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 pformat("format", "Multiple", "sanger-illumina-solexa", "sanger", "", "", "",false,false); parameters.push_back(pformat);
  20         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  21                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  22
  23                 vector<string> myArray;
  24                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  25                 return myArray;
  26         }
  27         catch(exception& e) {
  28                 m->errorOut(e, "ParseFastaQCommand", "setParameters");
  29                 exit(1);
  30         }
  31 }
  32 //**********************************************************************************************************************
  33 string ParseFastaQCommand::getHelpString(){
  34         try {
  35                 string helpString = "";
  36                 helpString += "The fastq.info command reads a fastq file and creates a fasta and quality file.\n";
  37                 helpString += "The fastq.info command parameters are fastq, fasta, qfile and format; fastq is required.\n";
  38         helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
  39                 helpString += "The format parameter is used to indicate whether your sequences are sanger, solexa or illumina, default=sanger.\n";
  40         helpString += "The fasta parameter allows you to indicate whether you want a fasta file generated. Default=T.\n";
  41         helpString += "The qfile parameter allows you to indicate whether you want a quality file generated. Default=T.\n";
  42                 helpString += "Example fastq.info(fastaq=test.fastaq).\n";
  43                 helpString += "Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n";
  44                 return helpString;
  45         }
  46         catch(exception& e) {
  47                 m->errorOut(e, "ParseFastaQCommand", "getHelpString");
  48                 exit(1);
  49         }
  50 }
  51 //**********************************************************************************************************************
  52 string ParseFastaQCommand::getOutputFileNameTag(string type, string inputName=""){
  53         try {
  54         string outputFileName = "";
  55                 map<string, vector<string> >::iterator it;
  56
  57         //is this a type this command creates
  58         it = outputTypes.find(type);
  59         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  60         else {
  61             if (type == "fasta") {  outputFileName =  "fasta"; }
  62             else if (type == "qfile") {  outputFileName =  "qual"; }
  63             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  64         }
  65         return outputFileName;
  66         }
  67         catch(exception& e) {
  68                 m->errorOut(e, "ParseFastaQCommand", "getOutputFileNameTag");
  69                 exit(1);
  70         }
  71 }
  72
  73 //**********************************************************************************************************************
  74 ParseFastaQCommand::ParseFastaQCommand(){
  75         try {
  76                 abort = true; calledHelp = true;
  77                 setParameters();
  78                 vector<string> tempOutNames;
  79                 outputTypes["fasta"] = tempOutNames;
  80                 outputTypes["qfile"] = tempOutNames;
  81         }
  82         catch(exception& e) {
  83                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  84                 exit(1);
  85         }
  86 }
  87 //**********************************************************************************************************************
  88 ParseFastaQCommand::ParseFastaQCommand(string option){
  89         try {
  90                 abort = false; calledHelp = false;
  91
  92                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  93                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  94
  95                 else {
  96                         vector<string> myArray = setParameters();
  97
  98                         OptionParser parser(option);
  99                         map<string,string> parameters = parser.getParameters();
 100
 101                         ValidParameters validParameter;
 102                         map<string,string>::iterator it;
 103
 104                         //check to make sure all parameters are valid for command
 105                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 106                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 107                         }
 108
 109                         //initialize outputTypes
 110                         vector<string> tempOutNames;
 111                         outputTypes["fasta"] = tempOutNames;
 112                         outputTypes["qfile"] = tempOutNames;
 113
 114                         //if the user changes the input directory command factory will send this info to us in the output parameter
 115                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 116                         if (inputDir == "not found"){   inputDir = "";          }
 117                         else {
 118                                 string path;
 119                                 it = parameters.find("fastq");
 120                                 //user has given a template file
 121                                 if(it != parameters.end()){
 122                                         path = m->hasPath(it->second);
 123                                         //if the user has not given a path then, add inputdir. else leave path alone.
