X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=a359607a3ea11fa118efed129407ee608d4903e5;hb=cc19310422f125d6628980bd1148e1e816792382;hp=cfd8ec21498120614b94d3adbdade7275eb6a8b7;hpb=a33a385cc5b7481488f92f794425f01fbf40a543;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index cfd8ec2..a359607 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -191,6 +191,59 @@ 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();
+		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");
+		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();
@@ -550,7 +603,7 @@ 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;
 			}
@@ -581,6 +634,11 @@ int Sequence::filterToPos(int start){
 		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;
@@ -596,6 +654,11 @@ 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 +668,7 @@ 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 +682,7 @@ 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] = '.';
 	}