X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=96662bc36d4b77e1cb8b7a884f4e8460969033f4;hb=16f9c4ab6f39769856b13e048eae2c8eaa413c02;hp=b9e1c4b3ffbc4951d23204b3b761a74708003edc;hpb=7a8fc1115b3871107c09a4b9b307b1f2ab7d3fcf;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index b9e1c4b..96662bc 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -74,14 +74,19 @@ Sequence::Sequence(istringstream& fastaString){
 				}
 			}
 			
-			while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){ break;	}	} // get rest of line if there's any crap there
+			
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaString, numAmbig);
 			
-			sequence = getSequenceString(fastaString);		
 			setAligned(sequence);	
 			//setUnaligned removes any gap characters for us						
-			setUnaligned(sequence);		
+			setUnaligned(sequence);	
+			
+			if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+		
 		}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");
@@ -116,13 +121,18 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
 				}
 			}
 			
-			while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			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);		
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaString, numAmbig);
 			
 			//setUnaligned removes any gap characters for us						
-			setUnaligned(sequence);		
+			setUnaligned(sequence);	
+			
+			if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+			
 		}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");
@@ -138,7 +148,7 @@ Sequence::Sequence(ifstream& fastaFile){
 		m = MothurOut::getInstance();
 		initialize();
 		fastaFile >> name;
-
+		
 		if (name.length() != 0) { 
 		
 			name = name.substr(1); 
@@ -160,15 +170,72 @@ 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
+			
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaFile, numAmbig);
 			
-			sequence = getSequenceString(fastaFile);		
-		
 			setAligned(sequence);	
 			//setUnaligned removes any gap characters for us						
 			setUnaligned(sequence);	
+			
+			if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+			
 		}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& extraInfo, bool getInfo){
+	try {
+		m = MothurOut::getInstance();
+		initialize();
+		fastaFile >> name;
+        extraInfo = "";
 		
+		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 info after sequence name
+			while (!fastaFile.eof())	{	
+                char c = fastaFile.get(); 
+                if (c == 10 || c == 13){  break;	}	
+                extraInfo += c;
+            } 
+			
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaFile, numAmbig);
+			
+			setAligned(sequence);	
+			//setUnaligned removes any gap characters for us						
+			setUnaligned(sequence);	
+			
+			if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+			
+		}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");
@@ -202,14 +269,18 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
 			}
 			
 			//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);		
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaFile, numAmbig);
 			
 			//setUnaligned removes any gap characters for us						
 			setUnaligned(sequence);	
+			
+			if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
+			
 		}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");
@@ -218,20 +289,25 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
 }
 
 //********************************************************************************************************************
-string Sequence::getSequenceString(ifstream& fastaFile) {
+string Sequence::getSequenceString(ifstream& fastaFile, int& numAmbig) {
 	try {
 		char letter;
 		string sequence = "";	
+		numAmbig = 0;
 		
 		while(fastaFile){
 			letter= fastaFile.get();
 			if(letter == '>'){
 				fastaFile.putback(letter);
 				break;
-			}
+			}else if (letter == ' ') {;}
 			else if(isprint(letter)){
 				letter = toupper(letter);
 				if(letter == 'U'){letter = 'T';}
+				if(letter != '.' && letter != '-' && letter != 'A' && letter != 'T' && letter != 'G'  && letter != 'C' && letter != 'N'){
+					letter = 'N';
+					numAmbig++;
+				}
 				sequence += letter;
 			}
 		}
@@ -253,7 +329,7 @@ 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;
 			}
 		}
@@ -266,10 +342,11 @@ string Sequence::getCommentString(ifstream& fastaFile) {
 	}
 }
 //********************************************************************************************************************
-string Sequence::getSequenceString(istringstream& fastaFile) {
+string Sequence::getSequenceString(istringstream& fastaFile, int& numAmbig) {
 	try {
 		char letter;
-		string sequence = "";	
+		string sequence = "";
+		numAmbig = 0;
 		
