X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=parsefastaqcommand.cpp;h=89f97acf13ad45650a1f8294a3e88895beaa347c;hb=eb71e28b7b7afd82540f4a8f0bac9429c5b9d713;hp=06ea359e6740bd822691e9f8ab6e3e53454cc724;hpb=bdd77df6a7df8a5775b57d0a7ec8b55de45e7b5b;p=mothur.git

diff --git a/parsefastaqcommand.cpp b/parsefastaqcommand.cpp
index 06ea359..89f97ac 100644
--- a/parsefastaqcommand.cpp
+++ b/parsefastaqcommand.cpp
@@ -13,11 +13,12 @@
 //**********************************************************************************************************************
 vector<string> ParseFastaQCommand::setParameters(){	
 	try {
-		CommandParameter pfastq("fastq", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfastq);
-		CommandParameter pfasta("fasta", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pfasta);
-		CommandParameter pqual("qfile", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pqual);
-		CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
-		CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+		CommandParameter pfastq("fastq", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(pfastq);
+		CommandParameter pfasta("fasta", "Boolean", "", "T", "", "", "","fasta",false,false); parameters.push_back(pfasta);
+		CommandParameter pqual("qfile", "Boolean", "", "T", "", "", "","qfile",false,false); parameters.push_back(pqual);
+ 		CommandParameter pformat("format", "Multiple", "sanger-illumina-solexa-illumina1.8+", "sanger", "", "", "","",false,false,true); parameters.push_back(pformat);
+        CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+		CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
 		
 		vector<string> myArray;
 		for (int i = 0; i < parameters.size(); i++) {	myArray.push_back(parameters[i].name);		}
@@ -33,8 +34,9 @@ string ParseFastaQCommand::getHelpString(){
 	try {
 		string helpString = "";
 		helpString += "The fastq.info command reads a fastq file and creates a fasta and quality file.\n";
-		helpString += "The fastq.info command parameters are fastq, fasta and qfile; fastq is required.\n";
-		helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
+		helpString += "The fastq.info command parameters are fastq, fasta, qfile and format; fastq is required.\n";
+        helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
+		helpString += "The format parameter is used to indicate whether your sequences are sanger, solexa, illumina1.8+ or illumina, default=sanger.\n";
         helpString += "The fasta parameter allows you to indicate whether you want a fasta file generated. Default=T.\n";
         helpString += "The qfile parameter allows you to indicate whether you want a quality file generated. Default=T.\n";
 		helpString += "Example fastq.info(fastaq=test.fastaq).\n";
@@ -47,6 +49,22 @@ string ParseFastaQCommand::getHelpString(){
 	}
 }
 //**********************************************************************************************************************
+string ParseFastaQCommand::getOutputPattern(string type) {
+    try {
+        string pattern = "";
+        
+        if (type == "fasta") {  pattern = "[filename],fasta"; } 
+        else if (type == "qfile") {  pattern = "[filename],qual"; } 
+        else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
+        
+        return pattern;
+    }
+    catch(exception& e) {
+        m->errorOut(e, "ParseFastaQCommand", "getOutputPattern");
+        exit(1);
+    }
+}
+//**********************************************************************************************************************
 ParseFastaQCommand::ParseFastaQCommand(){	
 	try {
 		abort = true; calledHelp = true; 
@@ -116,6 +134,13 @@ ParseFastaQCommand::ParseFastaQCommand(string option){
 			temp = validParameter.validFile(parameters, "qfile", false);	if(temp == "not found"){	temp = "T";	}
 			qual = m->isTrue(temp); 
 			
+            format = validParameter.validFile(parameters, "format", false);		if (format == "not found"){	format = "sanger";	}
+            
+            if ((format != "sanger") && (format != "illumina") && (format != "illumina1.8+") && (format != "solexa"))  { 
+				m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa, illumina1.8+ and illumina, aborting." ); m->mothurOutEndLine();
+				abort=true;
+			}
+
 			if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
 
 		}		
@@ -132,8 +157,10 @@ int ParseFastaQCommand::execute(){
 		if (abort == true) { if (calledHelp) { return 0; }  return 2;	}
 		
 		//open Output Files
-		string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "fasta";
-		string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "qual";
+        map<string, string> variables; 
+        variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
+		string fastaFile = getOutputFileName("fasta",variables);
+		string qualFile = getOutputFileName("qfile",variables);
 		ofstream outFasta, outQual;
 		
 		if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);	}
@@ -141,6 +168,12 @@ int ParseFastaQCommand::execute(){
 		
 		ifstream in;
 		m->openInputFile(fastaQFile, in);
+        
+        //fill convert table - goes from solexa to sanger. Used fq_all2std.pl as a reference.
+        for (int i = -64; i < 65; i++) { 
+            char temp = (char) ((int)(33 + 10*log(1+pow(10,(i/10.0)))/log(10)+0.499));
+            convertTable.push_back(temp);
+        }
 		
 		while (!in.eof()) {
 			
@@ -216,15 +249,28 @@ vector<int> ParseFastaQCommand::convertQual(string qual) {
 	try {
 		vector<int> qualScores;
 		
-		int controlChar = int('@');
-		
+        bool negativeScores = false;
+        
 		for (int i = 0; i < qual.length(); i++) { 
-			int temp = int(qual[i]);
-			temp -= controlChar;
-			
+            
+            int temp = 0;
+            temp = int(qual[i]);
+            if (format == "illumina") {
+                temp -= 64; //char '@'
+            }else if (format == "illumina1.8+") {
+                temp -= int('!'); //char '!'
+            }else if (format == "solexa") {
+                temp = int(convertTable[temp]); //convert to sanger
+                temp -= int('!'); //char '!'
+            }else {
+                temp -= int('!'); //char '!'
+            }
+            if (temp < -5) { negativeScores = true; }
 			qualScores.push_back(temp);
 		}
 		
+        if (negativeScores) { m->mothurOut("[ERROR]: finding negative quality scores, do you have the right format selected? http://en.wikipedia.org/wiki/FASTQ_format#Encoding \n");  m->control_pressed = true;  }
+        
 		return qualScores;
 	}
 	catch(exception& e) {