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