vector<string> ParseFastaQCommand::setParameters(){
try {
CommandParameter pfastq("fastq", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfastq);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
+ CommandParameter pfasta("fasta", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pfasta);
+ CommandParameter pqual("qfile", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pqual);
+ CommandParameter pformat("format", "Multiple", "sanger-illumina-solexa", "sanger", "", "", "",false,false); 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;
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 parameter is fastq, and it 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 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";
helpString += "Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n";
return helpString;
exit(1);
}
}
+//**********************************************************************************************************************
+string ParseFastaQCommand::getOutputFileNameTag(string type, string inputName=""){
+ try {
+ string outputFileName = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "fasta") { outputFileName = "fasta"; }
+ else if (type == "qfile") { outputFileName = "qual"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "ParseFastaQCommand", "getOutputFileNameTag");
+ exit(1);
+ }
+}
+
//**********************************************************************************************************************
ParseFastaQCommand::ParseFastaQCommand(){
try {
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){ outputDir = m->hasPath(fastaQFile); }
+
+ string temp;
+ temp = validParameter.validFile(parameters, "fasta", false); if(temp == "not found"){ temp = "T"; }
+ fasta = m->isTrue(temp);
+
+ 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 != "solexa")) {
+ m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa and illumina, aborting." ); m->mothurOutEndLine();
+ abort=true;
+ }
+
+ if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
}
}
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";
+ string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("fasta");
+ string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("qfile");
ofstream outFasta, outQual;
- m->openOutputFile(fastaFile, outFasta); outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);
- m->openOutputFile(qualFile, outQual); outputNames.push_back(qualFile); outputTypes["qfile"].push_back(qualFile);
+
+ if (fasta) { m->openOutputFile(fastaFile, outFasta); outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile); }
+ if (qual) { m->openOutputFile(qualFile, outQual); outputNames.push_back(qualFile); outputTypes["qfile"].push_back(qualFile); }
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()) {
+
+ if (m->control_pressed) { break; }
//read sequence name
string name = m->getline(in); m->gobble(in);
else { name2 = name2.substr(1); }
//read quality scores
- string qual = m->getline(in); m->gobble(in);
- if (qual == "") { m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
+ string quality = m->getline(in); m->gobble(in);
+ if (quality == "") { m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
//sanity check sequence length and number of quality scores match
if (name2 != "") { if (name != name2) { m->mothurOut("[ERROR]: names do not match. read " + name + " for fasta and " + name2 + " for quality."); m->mothurOutEndLine(); m->control_pressed = true; break; } }
- if (qual.length() != sequence.length()) { m->mothurOut("[ERROR]: lengths do not match. read " + toString(sequence.length()) + " characters for fasta and " + toString(qual.length()) + " characters for quality scores."); m->mothurOutEndLine(); m->control_pressed = true; break; }
-
- //convert quality scores
- vector<int> qualScores = convertQual(qual);
+ if (quality.length() != sequence.length()) { m->mothurOut("[ERROR]: Lengths do not match for sequence " + name + ". Read " + toString(sequence.length()) + " characters for fasta and " + toString(quality.length()) + " characters for quality scores."); m->mothurOutEndLine(); m->control_pressed = true; break; }
//print sequence info to files
- outFasta << ">" << name << endl << sequence << endl;
+ if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
- outQual << ">" << name << endl;
- for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
- outQual << endl;
+ if (qual) {
+ vector<int> qualScores = convertQual(quality);
+ outQual << ">" << name << endl;
+ for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
+ outQual << endl;
+ }
}
in.close();
- outFasta.close();
- outQual.close();
+ if (fasta) { outFasta.close(); }
+ if (qual) { outQual.close(); }
if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
try {
vector<int> qualScores;
- int controlChar = int('!');
-
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 == "solexa") {
+ temp = int(convertTable[temp]); //convert to sanger
+ temp -= 33; //char '!'
+ }else {
+ temp -= 33; //char '!'
+ }
qualScores.push_back(temp);
}
projects / mothur.git / blobdiff