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 pfasta("fasta", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pqual("qfile", "Bool", "", "T", "", "", "",false,false); parameters.push_back(pqual);
  19                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  20                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  21
  22                 vector<string> myArray;
  23                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  24                 return myArray;
  25         }
  26         catch(exception& e) {
  27                 m->errorOut(e, "ParseFastaQCommand", "setParameters");
  28                 exit(1);
  29         }
  30 }
  31 //**********************************************************************************************************************
  32 string ParseFastaQCommand::getHelpString(){
  33         try {
  34                 string helpString = "";
  35                 helpString += "The fastq.info command reads a fastq file and creates a fasta and quality file.\n";
  36                 helpString += "The fastq.info command parameters are fastq, fasta and qfile; fastq is required.\n";
  37                 helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
  38         helpString += "The fasta parameter allows you to indicate whether you want a fasta file generated. Default=T.\n";
  39         helpString += "The qfile parameter allows you to indicate whether you want a quality file generated. Default=T.\n";
  40                 helpString += "Example fastq.info(fastaq=test.fastaq).\n";
  41                 helpString += "Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n";
  42                 return helpString;
  43         }
  44         catch(exception& e) {
  45                 m->errorOut(e, "ParseFastaQCommand", "getHelpString");
  46                 exit(1);
  47         }
  48 }
  49 //**********************************************************************************************************************
  50 ParseFastaQCommand::ParseFastaQCommand(){
  51         try {
  52                 abort = true; calledHelp = true;
  53                 setParameters();
  54                 vector<string> tempOutNames;
  55                 outputTypes["fasta"] = tempOutNames;
  56                 outputTypes["qfile"] = tempOutNames;
  57         }
  58         catch(exception& e) {
  59                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  60                 exit(1);
  61         }
  62 }
  63 //**********************************************************************************************************************
  64 ParseFastaQCommand::ParseFastaQCommand(string option){
  65         try {
  66                 abort = false; calledHelp = false;
  67
  68                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  69                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  70
  71                 else {
  72                         vector<string> myArray = setParameters();
  73
  74                         OptionParser parser(option);
  75                         map<string,string> parameters = parser.getParameters();
  76
  77                         ValidParameters validParameter;
  78                         map<string,string>::iterator it;
  79
  80                         //check to make sure all parameters are valid for command
  81                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
  82                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  83                         }
  84
  85                         //initialize outputTypes
  86                         vector<string> tempOutNames;
  87                         outputTypes["fasta"] = tempOutNames;
  88                         outputTypes["qfile"] = tempOutNames;
  89
  90                         //if the user changes the input directory command factory will send this info to us in the output parameter
  91                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  92                         if (inputDir == "not found"){   inputDir = "";          }
  93                         else {
  94                                 string path;
  95                                 it = parameters.find("fastq");
  96                                 //user has given a template file
  97                                 if(it != parameters.end()){
  98                                         path = m->hasPath(it->second);
  99                                         //if the user has not given a path then, add inputdir. else leave path alone.
 100                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 101                                 }
 102                         }
 103
 104                         //check for required parameters
 105                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 106                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 107                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 108
 109                         //if the user changes the output directory command factory will send this info to us in the output parameter
 110                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 111
 112                         string temp;
 113                         temp = validParameter.validFile(parameters, "fasta", false);    if(temp == "not found"){        temp = "T";     }
 114                         fasta = m->isTrue(temp);
 115
 116                         temp = validParameter.validFile(parameters, "qfile", false);    if(temp == "not found"){        temp = "T";     }
 117                         qual = m->isTrue(temp);
 118
 119                         if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
 120
 121                 }
 122         }
 123         catch(exception& e) {
 124                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 125                 exit(1);
 126         }
 127 }
 128 //**********************************************************************************************************************
 129
 130 int ParseFastaQCommand::execute(){
 131         try {
 132                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 133
 134                 //open Output Files
 135                 string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "fasta";
 136                 string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "qual";
 137                 ofstream outFasta, outQual;
 138
 139                 if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);      }
 140                 if (qual) { m->openOutputFile(qualFile, outQual);       outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);             }
 141
 142                 ifstream in;
 143                 m->openInputFile(fastaQFile, in);
 144
 145                 while (!in.eof()) {
 146
 147                         if (m->control_pressed) { break; }
 148
 149                         //read sequence name
 150                         string name = m->getline(in); m->gobble(in);
 151                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 152                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 153                         else { name = name.substr(1); }
 154
 155                         //read sequence
 156                         string sequence = m->getline(in); m->gobble(in);
 157                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 158
 159                         //read sequence name
 160                         string name2 = m->getline(in); m->gobble(in);
 161                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 162                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 163                         else { name2 = name2.substr(1);  }
 164
 165                         //read quality scores
 166                         string quality = m->getline(in); m->gobble(in);
 167                         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 168
 169                         //sanity check sequence length and number of quality scores match
 170                         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; } }
 171                         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; }
 172
 173                         //print sequence info to files
 174                         if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
 175
 176                         if (qual) {
 177                                 vector<int> qualScores = convertQual(quality);
 178                                 outQual << ">" << name << endl;
 179                                 for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 180                                 outQual << endl;
 181                         }
 182                 }
 183
 184                 in.close();
 185                 if (fasta)      { outFasta.close();     }
 186                 if (qual)       { outQual.close();      }
 187
 188                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
 189
 190                 //set fasta file as new current fastafile
 191                 string current = "";
 192                 itTypes = outputTypes.find("fasta");
 193                 if (itTypes != outputTypes.end()) {
 194                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 195                 }
 196
 197                 itTypes = outputTypes.find("qfile");
 198                 if (itTypes != outputTypes.end()) {
 199                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 200                 }
 201
 202                 m->mothurOutEndLine();
 203                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 204                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 205                 m->mothurOutEndLine();
 206
 207                 return 0;
 208         }
 209         catch(exception& e) {
 210                 m->errorOut(e, "ParseFastaQCommand", "execute");
 211                 exit(1);
 212         }
 213 }
 214 //**********************************************************************************************************************
 215 vector<int> ParseFastaQCommand::convertQual(string qual) {
 216         try {
 217                 vector<int> qualScores;
 218
 219                 int controlChar = int('@');
 220
 221                 for (int i = 0; i < qual.length(); i++) {
 222                         int temp = int(qual[i]);
 223                         temp -= controlChar;
 224
 225                         qualScores.push_back(temp);
 226                 }
 227
 228                 return qualScores;
 229         }
 230         catch(exception& e) {
 231                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 232                 exit(1);
 233         }
 234 }
 235 //**********************************************************************************************************************
 236
 237
 238