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 string ParseFastaQCommand::getOutputFileNameTag(string type, string inputName=""){
  51         try {
  52         string outputFileName = "";
  53                 map<string, vector<string> >::iterator it;
  54
  55         //is this a type this command creates
  56         it = outputTypes.find(type);
  57         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  58         else {
  59             if (type == "fasta") {  outputFileName =  "fasta"; }
  60             else if (type == "qfile") {  outputFileName =  "qual"; }
  61             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  62         }
  63         return outputFileName;
  64         }
  65         catch(exception& e) {
  66                 m->errorOut(e, "ParseFastaQCommand", "getOutputFileNameTag");
  67                 exit(1);
  68         }
  69 }
  70
  71 //**********************************************************************************************************************
  72 ParseFastaQCommand::ParseFastaQCommand(){
  73         try {
  74                 abort = true; calledHelp = true;
  75                 setParameters();
  76                 vector<string> tempOutNames;
  77                 outputTypes["fasta"] = tempOutNames;
  78                 outputTypes["qfile"] = tempOutNames;
  79         }
  80         catch(exception& e) {
  81                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  82                 exit(1);
  83         }
  84 }
  85 //**********************************************************************************************************************
  86 ParseFastaQCommand::ParseFastaQCommand(string option){
  87         try {
  88                 abort = false; calledHelp = false;
  89
  90                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  91                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  92
  93                 else {
  94                         vector<string> myArray = setParameters();
  95
  96                         OptionParser parser(option);
  97                         map<string,string> parameters = parser.getParameters();
  98
  99                         ValidParameters validParameter;
 100                         map<string,string>::iterator it;
 101
 102                         //check to make sure all parameters are valid for command
 103                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 104                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 105                         }
 106
 107                         //initialize outputTypes
 108                         vector<string> tempOutNames;
 109                         outputTypes["fasta"] = tempOutNames;
 110                         outputTypes["qfile"] = tempOutNames;
 111
 112                         //if the user changes the input directory command factory will send this info to us in the output parameter
 113                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 114                         if (inputDir == "not found"){   inputDir = "";          }
 115                         else {
 116                                 string path;
 117                                 it = parameters.find("fastq");
 118                                 //user has given a template file
 119                                 if(it != parameters.end()){
 120                                         path = m->hasPath(it->second);
 121                                         //if the user has not given a path then, add inputdir. else leave path alone.
 122                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 123                                 }
 124                         }
 125
 126                         //check for required parameters
 127                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 128                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 129                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 130
 131                         //if the user changes the output directory command factory will send this info to us in the output parameter
 132                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 133
 134                         string temp;
 135                         temp = validParameter.validFile(parameters, "fasta", false);    if(temp == "not found"){        temp = "T";     }
 136                         fasta = m->isTrue(temp);
 137
 138                         temp = validParameter.validFile(parameters, "qfile", false);    if(temp == "not found"){        temp = "T";     }
 139                         qual = m->isTrue(temp);
 140
 141                         if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
 142
 143                 }
 144         }
 145         catch(exception& e) {
 146                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 147                 exit(1);
 148         }
 149 }
 150 //**********************************************************************************************************************
 151
 152 int ParseFastaQCommand::execute(){
 153         try {
 154                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 155
 156                 //open Output Files
 157                 string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("fasta");
 158                 string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("qfile");
 159                 ofstream outFasta, outQual;
 160
 161                 if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);      }
 162                 if (qual) { m->openOutputFile(qualFile, outQual);       outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);             }
 163
 164                 ifstream in;
 165                 m->openInputFile(fastaQFile, in);
 166
 167                 while (!in.eof()) {
 168
 169                         if (m->control_pressed) { break; }
 170
 171                         //read sequence name
 172                         string name = m->getline(in); m->gobble(in);
 173                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 174                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 175                         else { name = name.substr(1); }
 176
 177                         //read sequence
 178                         string sequence = m->getline(in); m->gobble(in);
 179                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 180
 181                         //read sequence name
 182                         string name2 = m->getline(in); m->gobble(in);
 183                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 184                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 185                         else { name2 = name2.substr(1);  }
 186
 187                         //read quality scores
 188                         string quality = m->getline(in); m->gobble(in);
 189                         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 190
 191                         //sanity check sequence length and number of quality scores match
 192                         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; } }
 193                         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; }
 194
 195                         //print sequence info to files
 196                         if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
 197
 198                         if (qual) {
 199                                 vector<int> qualScores = convertQual(quality);
 200                                 outQual << ">" << name << endl;
 201                                 for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 202                                 outQual << endl;
 203                         }
 204                 }
 205
 206                 in.close();
 207                 if (fasta)      { outFasta.close();     }
 208                 if (qual)       { outQual.close();      }
 209
 210                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
 211
 212                 //set fasta file as new current fastafile
 213                 string current = "";
 214                 itTypes = outputTypes.find("fasta");
 215                 if (itTypes != outputTypes.end()) {
 216                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 217                 }
 218
 219                 itTypes = outputTypes.find("qfile");
 220                 if (itTypes != outputTypes.end()) {
 221                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 222                 }
 223
 224                 m->mothurOutEndLine();
 225                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 226                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 227                 m->mothurOutEndLine();
 228
 229                 return 0;
 230         }
 231         catch(exception& e) {
 232                 m->errorOut(e, "ParseFastaQCommand", "execute");
 233                 exit(1);
 234         }
 235 }
 236 //**********************************************************************************************************************
 237 vector<int> ParseFastaQCommand::convertQual(string qual) {
 238         try {
 239                 vector<int> qualScores;
 240
 241                 int controlChar = int('@');
 242
 243                 for (int i = 0; i < qual.length(); i++) {
 244                         int temp = int(qual[i]);
 245                         temp -= controlChar;
 246
 247                         qualScores.push_back(temp);
 248                 }
 249
 250                 return qualScores;
 251         }
 252         catch(exception& e) {
 253                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 254                 exit(1);
 255         }
 256 }
 257 //**********************************************************************************************************************
 258
 259
 260