X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=6aa0c0f8ae2c46a32edde8a67276ccb966bc0003;hb=65f2be9d8f9b4cf72b4aafd800fdd11ad26435fd;hp=685f0726e6ec5042d32379874e54b83e8a6eff26;hpb=861f46b74c17adec8c6ad6d89f232ae7485797bf;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index 685f072..6aa0c0f 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -10,55 +10,311 @@
 #include "sequence.hpp"
 
 /***********************************************************************/
-
 Sequence::Sequence(){
+	m = MothurOut::getInstance();
 	initialize();
 }
-
 /***********************************************************************/
-
 Sequence::Sequence(string newName, string sequence) {
-
-	initialize();	
-	name = newName;
-	if(sequence.find_first_of('-') != string::npos) {
+	try {
+		m = MothurOut::getInstance();
+		initialize();	
+		name = newName;
+		
+		//setUnaligned removes any gap characters for us
+		setUnaligned(sequence);
 		setAligned(sequence);
 	}
-	setUnaligned(sequence);
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}			
+}
+/***********************************************************************/
+Sequence::Sequence(string newName, string sequence, string justUnAligned) {
+	try {
+		m = MothurOut::getInstance();
+		initialize();	
+		name = newName;
+		
+		//setUnaligned removes any gap characters for us
+		setUnaligned(sequence);
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}			
+}
+
+//********************************************************************************************************************
+//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(istringstream& fastaString){
+	try {
+		m = MothurOut::getInstance();
 	
+		initialize();
+		fastaString >> name;
+		
+		if (name.length() != 0) { 
+		
+			name = name.substr(1);
+			string sequence;
+		
+			//read comments
+			while ((name[0] == '#') && fastaString) { 
+				while (!fastaString.eof())	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+				sequence = getCommentString(fastaString);
+				
+				if (fastaString) {  
+					fastaString >> name;  
+					name = name.substr(1);	
+				}else { 
+					name = "";
+					break;
+				}
+			}
+			
+			while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){ break;	}	} // get rest of line if there's any crap there
+			
+			sequence = getSequenceString(fastaString);		
+			setAligned(sequence);	
+			//setUnaligned removes any gap characters for us						
+			setUnaligned(sequence);		
+		}else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}								
 }
 //********************************************************************************************************************
+//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(istringstream& fastaString, string JustUnaligned){
+	try {
+		m = MothurOut::getInstance();
+	
+		initialize();
+		fastaString >> name;
+		
+		if (name.length() != 0) { 
+		
+			name = name.substr(1);
+			string sequence;
+		
+			//read comments
+			while ((name[0] == '#') && fastaString) { 
+				while (!fastaString.eof())	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+				sequence = getCommentString(fastaString);
+				
+				if (fastaString) {  
+					fastaString >> name;  
+					name = name.substr(1);	
+				}else { 
+					name = "";
+					break;
+				}
+			}
+			
+			while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){ break;	}	} // get rest of line if there's any crap there
+			
+			sequence = getSequenceString(fastaString);		
+			
+			//setUnaligned removes any gap characters for us						
+			setUnaligned(sequence);		
+		}else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}								
+}
 
-Sequence::Sequence(ifstream& fastaFile){
 
-	initialize();
-	fastaFile >> name;
-	name = name.substr(1);
+//********************************************************************************************************************
+//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){
+	try {
+		m = MothurOut::getInstance();
+		initialize();
+		fastaFile >> name;
+		
+		if (name.length() != 0) { 
+		
+			name = name.substr(1); 
+			
+			string sequence;
+		
+			//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;
+				}
+			}
+			
+			//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);		
 	
-	while (!fastaFile.eof())	{	char c = fastaFile.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			setAligned(sequence);	
+			//setUnaligned removes any gap characters for us						
+			setUnaligned(sequence);	
+		}else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}							
+}
+//********************************************************************************************************************
+//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 JustUnaligned){
+	try {
+		m = MothurOut::getInstance();
+		initialize();
+		fastaFile >> name;
+		
+		if (name.length() != 0) { 
+			name = name.substr(1);
+			string sequence;
+			
+			//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;
+				}
+			}
+			
+			//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);		
+			
+			//setUnaligned removes any gap characters for us						
+			setUnaligned(sequence);	
+		}else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}							
+}
 
