X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;ds=sidebyside;f=sequence.cpp;h=6aa0c0f8ae2c46a32edde8a67276ccb966bc0003;hb=d0954e27635712cdbd8b86e3c4510670803a2665;hp=0afe3a971dcb1884160cd25f6c3d1b1d7012fa68;hpb=54a6cb0ef32b761ebd7ae35f1e60947edf44d699;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index 0afe3a9..6aa0c0f 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -54,29 +54,34 @@ Sequence::Sequence(istringstream& fastaString){
 	
 		initialize();
 		fastaString >> name;
-		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
+		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(); }
 		
-		sequence = getSequenceString(fastaString);		
-		setAligned(sequence);	
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
@@ -91,29 +96,34 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
 	
 		initialize();
 		fastaString >> name;
-		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
+		if (name.length() != 0) { 
+		
+			name = name.substr(1);
+			string sequence;
 		
-		sequence = getSequenceString(fastaString);		
+			//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(); }
 		
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
@@ -129,7 +139,7 @@ Sequence::Sequence(ifstream& fastaFile){
 		m = MothurOut::getInstance();
 		initialize();
 		fastaFile >> name;
-
+		
 		if (name.length() != 0) { 
 		
 			name = name.substr(1); 
@@ -151,15 +161,15 @@ Sequence::Sequence(ifstream& fastaFile){
 			}
 			
 			//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
+			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");
@@ -173,30 +183,34 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
 		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);
+			string sequence;
 			
-			if (fastaFile) {  
-				fastaFile >> name;  
-				name = name.substr(1);	
-			}else { 
-				name = "";
-				break;
+			//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);		
+			
+			//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(); }
 		
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);	
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
@@ -240,7 +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;
 			}
 		}
@@ -289,7 +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;
 			}
 		}
@@ -413,6 +427,13 @@ string Sequence::getAligned(){
 	else				{  return aligned;  }
 }
 
+//********************************************************************************************************************
+
+string Sequence::getInlineSeq(){
+	return name + '\t' + aligned;	
+}
+
+
 //********************************************************************************************************************
 
 string Sequence::getPairwise(){
@@ -467,6 +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;
@@ -488,7 +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;
@@ -503,6 +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--){
@@ -519,6 +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;
 }
@@ -539,4 +594,16 @@ void Sequence::reverseComplement(){
 	aligned = temp;
 	
 }
-/**************************************************************************************************/
+
+//********************************************************************************************************************
+
+void Sequence::trim(int length){
+	
+	if(numBases > length){
+		unaligned = unaligned.substr(0,length);
+		numBases = length;
+	}
+	
+}
+
+///**************************************************************************************************/