5 * Created by Pat Schloss on 12/22/08.
6 * Copyright 2008 Patrick D. Schloss. All rights reserved.
11 #include "database.hpp"
12 #include "sequence.hpp"
13 #include "blastdb.hpp"
15 /**************************************************************************************************/
17 BlastDB::BlastDB(float gO, float gE, float m, float mM) : Database(),
18 gapOpen(gO), gapExtend(gE), match(m), misMatch(mM) {
20 globaldata = GlobalData::getInstance();
23 int randNumber = rand();
24 dbFileName = toString(randNumber) + ".template.unaligned.fasta";
25 queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
26 blastFileName = toString(randNumber) + ".blast";
29 /**************************************************************************************************/
31 BlastDB::BlastDB() : Database() {
33 globaldata = GlobalData::getInstance();
36 int randNumber = rand();
37 dbFileName = toString(randNumber) + ".template.unaligned.fasta";
38 queryFileName = toString(randNumber) + ".candidate.unaligned.fasta";
39 blastFileName = toString(randNumber) + ".blast";
42 m->errorOut(e, "BlastDB", "BlastDB");
47 /**************************************************************************************************/
51 remove(queryFileName.c_str()); // let's clean stuff up and remove the temp files
52 remove(dbFileName.c_str()); // let's clean stuff up and remove the temp files
53 remove((dbFileName+".nsq").c_str()); // let's clean stuff up and remove the temp files
54 remove((dbFileName+".nsi").c_str()); // let's clean stuff up and remove the temp files
55 remove((dbFileName+".nsd").c_str()); // let's clean stuff up and remove the temp files
56 remove((dbFileName+".nin").c_str()); // let's clean stuff up and remove the temp files
57 remove((dbFileName+".nhr").c_str()); // let's clean stuff up and remove the temp files
58 remove(blastFileName.c_str()); // let's clean stuff up and remove the temp files
61 m->errorOut(e, "BlastDB", "~BlastDB");
65 /**************************************************************************************************/
66 //assumes you have added all the template sequences using the addSequence function and run generateDB.
67 vector<int> BlastDB::findClosestSequences(Sequence* seq, int n) {
69 vector<int> topMatches;
72 m->openOutputFile((queryFileName+seq->getName()), queryFile);
73 queryFile << '>' << seq->getName() << endl;
74 queryFile << seq->getUnaligned() << endl;
78 // the goal here is to quickly survey the database to find the closest match. To do this we are using the default
79 // wordsize used in megablast. I'm sure we're sacrificing accuracy for speed, but anyother way would take way too
80 // long. With this setting, it seems comparable in speed to the suffix tree approach.
82 string blastCommand = path + "blast/bin/blastall -p blastn -d " + dbFileName + " -m 8 -W 28 -v " + toString(n) + " -b " + toString(n);;
83 blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName());
84 system(blastCommand.c_str());
86 ifstream m8FileHandle;
87 m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
90 int templateAccession;
91 m->gobble(m8FileHandle);
93 while(!m8FileHandle.eof()){
94 m8FileHandle >> dummy >> templateAccession >> searchScore;
96 //get rest of junk in line
97 while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; } }
99 m->gobble(m8FileHandle);
100 topMatches.push_back(templateAccession);
102 m8FileHandle.close();
103 remove((queryFileName+seq->getName()).c_str());
104 remove((blastFileName+seq->getName()).c_str());
108 catch(exception& e) {
109 m->errorOut(e, "BlastDB", "findClosestSequences");
114 /**************************************************************************************************/
115 //assumes you have added all the template sequences using the addSequence function and run generateDB.
116 vector<int> BlastDB::findClosestMegaBlast(Sequence* seq, int n) {
118 vector<int> topMatches;
121 m->openOutputFile((queryFileName+seq->getName()), queryFile);
122 queryFile << '>' << seq->getName() << endl;
123 queryFile << seq->getUnaligned() << endl;
126 // the goal here is to quickly survey the database to find the closest match. To do this we are using the default
127 // wordsize used in megablast. I'm sure we're sacrificing accuracy for speed, but anyother way would take way too
128 // long. With this setting, it seems comparable in speed to the suffix tree approach.
130 string blastCommand = path + "blast/bin/megablast -e 1e-10 -d " + dbFileName + " -m 8 -b " + toString(n) + " -v " + toString(n); //-W 28 -p blastn
131 blastCommand += (" -i " + (queryFileName+seq->getName()) + " -o " + blastFileName+seq->getName());
132 system(blastCommand.c_str());
134 ifstream m8FileHandle;
135 m->openInputFile(blastFileName+seq->getName(), m8FileHandle, "no error");
138 int templateAccession;
139 m->gobble(m8FileHandle);
141 while(!m8FileHandle.eof()){
142 m8FileHandle >> dummy >> templateAccession >> searchScore;
144 //get rest of junk in line
145 while (!m8FileHandle.eof()) { char c = m8FileHandle.get(); if (c == 10 || c == 13){ break; } }
147 m->gobble(m8FileHandle);
148 topMatches.push_back(templateAccession);
149 //cout << templateAccession << endl;
151 m8FileHandle.close();
152 remove((queryFileName+seq->getName()).c_str());
153 remove((blastFileName+seq->getName()).c_str());
157 catch(exception& e) {
158 m->errorOut(e, "BlastDB", "findClosest");
162 /**************************************************************************************************/
163 void BlastDB::addSequence(Sequence seq) {
166 ofstream unalignedFastaFile;
167 m->openOutputFileAppend(dbFileName, unalignedFastaFile);
169 // generating a fasta file with unaligned template
170 unalignedFastaFile << '>' << count << endl; // sequences, which will be input to formatdb
171 unalignedFastaFile << seq.getUnaligned() << endl;
172 unalignedFastaFile.close();
176 catch(exception& e) {
177 m->errorOut(e, "BlastDB", "addSequence");
181 /**************************************************************************************************/
182 void BlastDB::generateDB() {
185 //m->mothurOut("Generating the temporary BLAST database...\t"); cout.flush();
187 path = globaldata->argv;
188 path = path.substr(0, (path.find_last_of('m')));
190 string formatdbCommand = path + "blast/bin/formatdb -p F -o T -i " + dbFileName; // format the database, -o option gives us the ability
191 system(formatdbCommand.c_str()); // to get the right sequence names, i think. -p F
192 // option tells formatdb that seqs are DNA, not prot
193 //m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine(); cout.flush();
195 catch(exception& e) {
196 m->errorOut(e, "BlastDB", "generateDB");
200 /**************************************************************************************************/
202 /**************************************************************************************************/