X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=sequence.cpp;h=d1cf7299a46c522255eb725dccb058673f6b1c78;hp=d877f4b93f33d8123ff6a1a9f7f839dc680fcf61;hb=1a20e24ee786195ab0e1cccd4f5aede7a88f3f4e;hpb=91a27e0483827c06c21c4fe89558923bbfe86573

diff --git a/sequence.cpp b/sequence.cpp
index d877f4b..d1cf729 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -20,6 +20,10 @@ Sequence::Sequence(string newName, string sequence) {
 		m = MothurOut::getInstance();
 		initialize();	
 		name = newName;
+        
+        for (int i = 0; i < name.length(); i++) {
+            if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
+        }
 		
 		//setUnaligned removes any gap characters for us
 		setUnaligned(sequence);
@@ -36,6 +40,10 @@ Sequence::Sequence(string newName, string sequence, string justUnAligned) {
 		m = MothurOut::getInstance();
 		initialize();	
 		name = newName;
+        
+        for (int i = 0; i < name.length(); i++) {
+            if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
+        }
 		
 		//setUnaligned removes any gap characters for us
 		setUnaligned(sequence);
@@ -53,11 +61,9 @@ Sequence::Sequence(istringstream& fastaString){
 		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
@@ -84,8 +90,7 @@ 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 +105,9 @@ 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 +134,7 @@ 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 +150,9 @@ Sequence::Sequence(ifstream& fastaFile){
 	try {
 		m = MothurOut::getInstance();
 		initialize();
-		fastaFile >> name;
-		
-		if (name.length() != 0) { 
+		name = getSequenceName(fastaFile);
 		
-			name = name.substr(1); 
+		if (!m->control_pressed) { 
 			
 			string sequence;
 		
@@ -181,7 +182,7 @@ 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) {
@@ -195,13 +196,11 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
 	try {
 		m = MothurOut::getInstance();
 		initialize();
-		fastaFile >> name;
         extraInfo = "";
 		
-		if (name.length() != 0) { 
-            
-			name = name.substr(1); 
-			
+		name = getSequenceName(fastaFile);
+		
+		if (!m->control_pressed) { 			
 			string sequence;
             
 			//read comments
@@ -221,7 +220,7 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
 			//read info after sequence name
 			while (!fastaFile.eof())	{	
                 char c = fastaFile.get(); 
-                if (c == 10 || c == 13){  break;	}	
+                if (c == 10 || c == 13 || c == -1){  break;	}
                 extraInfo += c;
             } 
 			
@@ -233,8 +232,7 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
 			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) {
@@ -248,10 +246,9 @@ 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
@@ -279,7 +276,7 @@ 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) {
@@ -287,7 +284,54 @@ 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); 
+            
+            for (int i = 0; i < name.length(); i++) {
+                if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
+            }
+            
+        }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); 
+            
+            for (int i = 0; i < name.length(); i++) {
+                if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; }
+            }
+            
+        }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 {
@@ -300,7 +344,7 @@ 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';}
@@ -354,7 +398,7 @@ 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';}
@@ -531,6 +575,15 @@ 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;
+}
 
 //********************************************************************************************************************
 
@@ -603,7 +656,7 @@ 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;
 			}
@@ -668,7 +721,7 @@ 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;
 			}
@@ -682,7 +735,7 @@ 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] = '.';
 	}
@@ -719,6 +772,8 @@ void Sequence::trim(int length){
 	if(numBases > length){
 		unaligned = unaligned.substr(0,length);
 		numBases = length;
+        aligned = "";
+        isAligned = 0;
 	}
 	
 }