 124                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 125                                 }
 126                         }
 127
 128                         //check for required parameters
 129                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 130                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 131                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 132
 133                         //if the user changes the output directory command factory will send this info to us in the output parameter
 134                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 135
 136                         string temp;
 137                         temp = validParameter.validFile(parameters, "fasta", false);    if(temp == "not found"){        temp = "T";     }
 138                         fasta = m->isTrue(temp);
 139
 140                         temp = validParameter.validFile(parameters, "qfile", false);    if(temp == "not found"){        temp = "T";     }
 141                         qual = m->isTrue(temp);
 142
 143             format = validParameter.validFile(parameters, "format", false);             if (format == "not found"){     format = "sanger";      }
 144
 145             if ((format != "sanger") && (format != "illumina") && (format != "solexa"))  {
 146                                 m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa and illumina, aborting." ); m->mothurOutEndLine();
 147                                 abort=true;
 148                         }
 149
 150                         if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
 151
 152                 }
 153         }
 154         catch(exception& e) {
 155                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 156                 exit(1);
 157         }
 158 }
 159 //**********************************************************************************************************************
 160
 161 int ParseFastaQCommand::execute(){
 162         try {
 163                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 164
 165                 //open Output Files
 166                 string fastaFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("fasta");
 167                 string qualFile = outputDir + m->getRootName(m->getSimpleName(fastaQFile)) + getOutputFileNameTag("qfile");
 168                 ofstream outFasta, outQual;
 169
 170                 if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);      }
 171                 if (qual) { m->openOutputFile(qualFile, outQual);       outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);             }
 172
 173                 ifstream in;
 174                 m->openInputFile(fastaQFile, in);
 175
 176         //fill convert table - goes from solexa to sanger. Used fq_all2std.pl as a reference.
 177         for (int i = -64; i < 65; i++) {
 178             char temp = (char) ((int)(33 + 10*log(1+pow(10,(i/10.0)))/log(10)+0.499));
 179             convertTable.push_back(temp);
 180         }
 181
 182                 while (!in.eof()) {
 183
 184                         if (m->control_pressed) { break; }
 185
 186                         //read sequence name
 187                         string name = m->getline(in); m->gobble(in);
 188                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 189                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 190                         else { name = name.substr(1); }
 191
 192                         //read sequence
 193                         string sequence = m->getline(in); m->gobble(in);
 194                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 195
 196                         //read sequence name
 197                         string name2 = m->getline(in); m->gobble(in);
 198                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 199                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 200                         else { name2 = name2.substr(1);  }
 201
 202                         //read quality scores
 203                         string quality = m->getline(in); m->gobble(in);
 204                         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 205
 206                         //sanity check sequence length and number of quality scores match
 207                         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; } }
 208                         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; }
 209
 210                         //print sequence info to files
 211                         if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
 212
 213                         if (qual) {
 214                                 vector<int> qualScores = convertQual(quality);
 215                                 outQual << ">" << name << endl;
 216                                 for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 217                                 outQual << endl;
 218                         }
 219                 }
 220
 221                 in.close();
 222                 if (fasta)      { outFasta.close();     }
 223                 if (qual)       { outQual.close();      }
 224
 225                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
 226
 227                 //set fasta file as new current fastafile
 228                 string current = "";
 229                 itTypes = outputTypes.find("fasta");
 230                 if (itTypes != outputTypes.end()) {
 231                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 232                 }
 233
 234                 itTypes = outputTypes.find("qfile");
 235                 if (itTypes != outputTypes.end()) {
 236                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 237                 }
 238
 239                 m->mothurOutEndLine();
 240                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 241                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 242                 m->mothurOutEndLine();
 243
 244                 return 0;
 245         }
 246         catch(exception& e) {
 247                 m->errorOut(e, "ParseFastaQCommand", "execute");
 248                 exit(1);
 249         }
 250 }
 251 //**********************************************************************************************************************
 252 vector<int> ParseFastaQCommand::convertQual(string qual) {
 253         try {
 254                 vector<int> qualScores;
 255
 256                 for (int i = 0; i < qual.length(); i++) {
 257
 258             int temp = 0;
 259             temp = int(qual[i]);
 260             if (format == "illumina") {
 261                 temp -= 64; //char '@'
 262             }else if (format == "solexa") {
 263                 temp = int(convertTable[temp]); //convert to sanger
 264                 temp -= 33; //char '!'
 265             }else {
 266                 temp -= 33; //char '!'
 267             }
 268                         qualScores.push_back(temp);
 269                 }
 270
 271                 return qualScores;
 272         }
 273         catch(exception& e) {
 274                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 275                 exit(1);
 276         }
 277 }
 278 //**********************************************************************************************************************
 279
 280
 281