X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=872a0f091d49fddac359d528c53c799a50ef3c66;hb=4116449310d17a847470b84728cdefee5197e67e;hp=3bd80a87fde16f95d1bf1f8ae9369c0f962eb181;hpb=30c1fd8c45b6f0d66c17f2714dbb58b8ddccdce2;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index 3bd80a8..872a0f0 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -10,14 +10,11 @@
 #include "sequence.hpp"
 
 /***********************************************************************/
-
 Sequence::Sequence(){
 	m = MothurOut::getInstance();
 	initialize();
 }
-
 /***********************************************************************/
-
 Sequence::Sequence(string newName, string sequence) {
 	try {
 		m = MothurOut::getInstance();
@@ -33,44 +30,100 @@ Sequence::Sequence(string newName, string 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();
-	int pid;
-	MPI_Comm_rank(MPI_COMM_WORLD, &pid); 
-	cout << pid << " after mothur instance " << &name << endl;
+	
 		initialize();
-	cout << "after mothur initialize" << endl;
 		fastaString >> name;
-	cout << pid << " after name "  << name << endl;
-		name = name.substr(1);
-
-		string sequence;
-cout << pid << " name = "  << name << endl;		
-		//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 (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;
+				}
+			}
 			
-			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;
+				}
 			}
-			cout << pid << "in while comment" << endl;	
-		}
-	cout << pid << "after mothur comment" << endl;	
-		while (!fastaString.eof())	{	char c = fastaString.get(); cout << pid << " char = " << int(c) << endl; if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
-	cout << pid << " after mothur name" << endl;	
-		sequence = getSequenceString(fastaString);		
-	cout << pid << " after mothur sequence" << endl;	
-		setAligned(sequence);	
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);		
+			
+			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");
@@ -78,6 +131,7 @@ cout << pid << " name = "  << name << endl;
 	}								
 }
 
+
 //********************************************************************************************************************
 //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){
@@ -85,37 +139,85 @@ Sequence::Sequence(ifstream& fastaFile){
 		m = MothurOut::getInstance();
 		initialize();
 		fastaFile >> name;
-		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 (name.length() != 0) { 
+		
+			name = name.substr(1); 
 			
-			if (fastaFile) {  
-				fastaFile >> name;  
-				name = name.substr(1);	
-			}else { 
-				name = "";
-				break;
-			}
-		}
+			string sequence;
 		
-		//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
+			//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);		
+	
+			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;
 		
-		sequence = getSequenceString(fastaFile);		
+		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(); }
 		
-		setAligned(sequence);	
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);	
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
 		exit(1);
 	}							
 }
+
 //********************************************************************************************************************
 string Sequence::getSequenceString(ifstream& fastaFile) {
 	try {
@@ -152,7 +254,7 @@ string Sequence::getCommentString(ifstream& fastaFile) {
 		while(fastaFile){
 			letter=fastaFile.get();
 			if((letter == '\r') || (letter == '\n')){  
-				gobble(fastaFile);  //in case its a \r\n situation
+				m->gobble(fastaFile);  //in case its a \r\n situation
 				break;
 			}
 		}
@@ -169,11 +271,10 @@ string Sequence::getSequenceString(istringstream& fastaFile) {
 	try {
 		char letter;
 		string sequence = "";	
-int pid;
-MPI_Comm_rank(MPI_COMM_WORLD, &pid); 		
+		
 		while(!fastaFile.eof()){
 			letter= fastaFile.get();
-	cout << pid << '\t' << letter << endl;
+	
 			if(letter == '>'){
 				fastaFile.putback(letter);
 				break;
@@ -202,7 +303,7 @@ string Sequence::getCommentString(istringstream& fastaFile) {
 		while(fastaFile){
 			letter=fastaFile.get();
 			if((letter == '\r') || (letter == '\n')){  
-				gobble(fastaFile);  //in case its a \r\n situation
+				m->gobble(fastaFile);  //in case its a \r\n situation
 				break;
 			}
 		}
@@ -322,7 +423,8 @@ string Sequence::getName(){
 //********************************************************************************************************************
 
 string Sequence::getAligned(){
-	return aligned;
+	if(isAligned == 0)	{ return unaligned; }
+	else				{  return aligned;  }
 }
 
 //********************************************************************************************************************
@@ -379,6 +481,18 @@ 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;
@@ -453,3 +567,14 @@ void Sequence::reverseComplement(){
 }
 
 //********************************************************************************************************************
+
+void Sequence::trim(int length){
+	
+	if(numBases > length){
+		unaligned = unaligned.substr(0,length);
+		numBases = length;
+	}
+	
+}
+
+///**************************************************************************************************/