************************************************************************************************************
 
-int Sequence::getLength(){
-	if(unaligned.length() > aligned.length())
-		return unaligned.length();
-	return aligned.length();
+int Sequence::getNumBases(){
+	return numBases;
 }
 
 //********************************************************************************************************************
 
 void Sequence::printSequence(ostream& out){
-	string toPrint = unaligned;
-	if(aligned.length() > unaligned.length())
-		toPrint = aligned;
-	out << ">" << name << "\n" << toPrint << "\n";
+
+	out << ">" << name << endl;
+	if(isAligned){
+		out << aligned << endl;
+	}
+	else{
+		out << unaligned << endl;
+	}
+}
+
+//********************************************************************************************************************
+
+int Sequence::getAlignLength(){
+	return alignmentLength;
+}
+
+//********************************************************************************************************************
+
+int Sequence::getAmbigBases(){
+	if(ambigBases == -1){
+		ambigBases = 0;
+		for(int j=0;j<numBases;j++){
+			if(unaligned[j] != 'A' && unaligned[j] != 'T' && unaligned[j] != 'G' && unaligned[j] != 'C'){
+				ambigBases++;
+			}
+		}
+	}	
+	
+	return ambigBases;
+}
+
+//********************************************************************************************************************
+
+int Sequence::getLongHomoPolymer(){
+	if(longHomoPolymer == -1){
+		longHomoPolymer = 1;
+		int homoPolymer = 1;
+		for(int j=1;j<numBases;j++){
+			if(unaligned[j] == unaligned[j-1]){
+				homoPolymer++;
+			}
+			else{
+				if(homoPolymer > longHomoPolymer){	longHomoPolymer = homoPolymer;	}
+				homoPolymer = 1;
+			}
+		}
+		if(homoPolymer > longHomoPolymer){	longHomoPolymer = homoPolymer;	}
+	}
+	return longHomoPolymer;
+}
+
+//********************************************************************************************************************
+
+int Sequence::getStartPos(){
+	if(endPos == -1){
+		for(int j = 0; j < alignmentLength; j++) {
+			if(aligned[j] != '.'){
+				startPos = j + 1;
+				break;
+			}
+		}
+	}
+	if(isAligned == 0){	startPos = 1;	}
+
+	return startPos;
+}
+
+//********************************************************************************************************************
+
+int Sequence::getEndPos(){
+	if(endPos == -1){
+		for(int j=alignmentLength-1;j>=0;j--){
+			if(aligned[j] != '.'){
+				endPos = j + 1;
+				break;
+			}
+		}
+	}
+	if(isAligned == 0){	endPos = numBases;	}
+	
+	return endPos;
+}
+
+//********************************************************************************************************************
+
+bool Sequence::getIsAligned(){
+	return isAligned;
+}
+
+//********************************************************************************************************************
+
+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;
+	
 }
 
 //********************************************************************************************************************