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;
  29                 //initialize outputTypes
  30                 vector<string> tempOutNames;
  31                 outputTypes["fasta"] = tempOutNames;
  32                 outputTypes["qual"] = tempOutNames;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 vector<string> ParseFastaQCommand::getRequiredParameters(){
  41         try {
  42                 string Array[] =  {"fastq"};
  43                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  44                 return myArray;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ParseFastaQCommand", "getRequiredParameters");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 vector<string> ParseFastaQCommand::getRequiredFiles(){
  53         try {
  54                 vector<string> myArray;
  55                 return myArray;
  56         }
  57         catch(exception& e) {
  58                 m->errorOut(e, "ParseFastaQCommand", "getRequiredFiles");
  59                 exit(1);
  60         }
  61 }
  62 //**********************************************************************************************************************
  63 ParseFastaQCommand::ParseFastaQCommand(string option){
  64         try {
  65                 abort = false;
  66
  67                 if(option == "help") {  help(); abort = true; }
  68
  69                 else {
  70                         //valid paramters for this command
  71                         string Array[] =  {"fastq", "outputdir", "inputdir"};
  72                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  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["qual"] = 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                 }
 113         }
 114         catch(exception& e) {
 115                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 116                 exit(1);
 117         }
 118 }
 119 //**********************************************************************************************************************
 120
 121 void ParseFastaQCommand::help(){
 122         try {
 123                 m->mothurOut("The fastq.info command reads a fastaQ file and creates a fasta and quality file.\n");
 124                 m->mothurOut("The fastq.info command parameter is fastq, and it is required.\n");
 125                 m->mothurOut("The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n");
 126                 m->mothurOut("Example fastq.info(fastaq=test.fastaq).\n");
 127                 m->mothurOut("Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n");
 128                 m->mothurOutEndLine();
 129         }
 130         catch(exception& e) {
 131                 m->errorOut(e, "ParseFastaQCommand", "help");
 132                 exit(1);
 133         }
 134 }
 135 //**********************************************************************************************************************
 136
 137 ParseFastaQCommand::~ParseFastaQCommand()       {       /*      do nothing      */      }
 138
 139 //**********************************************************************************************************************
 140
 141 int ParseFastaQCommand::execute(){
 142         try {
 143                 if (abort == true) {    return 0;       }
 144
 145                 //open Output Files
 146                 string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "fasta";
 147                 string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + "qual";
 148                 ofstream outFasta, outQual;
 149                 m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);
 150                 m->openOutputFile(qualFile, outQual);   outputNames.push_back(qualFile);  outputTypes["qual"].push_back(qualFile);
 151
 152                 ifstream in;
 153                 m->openInputFile(fastaQFile, in);
 154
 155                 while (!in.eof()) {
 156
 157                         //read sequence name
 158                         string name = m->getline(in); m->gobble(in);
 159                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 160                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 161                         else { name = name.substr(1); }
 162
 163                         //read sequence
 164                         string sequence = m->getline(in); m->gobble(in);
 165                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 166
 167                         //read sequence name
 168                         string name2 = m->getline(in); m->gobble(in);
 169                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 170                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 171                         else { name2 = name2.substr(1);  }
 172
 173                         //read quality scores
 174                         string qual = m->getline(in); m->gobble(in);
 175                         if (qual == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 176
 177                         //sanity check sequence length and number of quality scores match
 178                         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; }
 179                         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; }
 180
 181                         //convert quality scores
 182                         vector<int> qualScores = convertQual(qual);
 183
 184                         //print sequence info to files
 185                         outFasta << ">" << name << endl << sequence << endl;
 186
 187                         outQual << ">" << name << endl;
 188                         for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 189                         outQual << endl;
 190                 }
 191
 192                 in.close();
 193                 outFasta.close();
 194                 outQual.close();
 195
 196                 if (m->control_pressed) { outputTypes.clear(); remove(fastaFile.c_str()); remove(qualFile.c_str()); return 0; }
 197
 198                 m->mothurOutEndLine();
 199                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 200                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 201                 m->mothurOutEndLine();
 202
 203                 return 0;
 204         }
 205         catch(exception& e) {
 206                 m->errorOut(e, "ParseFastaQCommand", "execute");
 207                 exit(1);
 208         }
 209 }
 210 //**********************************************************************************************************************
 211 vector<int> ParseFastaQCommand::convertQual(string qual) {
 212         try {
 213                 vector<int> qualScores;
 214
 215                 int controlChar = int('!');
 216
 217                 for (int i = 0; i < qual.length(); i++) {
 218                         int temp = int(qual[i]);
 219                         temp -= controlChar;
 220
 221                         qualScores.push_back(temp);
 222                 }
 223
 224                 return qualScores;
 225         }
 226         catch(exception& e) {
 227                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 228                 exit(1);
 229         }
 230 }
 231 //**********************************************************************************************************************
 232
 233
 234