-	char letter;
-	string 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;
+	}
+	catch(exception& e) {
+		m->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')){  
+				m->gobble(fastaFile);  //in case its a \r\n situation
+				break;
+			}
+		}
+		
+		return sequence;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getCommentString");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+string Sequence::getSequenceString(istringstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";	
+		
+		while(!fastaFile.eof()){
+			letter= fastaFile.get();
 	
-	while(fastaFile){
-		letter= fastaFile.get();
-		if(letter == '>'){
-			fastaFile.putback(letter);
-			break;
+			if(letter == '>'){
+				fastaFile.putback(letter);
+				break;
+			}
+			else if(isprint(letter)){
+				letter = toupper(letter);
+				if(letter == 'U'){letter = 'T';}
+				sequence += letter;
+			}
 		}
-		else if(isprint(letter)){
-			letter = toupper(letter);
-			if(letter == 'U'){letter = 'T';}
-			sequence += letter;
+		
+		return sequence;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getSequenceString");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(istringstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";
+		
+		while(fastaFile){
+			letter=fastaFile.get();
+			if((letter == '\r') || (letter == '\n')){  
+				m->gobble(fastaFile);  //in case its a \r\n situation
+				break;
+			}
 		}
+		
+		return sequence;
 	}
-
-	if((sequence.find_first_of('-') != string::npos) || (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) {
+		m->errorOut(e, "Sequence", "getCommentString");
+		exit(1);
 	}
-	setUnaligned(sequence);								//	also set the unaligned sequence file
 }
-
 //********************************************************************************************************************
 
 void Sequence::initialize(){
@@ -110,6 +366,7 @@ 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++){
@@ -166,9 +423,17 @@ string Sequence::getName(){
 //********************************************************************************************************************
 
 string Sequence::getAligned(){
-	return aligned;
+	if(isAligned == 0)	{ return unaligned; }
+	else				{  return aligned;  }
 }
 
+//********************************************************************************************************************
+
+string Sequence::getInlineSeq(){
+	return name + '\t' + aligned;	
+}
+
+
 //********************************************************************************************************************
 
 string Sequence::getPairwise(){
@@ -223,6 +488,18 @@ int Sequence::getAmbigBases(){
 
 //********************************************************************************************************************
 
+void Sequence::removeAmbigBases(){
+	
+	for(int j=0;j<alignmentLength;j++){
+		if(aligned[j] != 'A' && aligned[j] != 'T' && aligned[j] != 'G' && aligned[j] != 'C'){
+			aligned[j] = '-';
+		}
+	}
+	setUnaligned(aligned);
+}
+	
+//********************************************************************************************************************
+
 int Sequence::getLongHomoPolymer(){
 	if(longHomoPolymer == -1){
 		longHomoPolymer = 1;
@@ -244,7 +521,7 @@ int Sequence::getLongHomoPolymer(){
 //********************************************************************************************************************
 
 int Sequence::getStartPos(){
-	if(endPos == -1){
+	if(startPos == -1){
 		for(int j = 0; j < alignmentLength; j++) {
 			if(aligned[j] != '.'){
 				startPos = j + 1;
@@ -259,6 +536,17 @@ int Sequence::getStartPos(){
 
 //********************************************************************************************************************
 
+void Sequence::padToPos(int start){
+
+	for(int j = startPos-1; j < start-1; j++) {
+		aligned[j] = '.';
+	}
+	startPos = start;
+
+}
+
+//********************************************************************************************************************
+
 int Sequence::getEndPos(){
 	if(endPos == -1){
 		for(int j=alignmentLength-1;j>=0;j--){
@@ -275,6 +563,17 @@ int Sequence::getEndPos(){
 
 //********************************************************************************************************************
 
+void Sequence::padFromPos(int end){
+	
+	for(int j = end; j < endPos; j++) {
+		aligned[j] = '.';
+	}
+	endPos = end;
+	
+}
+
+//********************************************************************************************************************
+
 bool Sequence::getIsAligned(){
 	return isAligned;
 }
@@ -292,7 +591,19 @@ void Sequence::reverseComplement(){
 		else						{	temp += 'N';	}
 	}
 	unaligned = temp;
+	aligned = temp;
 	
 }
 
 //********************************************************************************************************************
+
+void Sequence::trim(int length){
+	
+	if(numBases > length){
+		unaligned = unaligned.substr(0,length);
+		numBases = length;
+	}
+	
+}
+
+///**************************************************************************************************/