 		while(!fastaFile.eof()){
 			letter= fastaFile.get();
@@ -277,10 +354,14 @@ string Sequence::getSequenceString(istringstream& fastaFile) {
 			if(letter == '>'){
 				fastaFile.putback(letter);
 				break;
-			}
+			}else if (letter == ' ') {;}
 			else if(isprint(letter)){
 				letter = toupper(letter);
 				if(letter == 'U'){letter = 'T';}
+				if(letter != '.' && letter != '-' && letter != 'A' && letter != 'T' && letter != 'G'  && letter != 'C' && letter != 'N'){
+					letter = 'N';
+					numAmbig++;
+				}
 				sequence += letter;
 			}
 		}
@@ -302,7 +383,7 @@ 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;
 			}
 		}
@@ -426,6 +507,13 @@ string Sequence::getAligned(){
 	else				{  return aligned;  }
 }
 
+//********************************************************************************************************************
+
+string Sequence::getInlineSeq(){
+	return name + '\t' + aligned;	
+}
+
+
 //********************************************************************************************************************
 
 string Sequence::getPairwise(){
@@ -480,6 +568,18 @@ 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;
@@ -501,9 +601,9 @@ int Sequence::getLongHomoPolymer(){
 //********************************************************************************************************************
 
 int Sequence::getStartPos(){
-	if(endPos == -1){
+	if(startPos == -1){
 		for(int j = 0; j < alignmentLength; j++) {
-			if(aligned[j] != '.'){
+			if((aligned[j] != '.')&&(aligned[j] != '-')){
 				startPos = j + 1;
 				break;
 			}
@@ -516,10 +616,59 @@ int Sequence::getStartPos(){
 
 //********************************************************************************************************************
 
+void Sequence::padToPos(int start){
+
+	for(int j = startPos-1; j < start-1; j++) {
+		aligned[j] = '.';
+	}
+	startPos = start;
+
+}
+//********************************************************************************************************************
+
+int Sequence::filterToPos(int start){
+    
+    if (start > aligned.length()) { start = aligned.length(); m->mothurOut("[ERROR]: start to large.\n"); }
+    
+	for(int j = 0; j < start-1; j++) {
+		aligned[j] = '.';
+	}
+	
+    //things like ......----------AT become ................AT
+    for(int j = start-1; j < aligned.length(); j++) {
+        if (isalpha(aligned[j])) { break; }
+        else { aligned[j] = '.'; }
+    }
+    setUnaligned(aligned);
+    
+    return 0;
+    
+}
+//********************************************************************************************************************
+
+int Sequence::filterFromPos(int end){
+    
+    if (end > aligned.length()) { end = aligned.length(); m->mothurOut("[ERROR]: end to large.\n"); }
+    
+	for(int j = end; j < aligned.length(); j++) {
+		aligned[j] = '.';
+	}
+	
+    for(int j = aligned.length()-1; j < 0; j--) {
+        if (isalpha(aligned[j])) { break; }
+        else { aligned[j] = '.'; }
+    }
+    
+    setUnaligned(aligned);
+    
+    return 0;
+}
+//********************************************************************************************************************
+
 int Sequence::getEndPos(){
 	if(endPos == -1){
 		for(int j=alignmentLength-1;j>=0;j--){
-			if(aligned[j] != '.'){
+			if((aligned[j] != '.')&&(aligned[j] != '-')){
 				endPos = j + 1;
 				break;
 			}
@@ -532,10 +681,20 @@ int Sequence::getEndPos(){
 
 //********************************************************************************************************************
 
+void Sequence::padFromPos(int end){
+	//cout << end << '\t' << endPos << endl;
+	for(int j = end; j < endPos; j++) {
+		aligned[j] = '.';
+	}
+	endPos = end;
+	
+}
+
+//********************************************************************************************************************
+
 bool Sequence::getIsAligned(){
 	return isAligned;
 }
-
 //********************************************************************************************************************
 
 void Sequence::reverseComplement(){
@@ -552,4 +711,16 @@ void Sequence::reverseComplement(){
 	aligned = temp;
 	
 }
-/**************************************************************************************************/
+
+//********************************************************************************************************************
+
+void Sequence::trim(int length){
+	
+	if(numBases > length){
+		unaligned = unaligned.substr(0,length);
+		numBases = length;
+	}
+	
+}
+
+///**************************************************************************************************/