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 ppacbio("pacbio", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(ppacbio);
  20                 CommandParameter pformat("format", "Multiple", "sanger-illumina-solexa-illumina1.8+", "sanger", "", "", "","",false,false,true); parameters.push_back(pformat);
  21         CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "ParseFastaQCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string ParseFastaQCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The fastq.info command reads a fastq file and creates a fasta and quality file.\n";
  38                 helpString += "The fastq.info command parameters are fastq, fasta, qfile and format; fastq is required.\n";
  39         helpString += "The fastq.info command should be in the following format: fastq.info(fastaq=yourFastaQFile).\n";
  40                 helpString += "The format parameter is used to indicate whether your sequences are sanger, solexa, illumina1.8+ or illumina, default=sanger.\n";
  41         helpString += "The fasta parameter allows you to indicate whether you want a fasta file generated. Default=T.\n";
  42         helpString += "The qfile parameter allows you to indicate whether you want a quality file generated. Default=T.\n";
  43         helpString += "The pacbio parameter allows you to indicate .... When set to true, quality scores of 0 will results in a corresponding base of N. Default=F.\n";
  44                 helpString += "Example fastq.info(fastaq=test.fastaq).\n";
  45                 helpString += "Note: No spaces between parameter labels (i.e. fastq), '=' and yourFastQFile.\n";
  46                 return helpString;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "ParseFastaQCommand", "getHelpString");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 string ParseFastaQCommand::getOutputPattern(string type) {
  55     try {
  56         string pattern = "";
  57
  58         if (type == "fasta") {  pattern = "[filename],fasta"; }
  59         else if (type == "qfile") {  pattern = "[filename],qual"; }
  60         else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true;  }
  61
  62         return pattern;
  63     }
  64     catch(exception& e) {
  65         m->errorOut(e, "ParseFastaQCommand", "getOutputPattern");
  66         exit(1);
  67     }
  68 }
  69 //**********************************************************************************************************************
  70 ParseFastaQCommand::ParseFastaQCommand(){
  71         try {
  72                 abort = true; calledHelp = true;
  73                 setParameters();
  74                 vector<string> tempOutNames;
  75                 outputTypes["fasta"] = tempOutNames;
  76                 outputTypes["qfile"] = tempOutNames;
  77         }
  78         catch(exception& e) {
  79                 m->errorOut(e, "ParseFastaQCommand", "ParseFastaQCommand");
  80                 exit(1);
  81         }
  82 }
  83 //**********************************************************************************************************************
  84 ParseFastaQCommand::ParseFastaQCommand(string option){
  85         try {
  86                 abort = false; calledHelp = false;
  87
  88                 if(option == "help") {  help(); abort = true; calledHelp = true; }
  89                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  90
  91                 else {
  92                         vector<string> myArray = setParameters();
  93
  94                         OptionParser parser(option);
  95                         map<string,string> parameters = parser.getParameters();
  96
  97                         ValidParameters validParameter;
  98                         map<string,string>::iterator it;
  99
 100                         //check to make sure all parameters are valid for command
 101                         for (map<string,string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
 102                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 103                         }
 104
 105                         //initialize outputTypes
 106                         vector<string> tempOutNames;
 107                         outputTypes["fasta"] = tempOutNames;
 108                         outputTypes["qfile"] = tempOutNames;
 109
 110                         //if the user changes the input directory command factory will send this info to us in the output parameter
 111                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 112                         if (inputDir == "not found"){   inputDir = "";          }
 113                         else {
 114                                 string path;
 115                                 it = parameters.find("fastq");
 116                                 //user has given a template file
 117                                 if(it != parameters.end()){
 118                                         path = m->hasPath(it->second);
 119                                         //if the user has not given a path then, add inputdir. else leave path alone.
 120                                         if (path == "") {       parameters["fastq"] = inputDir + it->second;            }
 121                                 }
 122                         }
 123
 124                         //check for required parameters
 125                         fastaQFile = validParameter.validFile(parameters, "fastq", true);
 126                         if (fastaQFile == "not found") {        m->mothurOut("fastq is a required parameter for the fastq.info command.");      m->mothurOutEndLine();  abort = true;   }
 127                         else if (fastaQFile == "not open")      {       fastaQFile = ""; abort = true;  }
 128
 129                         //if the user changes the output directory command factory will send this info to us in the output parameter
 130                         outputDir = validParameter.validFile(parameters, "outputdir", false);   if (outputDir == "not found"){  outputDir = m->hasPath(fastaQFile);     }
 131
 132                         string temp;
 133                         temp = validParameter.validFile(parameters, "fasta", false);    if(temp == "not found"){        temp = "T";     }
 134                         fasta = m->isTrue(temp);
 135
 136                         temp = validParameter.validFile(parameters, "qfile", false);    if(temp == "not found"){        temp = "T";     }
 137                         qual = m->isTrue(temp);
 138
 139             temp = validParameter.validFile(parameters, "pacbio", false);       if(temp == "not found"){        temp = "F";     }
 140                         pacbio = m->isTrue(temp);
 141
 142
 143             format = validParameter.validFile(parameters, "format", false);             if (format == "not found"){     format = "sanger";      }
 144
 145             if ((format != "sanger") && (format != "illumina") && (format != "illumina1.8+") && (format != "solexa"))  {
 146                                 m->mothurOut(format + " is not a valid format. Your format choices are sanger, solexa, illumina1.8+ 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         map<string, string> variables;
 167         variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(fastaQFile));
 168                 string fastaFile = getOutputFileName("fasta",variables);
 169                 string qualFile = getOutputFileName("qfile",variables);
 170                 ofstream outFasta, outQual;
 171
 172                 if (fasta) { m->openOutputFile(fastaFile, outFasta);  outputNames.push_back(fastaFile); outputTypes["fasta"].push_back(fastaFile);      }
 173                 if (qual) { m->openOutputFile(qualFile, outQual);       outputNames.push_back(qualFile);  outputTypes["qfile"].push_back(qualFile);             }
 174
 175                 ifstream in;
 176                 m->openInputFile(fastaQFile, in);
 177
 178         //fill convert table - goes from solexa to sanger. Used fq_all2std.pl as a reference.
