parsefastaqcommand.cpp

   1 /*
   2  *  parsefastaqcommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/30/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "parsefastaqcommand.h"
  11 #include "sequence.hpp"
  12
  13 //**********************************************************************************************************************
  14 vector<string> ParseFastaQCommand::setParameters(){
  15         try {
  16                 CommandParameter pfastq("fastq", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfastq);
  17                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  18                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  19
  20                 vector<string> myArray;
  21                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  22                 return myArray;
  23         }
  24         catch(exception& e) {
  25                 m->errorOut(e, "ParseFastaQCommand", "setParameters");
  26                 exit(1);
  27         }
  28 }
  29 //**********************************************************************************************************************
  30 string ParseFastaQCommand::getHelpString(){
  31         try {
  32                 string helpString = "";
  33                 helpString += "The fastq.info command reads a fastq file and creates a fasta and quality file.\n";
  34                 helpString += "The fastq.info command parameter is fastq, and it is required.\n";
  35                 helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
  36                 helpString += "Example fastq.info(fastaq=test.fastaq).\n";
  37                 helpString += "Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n";
  38                 return helpString;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "ParseFastaQCommand", "getHelpString");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 ParseFastaQCommand::ParseFastaQCommand(){
  47         try {
  48                 abort = true; calledHelp = true;
  49                 setParameters();
  50                 vector<string> tempOutNames;
  51                 outputTypes["fasta"] = tempOutNames;
  52                 outputTypes["qfile"] = tempOutNames;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 ParseFastaQCommand::ParseFastaQCommand(string option){
  61         try {
  62                 abort = false; calledHelp = false;
  63
  64                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  65                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  66
  67                 else {
  68                         vector<string> myArray = setParameters();
  69
  70                         OptionParser parser(option);
  71                         map<string,string> parameters = parser.getParameters();
  72
  73                         ValidParameters validParameter;
  74                         map<string,string>::iterator it;
  75
  76                         //check to make sure all parameters are valid for command
  77                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  78                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  79                         }
  80
  81                         //initialize outputTypes
  82                         vector<string> tempOutNames;
  83                         outputTypes["fasta"] = tempOutNames;
  84                         outputTypes["qfile"] = tempOutNames;
  85
  86                         //if the user changes the input directory command factory will send this info to us in the output parameter
  87                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  88                         if (inputDir == "not found"){   inputDir = "";          }
  89                         else {
  90                                 string path;
  91                                 it = parameters.find("fastq");
  92                                 //user has given a template file
  93                                 if(it != parameters.end()){
  94                                         path = m->hasPath(it->second);
  95                                         //if the user has not given a path then, add inputdir. else leave path alone.
  96                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
  97                                 }
  98                         }
  99
 100                         //check for required parameters
 101                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 102                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 103                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 104
 105                         //if the user changes the output directory command factory will send this info to us in the output parameter
 106                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 107
 108                 }
 109         }
 110         catch(exception& e) {
 111                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 112                 exit(1);
 113         }
 114 }
 115 //**********************************************************************************************************************
 116
 117 int ParseFastaQCommand::execute(){
 118         try {
 119                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 120
 121                 //open Output Files
 122                 string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "fasta";
 123                 string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "qual";
 124                 ofstream outFasta, outQual;
 125                 m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);
 126                 m->openOutputFile(qualFile, outQual);   outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);
 127
 128                 ifstream in;
 129                 m->openInputFile(fastaQFile, in);
 130
 131                 while (!in.eof()) {
 132
 133                         //read sequence name
 134                         string name = m->getline(in); m->gobble(in);
 135                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 136                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 137                         else { name = name.substr(1); }
 138
 139                         //read sequence
 140                         string sequence = m->getline(in); m->gobble(in);
 141                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 142
 143                         //read sequence name
 144                         string name2 = m->getline(in); m->gobble(in);
 145                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 146                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 147                         else { name2 = name2.substr(1);  }
 148
 149                         //read quality scores
 150                         string qual = m->getline(in); m->gobble(in);
 151                         if (qual == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 152
 153                         //sanity check sequence length and number of quality scores match
 154                         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; } }
 155                         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; }
 156
 157                         //convert quality scores
 158                         vector<int> qualScores = convertQual(qual);
 159
 160                         //print sequence info to files
 161                         outFasta << ">" << name << endl << sequence << endl;
 162
 163                         outQual << ">" << name << endl;
 164                         for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 165                         outQual << endl;
 166                 }
 167
 168                 in.close();
 169                 outFasta.close();
 170                 outQual.close();
 171
 172                 if (m->control_pressed) { outputTypes.clear(); remove(fastaFile.c_str()); remove(qualFile.c_str()); return 0; }
 173
 174                 //set fasta file as new current fastafile
 175                 string current = "";
 176                 itTypes = outputTypes.find("fasta");
 177                 if (itTypes != outputTypes.end()) {
 178                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 179                 }
 180
 181                 itTypes = outputTypes.find("qfile");
 182                 if (itTypes != outputTypes.end()) {
 183                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 184                 }
 185
 186                 m->mothurOutEndLine();
 187                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 188                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 189                 m->mothurOutEndLine();
 190
 191                 return 0;
 192         }
 193         catch(exception& e) {
 194                 m->errorOut(e, "ParseFastaQCommand", "execute");
 195                 exit(1);
 196         }
 197 }
 198 //**********************************************************************************************************************
 199 vector<int> ParseFastaQCommand::convertQual(string qual) {
 200         try {
 201                 vector<int> qualScores;
 202
 203                 int controlChar = int('!');
 204
 205                 for (int i = 0; i < qual.length(); i++) {
 206                         int temp = int(qual[i]);
 207                         temp -= controlChar;
 208
 209                         qualScores.push_back(temp);
 210                 }
 211
 212                 return qualScores;
 213         }
 214         catch(exception& e) {
 215                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 216                 exit(1);
 217         }
 218 }
 219 //**********************************************************************************************************************
 220
 221
 222