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,true); parameters.push_back(pfastq);
  17                 CommandParameter pfasta("fasta", "Boolean", "", "T", "", "", "","fasta",false,false); parameters.push_back(pfasta);
  18                 CommandParameter pqual("qfile", "Boolean", "", "T", "", "", "","qfile",false,false); parameters.push_back(pqual);
  19                 CommandParameter pformat("format", "Multiple", "sanger-illumina-solexa-illumina1.8+", "sanger", "", "", "","",false,false,true); 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, illumina1.8+ 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::getOutputPattern(string type) {
  53     try {
  54         string pattern = "";
  55
  56         if (type == "fasta") {  pattern = "[filename],fasta"; }
  57         else if (type == "qfile") {  pattern = "[filename],qual"; }
  58         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  59
  60         return pattern;
  61     }
  62     catch(exception& e) {
  63         m->errorOut(e, "ParseFastaQCommand", "getOutputPattern");
  64         exit(1);
  65     }
  66 }
  67 //**********************************************************************************************************************
  68 ParseFastaQCommand::ParseFastaQCommand(){
  69         try {
  70                 abort = true; calledHelp = true;
  71                 setParameters();
  72                 vector<string> tempOutNames;
  73                 outputTypes["fasta"] = tempOutNames;
  74                 outputTypes["qfile"] = tempOutNames;
  75         }
  76         catch(exception& e) {
  77                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  78                 exit(1);
  79         }
  80 }
  81 //**********************************************************************************************************************
  82 ParseFastaQCommand::ParseFastaQCommand(string option){
  83         try {
  84                 abort = false; calledHelp = false;
  85
  86                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  87                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  88
  89                 else {
  90                         vector<string> myArray = setParameters();
  91
  92                         OptionParser parser(option);
  93                         map<string,string> parameters = parser.getParameters();
  94
  95                         ValidParameters validParameter;
  96                         map<string,string>::iterator it;
  97
  98                         //check to make sure all parameters are valid for command
  99                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 100                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 101                         }
 102
 103                         //initialize outputTypes
 104                         vector<string> tempOutNames;
 105                         outputTypes["fasta"] = tempOutNames;
 106                         outputTypes["qfile"] = tempOutNames;
 107
 108                         //if the user changes the input directory command factory will send this info to us in the output parameter
 109                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 110                         if (inputDir == "not found"){   inputDir = "";          }
 111                         else {
 112                                 string path;
 113                                 it = parameters.find("fastq");
 114                                 //user has given a template file
 115                                 if(it != parameters.end()){
 116                                         path = m->hasPath(it->second);
 117                                         //if the user has not given a path then, add inputdir. else leave path alone.
 118                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 119                                 }
 120                         }
 121
 122                         //check for required parameters
 123                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 124                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 125                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 126
 127                         //if the user changes the output directory command factory will send this info to us in the output parameter
 128                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 129
 130                         string temp;
 131                         temp = validParameter.validFile(parameters, "fasta", false);    if(temp == "not found"){        temp = "T";     }
 132                         fasta = m->isTrue(temp);
 133
 134                         temp = validParameter.validFile(parameters, "qfile", false);    if(temp == "not found"){        temp = "T";     }
 135                         qual = m->isTrue(temp);
 136
 137             format = validParameter.validFile(parameters, "format", false);             if (format == "not found"){     format = "sanger";      }
 138
 139             if ((format != "sanger") && (format != "illumina") && (format != "illumina1.8+") && (format != "solexa"))  {
 140                                 m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa, illumina1.8+ and illumina, aborting." ); m->mothurOutEndLine();
 141                                 abort=true;
 142                         }
 143
 144                         if ((!fasta) && (!qual)) { m->mothurOut("[ERROR]: no outputs selected. Aborting."); m->mothurOutEndLine(); abort=true; }
 145
 146                 }
 147         }
 148         catch(exception& e) {
 149                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
 150                 exit(1);
 151         }
 152 }
 153 //**********************************************************************************************************************
 154
 155 int ParseFastaQCommand::execute(){
 156         try {
 157                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 158
 159                 //open Output Files
 160         map<string, string> variables;
 161         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
 162                 string fastaFile = getOutputFileName("fasta",variables);
 163                 string qualFile = getOutputFileName("qfile",variables);
 164                 ofstream outFasta, outQual;
 165
 166                 if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);      }
 167                 if (qual) { m->openOutputFile(qualFile, outQual);       outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);             }
 168
 169                 ifstream in;
 170                 m->openInputFile(fastaQFile, in);
 171
 172         //fill convert table - goes from solexa to sanger. Used fq_all2std.pl as a reference.
