makefastqcommand.cpp

   1 /*
   2  *  makefastqcommand.cpp
   3  *  mothur
   4  *
   5  *  Created by westcott on 2/14/11.
   6  *  Copyright 2011 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "makefastqcommand.h"
  11 #include "sequence.hpp"
  12 #include "qualityscores.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> MakeFastQCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none","fastq",false,true,true); parameters.push_back(pfasta);
  18                 CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "none","fastq",false,true,true); parameters.push_back(pqfile);
  19                 CommandParameter pformat("format", "Multiple", "sanger-illumina-illumina1.8+", "sanger", "", "", "","",false,false); parameters.push_back(pformat);
  20         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "MakeFastQCommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string MakeFastQCommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The make.fastq command reads a fasta and quality file and creates a fastq file.\n";
  37                 helpString += "The make.fastq command parameters are fasta, qfile and format.  fasta and qfile are required.\n";
  38                 helpString += "The format parameter is used to indicate whether your sequences are sanger, illumina1.8+ or illumina, default=sanger.\n";
  39                 helpString += "The make.fastq command should be in the following format: make.fastq(qfile=yourQualityFile, fasta=yourFasta).\n";
  40                 helpString += "Example make.fastq(fasta=amazon.fasta, qfile=amazon.qual).\n";
  41                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  42                 return helpString;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "MakeFastQCommand", "getHelpString");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 string MakeFastQCommand::getOutputPattern(string type) {
  51     try {
  52         string pattern = "";
  53
  54         if (type == "fastq") {  pattern = "[filename],fastq"; }
  55         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  56
  57         return pattern;
  58     }
  59     catch(exception& e) {
  60         m->errorOut(e, "MakeFastQCommand", "getOutputPattern");
  61         exit(1);
  62     }
  63 }
  64 //**********************************************************************************************************************
  65 MakeFastQCommand::MakeFastQCommand(){
  66         try {
  67                 abort = true; calledHelp = true;
  68                 setParameters();
  69                 vector<string> tempOutNames;
  70                 outputTypes["fastq"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "MakeFastQCommand", "MakeFastQCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 MakeFastQCommand::MakeFastQCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["fastq"] = tempOutNames;
 103
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("fasta");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 116                                 }
 117
 118                                 it = parameters.find("qfile");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["qfile"] = inputDir + it->second;            }
 124                                 }
 125                         }
 126
 127
 128                         //check for required parameters
 129                         fastafile = validParameter.validFile(parameters, "fasta", true);
 130                         if (fastafile == "not open") { abort = true; fastafile = ""; }
 131                         else if (fastafile == "not found") {
 132                                 fastafile = m->getFastaFile();
 133                                 if (fastafile != "") {  m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 134                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 135                         }else { m->setFastaFile(fastafile); }
 136
 137                         qualfile = validParameter.validFile(parameters, "qfile", true);
 138                         if (qualfile == "not open") { abort = true; qualfile = ""; }
 139                         else if (qualfile == "not found") {
 140                                 qualfile = m->getQualFile();
 141                                 if (qualfile != "") {  m->mothurOut("Using " + qualfile + " as input file for the qfile parameter."); m->mothurOutEndLine(); }
 142                                 else {  m->mothurOut("You have no current qualfile and the qfile parameter is required."); m->mothurOutEndLine(); abort = true; }
 143                         }else { m->setQualFile(qualfile); }
 144
 145                         //if the user changes the output directory command factory will send this info to us in the output parameter
 146                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(fastafile);              }
 147
 148             format = validParameter.validFile(parameters, "format", false);             if (format == "not found"){     format = "sanger";      }
 149
 150             if ((format != "sanger") && (format != "illumina") && (format != "illumina1.8+"))  {
 151                                 m->mothurOut(format + " is not a valid format. Your format choices are sanger, illumina1.8+ and illumina, aborting." ); m->mothurOutEndLine();
 152                                 abort=true;
 153                         }
 154
 155
 156                 }
 157
 158         }
 159         catch(exception& e) {
 160                 m->errorOut(e, "MakeFastQCommand", "MakeFastQCommand");
 161                 exit(1);
 162         }
 163 }
 164 //**********************************************************************************************************************
 165
 166 int MakeFastQCommand::execute(){
 167         try {
 168
 169                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 170
 171         map<string, string> variables;
 172         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastafile));
 173                 string outputFile = getOutputFileName("fastq",variables);
 174                 outputNames.push_back(outputFile); outputTypes["fastq"].push_back(outputFile);
 175
 176                 ofstream out;
 177                 m->openOutputFile(outputFile, out);
 178
 179                 ifstream qFile;
 180                 m->openInputFile(qualfile, qFile);
 181
 182                 ifstream fFile;
 183                 m->openInputFile(fastafile, fFile);
 184
 185                 while (!fFile.eof() && !qFile.eof()) {
 186
 187                         if (m->control_pressed) { break; }
 188
 189                         Sequence currSeq(fFile); m->gobble(fFile);
 190                         QualityScores currQual(qFile);  m->gobble(qFile);
 191
 192                         if (currSeq.getName() != currQual.getName()) { m->mothurOut("[ERROR]: mismatch between fasta and quality files. Found " + currSeq.getName() + " in fasta file and " + currQual.getName() + " in quality file."); m->mothurOutEndLine(); m->control_pressed = true; }
 193                         else {
 194                                 //print sequence
 195                                 out << '@' << currSeq.getName() << endl << currSeq.getAligned() << endl;
 196
 197                                 string qualityString = convertQual(currQual.getQualityScores());
 198
 199                                 //print quality info
 200                                 out << '+' << currQual.getName() << endl << qualityString << endl;
 201                         }
 202
 203                 }
 204
 205                 fFile.close();
 206                 qFile.close();
 207                 out.close();
 208
 209                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 210
 211                 m->mothurOutEndLine();
 212                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 213                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 214                 m->mothurOutEndLine();
 215
 216                 return 0;
 217
 218         }
 219         catch(exception& e) {
 220                 m->errorOut(e, "MakeFastQCommand", "execute");
 221                 exit(1);
 222         }
 223 }
 224 //**********************************************************************************************************************
 225 string MakeFastQCommand::convertQual(vector<int> qual) {
 226         try {
 227                 string qualScores;
 228
 229         for (int i = 0; i < qual.size(); i++) {
 230             int controlChar = int('!');
 231             if (format == "illumina") {  controlChar = int('@');  }
 232
 233                         int temp = qual[i] + controlChar;
 234                         char qualChar = (char) temp;
 235
 236                         qualScores += qualChar;
 237                 }
 238
 239                 return qualScores;
 240         }
 241         catch(exception& e) {
 242                 m->errorOut(e, "MakeFastQCommand", "convertQual");
 243                 exit(1);
 244         }
 245 }
 246 //**********************************************************************************************************************
 247
 248
 249
 250
 251