blastdb.cpp

   1 /*
   2  *  blastdb.cpp
   3  *
   4  *
   5  *  Created by Pat Schloss on 12/22/08.
   6  *  Copyright 2008 Patrick D. Schloss. All rights reserved.
   7  *
   8  */
   9
  10
  11 #include "database.hpp"
  12 #include "sequence.hpp"
  13 #include "blastdb.hpp"
  14
  15 /**************************************************************************************************/
  16
  17 BlastDB::BlastDB(float gO, float gE, float m, float mM) : Database(),
  18 gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) {
  19
  20         globaldata = GlobalData::getInstance();
  21         count = 0;
  22
  23         int randNumber = rand();
  24         dbFileName = toString(randNumber) + ".template.unaligned.fasta";
  25         queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
  26         blastFileName = toString(randNumber) + ".blast";
  27
  28 }
  29 /**************************************************************************************************/
  30
  31 BlastDB::BlastDB() : Database() {
  32
  33         globaldata = GlobalData::getInstance();
  34         count = 0;
  35
  36         int randNumber = rand();
  37         dbFileName = toString(randNumber) + ".template.unaligned.fasta";
  38         queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
  39         blastFileName = toString(randNumber) + ".blast";
  40
  41 }
  42
  43 /**************************************************************************************************/
  44
  45 BlastDB::~BlastDB(){
  46         remove(queryFileName.c_str());                          //      let's clean stuff up and remove the temp files
  47         remove(dbFileName.c_str());                                     //      let's clean stuff up and remove the temp files
  48         remove(blastFileName.c_str());                          //      let's clean stuff up and remove the temp files
  49 }
  50 /**************************************************************************************************/
  51 //assumes you have added all the template sequences using the addSequence function and run generateDB.
  52 vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
  53         try{
  54                 vector<int> topMatches;
  55
  56                 ofstream queryFile;
  57                 openOutputFile((queryFileName+seq->getName()), queryFile);
  58                 queryFile << '>' << seq->getName() << endl;
  59                 queryFile << seq->getUnaligned() << endl;
  60                 queryFile.close();
  61
  62
  63                 //      the goal here is to quickly survey the database to find the closest match.  To do this we are using the default
  64                 //      wordsize used in megablast.  I'm sure we're sacrificing accuracy for speed, but anyother way would take way too
  65                 //      long.  With this setting, it seems comparable in speed to the suffix tree approach.
  66
  67                 string blastCommand = path + "blast/bin/blastall -p blastn -d " + dbFileName + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);;
  68                 blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName());
  69                 system(blastCommand.c_str());
  70
  71                 ifstream m8FileHandle;
  72                 openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
  73
  74                 string dummy;
  75                 int templateAccession;
  76                 gobble(m8FileHandle);
  77
  78                 while(!m8FileHandle.eof()){
  79                         m8FileHandle >> dummy >> templateAccession >> searchScore;
  80
  81                         //get rest of junk in line
  82                         while (!m8FileHandle.eof())     {       char c = m8FileHandle.get(); if (c == 10 || c == 13){   break;  }       }
  83
  84                         gobble(m8FileHandle);
  85                         topMatches.push_back(templateAccession);
  86                 }
  87                 m8FileHandle.close();
  88                 remove((queryFileName+seq->getName()).c_str());
  89                 remove((blastFileName+seq->getName()).c_str());
  90
  91                 return topMatches;
  92         }
  93         catch(exception& e) {
  94                 m->errorOut(e, "BlastDB", "findClosestSequences");
  95                 exit(1);
  96         }
  97
  98 }
  99 /**************************************************************************************************/
 100 //assumes you have added all the template sequences using the addSequence function and run generateDB.
 101 vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
 102         try{
 103                 vector<int> topMatches;
 104
 105                 ofstream queryFile;
 106                 openOutputFile((queryFileName+seq->getName()), queryFile);
 107                 queryFile << '>' << seq->getName() << endl;
 108                 queryFile << seq->getUnaligned() << endl;
 109                 queryFile.close();
 110
 111                 //      the goal here is to quickly survey the database to find the closest match.  To do this we are using the default
 112                 //      wordsize used in megablast.  I'm sure we're sacrificing accuracy for speed, but anyother way would take way too
 113                 //      long.  With this setting, it seems comparable in speed to the suffix tree approach.
 114
 115                 string blastCommand = path + "blast/bin/megablast -e 1e-10 -d " + dbFileName + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn
 116                 blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName());
 117                 system(blastCommand.c_str());
 118
 119                 ifstream m8FileHandle;
 120                 openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
 121
 122                 string dummy;
 123                 int templateAccession;
 124                 gobble(m8FileHandle);
 125
 126                 while(!m8FileHandle.eof()){
 127                         m8FileHandle >> dummy >> templateAccession >> searchScore;
 128
 129                         //get rest of junk in line
 130                         while (!m8FileHandle.eof())     {       char c = m8FileHandle.get(); if (c == 10 || c == 13){   break;  }       }
 131
 132                         gobble(m8FileHandle);
 133                         topMatches.push_back(templateAccession);
 134 //cout << templateAccession << endl;
 135                 }
 136                 m8FileHandle.close();
 137                 remove((queryFileName+seq->getName()).c_str());
 138                 remove((blastFileName+seq->getName()).c_str());
 139 //cout << "\n\n" ;
 140                 return topMatches;
 141         }
 142         catch(exception& e) {
 143                 m->errorOut(e, "BlastDB", "findClosest");
 144                 exit(1);
 145         }
 146 }
 147 /**************************************************************************************************/
 148 void BlastDB::addSequence(Sequence seq) {
 149         try {
 150
 151                 ofstream unalignedFastaFile;
 152                 openOutputFileAppend(dbFileName, unalignedFastaFile);
 153
 154                 //      generating a fasta file with unaligned template
 155                 unalignedFastaFile << '>' << count << endl;                                     //      sequences, which will be input to formatdb
 156                 unalignedFastaFile << seq.getUnaligned() << endl;
 157                 unalignedFastaFile.close();
 158
 159                 count++;
 160         }
 161         catch(exception& e) {
 162                 m->errorOut(e, "BlastDB", "addSequence");
 163                 exit(1);
 164         }
 165 }
 166 /**************************************************************************************************/
 167 void BlastDB::generateDB() {
 168         try {
 169
 170                 //m->mothurOut("Generating the temporary BLAST database...\t"); cout.flush();
 171
 172                 path = globaldata->argv;
 173                 path = path.substr(0, (path.find_last_of('m')));
 174
 175                 string formatdbCommand = path + "blast/bin/formatdb -p F -o T -i " + dbFileName;        //      format the database, -o option gives us the ability
 176                 system(formatdbCommand.c_str());                                                                //      to get the right sequence names, i think. -p F
 177                                                                                                                                         //      option tells formatdb that seqs are DNA, not prot
 178                 //m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine(); cout.flush();
 179         }
 180         catch(exception& e) {
 181                 m->errorOut(e, "BlastDB", "generateDB");
 182                 exit(1);
 183         }
 184 }
 185 /**************************************************************************************************/
 186
 187 /**************************************************************************************************/
 188