 173         for (int i = -64; i < 65; i++) {
 174             char temp = (char) ((int)(33 + 10*log(1+pow(10,(i/10.0)))/log(10)+0.499));
 175             convertTable.push_back(temp);
 176         }
 177
 178                 while (!in.eof()) {
 179
 180                         if (m->control_pressed) { break; }
 181
 182                         //read sequence name
 183                         string name = m->getline(in); m->gobble(in);
 184                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 185                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 186                         else {
 187                 name = name.substr(1);
 188                 for (int i = 0; i < name.length(); i++) { if (name[i] == ':') { name[i] = '_'; m->changedSeqNames = true; } }
 189             }
 190
 191                         //read sequence
 192                         string sequence = m->getline(in); m->gobble(in);
 193                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 194
 195                         //read sequence name
 196                         string name2 = m->getline(in); m->gobble(in);
 197                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 198                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 199                         else {
 200                 name2 = name2.substr(1);
 201                 for (int i = 0; i < name2.length(); i++) { if (name2[i] == ':') { name2[i] = '_'; m->changedSeqNames = true; } }
 202             }
 203
 204                         //read quality scores
 205                         string quality = m->getline(in); m->gobble(in);
 206                         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 207
 208                         //sanity check sequence length and number of quality scores match
 209                         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; } }
 210                         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; }
 211
 212                         //print sequence info to files
 213                         if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
 214
 215                         if (qual) {
 216                                 vector<int> qualScores = convertQual(quality);
 217                                 outQual << ">" << name << endl;
 218                                 for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 219                                 outQual << endl;
 220                         }
 221                 }
 222
 223                 in.close();
 224                 if (fasta)      { outFasta.close();     }
 225                 if (qual)       { outQual.close();      }
 226
 227                 if (m->control_pressed) { outputTypes.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
 228
 229                 //set fasta file as new current fastafile
 230                 string current = "";
 231                 itTypes = outputTypes.find("fasta");
 232                 if (itTypes != outputTypes.end()) {
 233                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 234                 }
 235
 236                 itTypes = outputTypes.find("qfile");
 237                 if (itTypes != outputTypes.end()) {
 238                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 239                 }
 240
 241                 m->mothurOutEndLine();
 242                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 243                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 244                 m->mothurOutEndLine();
 245
 246                 return 0;
 247         }
 248         catch(exception& e) {
 249                 m->errorOut(e, "ParseFastaQCommand", "execute");
 250                 exit(1);
 251         }
 252 }
 253 //**********************************************************************************************************************
 254 vector<int> ParseFastaQCommand::convertQual(string qual) {
 255         try {
 256                 vector<int> qualScores;
 257
 258         bool negativeScores = false;
 259
 260                 for (int i = 0; i < qual.length(); i++) {
 261
 262             int temp = 0;
 263             temp = int(qual[i]);
 264             if (format == "illumina") {
 265                 temp -= 64; //char '@'
 266             }else if (format == "illumina1.8+") {
 267                 temp -= int('!'); //char '!'
 268             }else if (format == "solexa") {
 269                 temp = int(convertTable[temp]); //convert to sanger
 270                 temp -= int('!'); //char '!'
 271             }else {
 272                 temp -= int('!'); //char '!'
 273             }
 274             if (temp < -5) { negativeScores = true; }
 275                         qualScores.push_back(temp);
 276                 }
 277
 278         if (negativeScores) { m->mothurOut("[ERROR]: finding negative quality scores, do you have the right format selected? http://en.wikipedia.org/wiki/FASTQ_format#Encoding \n");  m->control_pressed = true;  }
 279
 280                 return qualScores;
 281         }
 282         catch(exception& e) {
 283                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 284                 exit(1);
 285         }
 286 }
 287 //**********************************************************************************************************************
 288
 289
 290