X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=sequence.cpp;h=d6073d75da3ed630b27b626bbba8615d79b58858;hp=b73bfab7caefdb0a97e5f7cc8f3271ec3e51cb57;hb=a8e2df1b96a57f5f29576b08361b86a96a8eff4f;hpb=17a6a53298a907c005fa93fb82af9e533adcda09

diff --git a/sequence.cpp b/sequence.cpp
index b73bfab..d6073d7 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -10,19 +10,18 @@
 #include "sequence.hpp"
 
 /***********************************************************************/
-
 Sequence::Sequence(){
 	m = MothurOut::getInstance();
 	initialize();
 }
-
 /***********************************************************************/
-
 Sequence::Sequence(string newName, string sequence) {
 	try {
 		m = MothurOut::getInstance();
 		initialize();	
 		name = newName;
+        
+        m->checkName(name);
 		
 		//setUnaligned removes any gap characters for us
 		setUnaligned(sequence);
@@ -33,43 +32,106 @@ Sequence::Sequence(string newName, string sequence) {
 		exit(1);
 	}			
 }
+/***********************************************************************/
+Sequence::Sequence(string newName, string sequence, string justUnAligned) {
+	try {
+		m = MothurOut::getInstance();
+		initialize();	
+		name = newName;
+        
+        m->checkName(name);
+		
+		//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 << "after name "  << endl;
-		name = name.substr(1);
-		string sequence;
+        name = getSequenceName(fastaString);
 		
-		//read comments
-		while ((name[0] == '#') && fastaString) { 
-			while (fastaString)	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
-			sequence = getCommentString(fastaString);
+		if (!m->control_pressed) { 
+			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
+			
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaString, 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(); }
+		}
+		
+	}
+	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();
+		name = getSequenceName(fastaString);
+		
+		if (!m->control_pressed) { 
+			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
+			
+			int numAmbig = 0;
+			sequence = getSequenceString(fastaString, 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(); }
+			
 		}
-	cout << "after mothur comment" << endl;	
-		//read real sequence
-		while (fastaString)	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
-	cout << "after mothur name" << endl;	
-		sequence = getSequenceString(fastaString);		
-	cout << "after mothur sequence" << endl;	
-		setAligned(sequence);	
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);		
+		
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
@@ -77,38 +139,141 @@ Sequence::Sequence(istringstream& fastaString){
 	}								
 }
 
+
 //********************************************************************************************************************
 //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;
-		name = name.substr(1);
-		string sequence;
+		name = getSequenceName(fastaFile);
 		
-		//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 (!m->control_pressed) { 
 			
-			if (fastaFile) {  
-				fastaFile >> name;  
-				name = name.substr(1);	
-			}else { 
-				name = "";
-				break;
+			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
+			
+			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(); }
+			
 		}
+
+	}
+	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();
+        extraInfo = "";
+		
+		name = getSequenceName(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
+		if (!m->control_pressed) { 			
+			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 || c == -1){  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(); }
+		}
+        
+	}
+	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();
+		name = getSequenceName(fastaFile);
 		
-		sequence = getSequenceString(fastaFile);		
+		if (!m->control_pressed) { 
+			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
+			
+			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(); }
+			
+		}
 		
-		setAligned(sequence);	
-		//setUnaligned removes any gap characters for us						
-		setUnaligned(sequence);	
 	}
 	catch(exception& e) {
 		m->errorOut(e, "Sequence", "Sequence");
@@ -116,20 +281,69 @@ Sequence::Sequence(ifstream& fastaFile){
 	}							
 }
 //********************************************************************************************************************
-string Sequence::getSequenceString(ifstream& fastaFile) {
+string Sequence::getSequenceName(ifstream& fastaFile) {
+	try {
+		string name = "";
+		
+        fastaFile >> name;
+		
+		if (name.length() != 0) { 
+            
+			name = name.substr(1); 
+            
+            m->checkName(name);
+            
+        }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
+        
+		return name;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getSequenceName");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+string Sequence::getSequenceName(istringstream& fastaFile) {
+	try {
+		string name = "";
+		
+        fastaFile >> name;
+		
+		if (name.length() != 0) { 
+            
+			name = name.substr(1); 
+            
+            m->checkName(name);
+            
+        }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
+        
+		return name;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getSequenceName");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+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;
 			}
 		}
@@ -151,7 +365,7 @@ 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;
 			}
 		}
@@ -164,20 +378,26 @@ 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){
+		while(!fastaFile.eof()){
 			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;
 			}
 		}
@@ -199,7 +419,7 @@ 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;
 			}
 		}
@@ -319,9 +539,17 @@ 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(){
@@ -339,6 +567,15 @@ string Sequence::getUnaligned(){
 int Sequence::getNumBases(){
 	return numBases;
 }
+//********************************************************************************************************************
+
+int Sequence::getNumNs(){
+    int numNs = 0;
+	for (int i = 0; i < unaligned.length(); i++) {
+        if(toupper(unaligned[i]) == 'N') { numNs++; }
+    }
+    return numNs;
+}
 
 //********************************************************************************************************************
 
@@ -376,6 +613,18 @@ 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;
@@ -397,9 +646,9 @@ 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;
 			}
@@ -412,10 +661,59 @@ 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; j++) {
+		aligned[j] = '.';
+	}
+	
+    //things like ......----------AT become ................AT
+    for(int j = start; 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;
 			}
@@ -428,10 +726,20 @@ 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(){
@@ -450,3 +758,16 @@ void Sequence::reverseComplement(){
 }
 
 //********************************************************************************************************************
+
+void Sequence::trim(int length){
+	
+	if(numBases > length){
+		unaligned = unaligned.substr(0,length);
+		numBases = length;
+        aligned = "";
+        isAligned = 0;
+	}
+	
+}
+
+///**************************************************************************************************/