X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=sequence.cpp;h=d6073d75da3ed630b27b626bbba8615d79b58858;hp=cfd8ec21498120614b94d3adbdade7275eb6a8b7;hb=a8e2df1b96a57f5f29576b08361b86a96a8eff4f;hpb=a33a385cc5b7481488f92f794425f01fbf40a543

diff --git a/sequence.cpp b/sequence.cpp
index cfd8ec2..d6073d7 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -20,6 +20,8 @@ Sequence::Sequence(string newName, string sequence) {
 		m = MothurOut::getInstance();
 		initialize();	
 		name = newName;
+        
+        m->checkName(name);
 		
 		//setUnaligned removes any gap characters for us
 		setUnaligned(sequence);
@@ -36,6 +38,8 @@ Sequence::Sequence(string newName, string sequence, string justUnAligned) {
 		m = MothurOut::getInstance();
 		initialize();	
 		name = newName;
+        
+        m->checkName(name);
 		
 		//setUnaligned removes any gap characters for us
 		setUnaligned(sequence);
@@ -53,11 +57,9 @@ Sequence::Sequence(istringstream& fastaString){
 		m = MothurOut::getInstance();
 	
 		initialize();
-		fastaString >> name;
+        name = getSequenceName(fastaString);
 		
-		if (name.length() != 0) { 
-		
-			name = name.substr(1);
+		if (!m->control_pressed) { 
 			string sequence;
 		
 			//read comments
@@ -84,8 +86,7 @@ Sequence::Sequence(istringstream& fastaString){
 			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) {
@@ -100,11 +101,9 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
 		m = MothurOut::getInstance();
 	
 		initialize();
-		fastaString >> name;
-		
-		if (name.length() != 0) { 
+		name = getSequenceName(fastaString);
 		
-			name = name.substr(1);
+		if (!m->control_pressed) { 
 			string sequence;
 		
 			//read comments
@@ -131,7 +130,7 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
 			
 			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) {
@@ -147,11 +146,9 @@ Sequence::Sequence(ifstream& fastaFile){
 	try {
 		m = MothurOut::getInstance();
 		initialize();
-		fastaFile >> name;
+		name = getSequenceName(fastaFile);
 		
-		if (name.length() != 0) { 
-		
-			name = name.substr(1); 
+		if (!m->control_pressed) { 
 			
 			string sequence;
 		
@@ -181,7 +178,7 @@ Sequence::Sequence(ifstream& fastaFile){
 			
 			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) {
@@ -191,14 +188,63 @@ Sequence::Sequence(ifstream& fastaFile){
 }
 //********************************************************************************************************************
 //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);
+		
+		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();
-		fastaFile >> name;
+		name = getSequenceName(fastaFile);
 		
-		if (name.length() != 0) { 
-			name = name.substr(1);
+		if (!m->control_pressed) { 
 			string sequence;
 			
 			//read comments
@@ -226,7 +272,7 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
 			
 			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) {
@@ -234,7 +280,50 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
 		exit(1);
 	}							
 }
-
+//********************************************************************************************************************
+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 {
@@ -247,7 +336,7 @@ string Sequence::getSequenceString(ifstream& fastaFile, int& numAmbig) {
 			if(letter == '>'){
 				fastaFile.putback(letter);
 				break;
-			}
+			}else if (letter == ' ') {;}
 			else if(isprint(letter)){
 				letter = toupper(letter);
 				if(letter == 'U'){letter = 'T';}
@@ -301,7 +390,7 @@ string Sequence::getSequenceString(istringstream& fastaFile, int& numAmbig) {
 			if(letter == '>'){
 				fastaFile.putback(letter);
 				break;
-			}
+			}else if (letter == ' ') {;}
 			else if(isprint(letter)){
 				letter = toupper(letter);
 				if(letter == 'U'){letter = 'T';}
@@ -478,6 +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;
+}
 
 //********************************************************************************************************************
 
@@ -550,7 +648,7 @@ int Sequence::getLongHomoPolymer(){
 int Sequence::getStartPos(){
 	if(startPos == -1){
 		for(int j = 0; j < alignmentLength; j++) {
-			if(aligned[j] != '.'){
+			if((aligned[j] != '.')&&(aligned[j] != '-')){
 				startPos = j + 1;
 				break;
 			}
@@ -577,10 +675,15 @@ 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++) {
+	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;
@@ -596,6 +699,11 @@ int Sequence::filterFromPos(int end){
 		aligned[j] = '.';
 	}
 	
+    for(int j = aligned.length()-1; j < 0; j--) {
+        if (isalpha(aligned[j])) { break; }
+        else { aligned[j] = '.'; }
+    }
+    
     setUnaligned(aligned);
     
     return 0;
@@ -605,7 +713,7 @@ int Sequence::filterFromPos(int end){
 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;
 			}
@@ -619,7 +727,7 @@ int Sequence::getEndPos(){
 //********************************************************************************************************************
 
 void Sequence::padFromPos(int end){
-	cout << end << '\t' << endPos << endl;
+	//cout << end << '\t' << endPos << endl;
 	for(int j = end; j < endPos; j++) {
 		aligned[j] = '.';
 	}
@@ -656,6 +764,8 @@ void Sequence::trim(int length){
 	if(numBases > length){
 		unaligned = unaligned.substr(0,length);
 		numBases = length;
+        aligned = "";
+        isAligned = 0;
 	}
 	
 }