 179         for (int i = -64; i < 65; i++) {
 180             char temp = (char) ((int)(33 + 10*log(1+pow(10,(i/10.0)))/log(10)+0.499));
 181             convertTable.push_back(temp);
 182         }
 183
 184                 while (!in.eof()) {
 185
 186                         if (m->control_pressed) { break; }
 187
 188                         //read sequence name
 189                         string name = m->getline(in); m->gobble(in);
 190                         if (name == "") {  m->mothurOut("[ERROR]: Blank fasta name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 191                         else if (name[0] != '@') { m->mothurOut("[ERROR]: reading " + name + " expected a name with @ as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 192                         else {
 193                 name = name.substr(1);
 194                 m->checkName(name);
 195             }
 196
 197                         //read sequence
 198                         string sequence = m->getline(in); m->gobble(in);
 199                         if (sequence == "") {  m->mothurOut("[ERROR]: missing sequence for " + name); m->mothurOutEndLine(); m->control_pressed = true; break; }
 200
 201                         //read sequence name
 202                         string name2 = m->getline(in); m->gobble(in);
 203                         if (name2 == "") {  m->mothurOut("[ERROR]: Blank quality name."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 204                         else if (name2[0] != '+') { m->mothurOut("[ERROR]: reading " + name2 + " expected a name with + as a leading character."); m->mothurOutEndLine(); m->control_pressed = true; break; }
 205                         else {
 206                 name2 = name2.substr(1);
 207                 m->checkName(name2);
 208             }
 209
 210                         //read quality scores
 211                         string quality = m->getline(in); m->gobble(in);
 212                         if (quality == "") {  m->mothurOut("[ERROR]: missing quality for " + name2); m->mothurOutEndLine(); m->control_pressed = true; break; }
 213
 214                         //sanity check sequence length and number of quality scores match
 215                         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; } }
 216                         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; }
 217
 218             vector<int> qualScores;
 219             if (qual) {
 220                                 qualScores = convertQual(quality);
 221                                 outQual << ">" << name << endl;
 222                                 for (int i = 0; i < qualScores.size(); i++) { outQual << qualScores[i] << " "; }
 223                                 outQual << endl;
 224                         }
 225
 226             if (m->control_pressed) { break; }
 227
 228             if (pacbio) {
 229                 if (!qual) { qualScores = convertQual(quality); } //get scores if we didn't already
 230                 for (int i = 0; i < qualScores.size(); i++) {
 231                     if (qualScores[i] == 0){ sequence[i] = 'N'; }
 232                 }
 233             }
 234
 235                         //print sequence info to files
 236                         if (fasta) { outFasta << ">" << name << endl << sequence << endl; }
 237
 238                 }
 239
 240                 in.close();
 241                 if (fasta)      { outFasta.close();     }
 242                 if (qual)       { outQual.close();      }
 243
 244                 if (m->control_pressed) { outputTypes.clear(); outputNames.clear(); m->mothurRemove(fastaFile); m->mothurRemove(qualFile); return 0; }
 245
 246                 //set fasta file as new current fastafile
 247                 string current = "";
 248                 itTypes = outputTypes.find("fasta");
 249                 if (itTypes != outputTypes.end()) {
 250                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 251                 }
 252
 253                 itTypes = outputTypes.find("qfile");
 254                 if (itTypes != outputTypes.end()) {
 255                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setQualFile(current); }
 256                 }
 257
 258                 m->mothurOutEndLine();
 259                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 260                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 261                 m->mothurOutEndLine();
 262
 263                 return 0;
 264         }
 265         catch(exception& e) {
 266                 m->errorOut(e, "ParseFastaQCommand", "execute");
 267                 exit(1);
 268         }
 269 }
 270 //**********************************************************************************************************************
 271 vector<int> ParseFastaQCommand::convertQual(string qual) {
 272         try {
 273                 vector<int> qualScores;
 274
 275         bool negativeScores = false;
 276
 277                 for (int i = 0; i < qual.length(); i++) {
 278
 279             int temp = 0;
 280             temp = int(qual[i]);
 281             if (format == "illumina") {
 282                 temp -= 64; //char '@'
 283             }else if (format == "illumina1.8+") {
 284                 temp -= int('!'); //char '!'
 285             }else if (format == "solexa") {
 286                 temp = int(convertTable[temp]); //convert to sanger
 287                 temp -= int('!'); //char '!'
 288             }else {
 289                 temp -= int('!'); //char '!'
 290             }
 291             if (temp < -5) { negativeScores = true; }
 292                         qualScores.push_back(temp);
 293                 }
 294
 295         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;  }
 296
 297                 return qualScores;
 298         }
 299         catch(exception& e) {
 300                 m->errorOut(e, "ParseFastaQCommand", "convertQual");
 301                 exit(1);
 302         }
 303 }
 304 //**********************************************************************************************************************
 305
 306
 307