5 * Created by westcott on 11/4/09.
\r
6 * Copyright 2009 Schloss Lab. All rights reserved.
\r
10 #include "alignmentdb.h"
\r
11 #include "kmerdb.hpp"
\r
12 #include "suffixdb.hpp"
\r
13 #include "blastdb.hpp"
\r
16 /**************************************************************************************************/
\r
17 AlignmentDB::AlignmentDB(string fastaFileName, string s, int kmerSize, float gapOpen, float gapExtend, float match, float misMatch){ // This assumes that the template database is in fasta format, may
\r
18 try { // need to alter this in the future?
\r
19 m = MothurOut::getInstance();
\r
22 bool needToGenerate = true;
\r
24 m->mothurOutEndLine();
\r
25 m->mothurOut("Reading in the " + fastaFileName + " template sequences...\t"); cout.flush();
\r
28 int pid, processors;
\r
29 vector<long> positions;
\r
33 MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
\r
34 MPI_Comm_size(MPI_COMM_WORLD, &processors);
\r
37 char inFileName[1024];
\r
38 strcpy(inFileName, fastaFileName.c_str());
\r
40 MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
\r
43 positions = setFilePosFasta(fastaFileName, numSeqs); //fills MPIPos, returns numSeqs
\r
45 //send file positions to all processes
\r
46 for(int i = 1; i < processors; i++) {
\r
47 MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
\r
48 MPI_Send(&positions[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
\r
51 MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
\r
52 positions.resize(numSeqs+1);
\r
53 MPI_Recv(&positions[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
\r
57 for(int i=0;i<numSeqs;i++){
\r
59 if (m->control_pressed) { templateSequences.clear(); break; }
\r
61 //read next sequence
\r
62 int length = positions[i+1] - positions[i];
\r
63 char* buf4 = new char[length];
\r
65 MPI_File_read_at(inMPI, positions[i], buf4, length, MPI_CHAR, &status);
\r
67 string tempBuf = buf4;
\r
68 if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
\r
71 istringstream iss (tempBuf,istringstream::in);
\r
73 Sequence temp(iss);
\r
74 if (temp.getName() != "") {
\r
75 templateSequences.push_back(temp);
\r
77 if (temp.getUnaligned().length() > longest) { longest = temp.getUnaligned().length()+1; }
\r
81 MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
\r
83 MPI_File_close(&inMPI);
\r
87 openInputFile(fastaFileName, fastaFile);
\r
89 while (!fastaFile.eof()) {
\r
90 Sequence temp(fastaFile); gobble(fastaFile);
\r
92 if (m->control_pressed) { templateSequences.clear(); break; }
\r
94 if (temp.getName() != "") {
\r
95 templateSequences.push_back(temp);
\r
97 if (temp.getUnaligned().length() > longest) { longest = temp.getUnaligned().length()+1; }
\r
104 numSeqs = templateSequences.size();
\r
105 //all of this is elsewhere already!
\r
107 m->mothurOut("DONE.");
\r
108 m->mothurOutEndLine(); cout.flush();
\r
110 //in case you delete the seqs and then ask for them
\r
111 emptySequence = Sequence();
\r
112 emptySequence.setName("no_match");
\r
113 emptySequence.setUnaligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX");
\r
114 emptySequence.setAligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX");
\r
118 if(method == "kmer") {
\r
119 search = new KmerDB(fastaFileName, kmerSize);
\r
123 kmerDBName = fastaFileName.substr(0,fastaFileName.find_last_of(".")+1) + char('0'+ kmerSize) + "mer";
\r
124 ifstream kmerFileTest(kmerDBName.c_str());
\r
126 if(kmerFileTest){ needToGenerate = false; }
\r
129 else if(method == "suffix") { search = new SuffixDB(numSeqs); }
\r
130 else if(method == "blast") { search = new BlastDB(gapOpen, gapExtend, match, misMatch); }
\r
132 m->mothurOut(method + " is not a valid search option. I will run the command using kmer, ksize=8.");
\r
133 m->mothurOutEndLine();
\r
134 search = new KmerDB(fastaFileName, 8);
\r
137 if (!(m->control_pressed)) {
\r
138 if (needToGenerate) {
\r
139 //add sequences to search
\r
140 for (int i = 0; i < templateSequences.size(); i++) {
\r
141 search->addSequence(templateSequences[i]);
\r
143 if (m->control_pressed) { templateSequences.clear(); break; }
\r
146 if (m->control_pressed) { templateSequences.clear(); }
\r
148 search->generateDB();
\r
150 }else if ((method == "kmer") && (!needToGenerate)) {
\r
151 ifstream kmerFileTest(kmerDBName.c_str());
\r
152 search->readKmerDB(kmerFileTest);
\r
155 search->setNumSeqs(numSeqs);
\r
158 catch(exception& e) {
\r
159 m->errorOut(e, "AlignmentDB", "AlignmentDB");
\r
163 /**************************************************************************************************/
\r
164 AlignmentDB::AlignmentDB(string s){
\r
166 m = MothurOut::getInstance();
\r
169 if(method == "suffix") { search = new SuffixDB(); }
\r
170 else if(method == "blast") { search = new BlastDB(); }
\r
171 else { search = new KmerDB(); }
\r
174 //in case you delete the seqs and then ask for them
\r
175 emptySequence = Sequence();
\r
176 emptySequence.setName("no_match");
\r
177 emptySequence.setUnaligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX");
\r
178 emptySequence.setAligned("XXXXXXXXXXXXXXXXXXXXXXXXXXXXX");
\r
181 catch(exception& e) {
\r
182 m->errorOut(e, "AlignmentDB", "AlignmentDB");
\r
186 /**************************************************************************************************/
\r
187 AlignmentDB::~AlignmentDB() { delete search; }
\r
188 /**************************************************************************************************/
\r
189 Sequence AlignmentDB::findClosestSequence(Sequence* seq) {
\r
192 vector<int> spot = search->findClosestSequences(seq, 1);
\r
194 if (spot.size() != 0) { return templateSequences[spot[0]]; }
\r
195 else { return emptySequence; }
\r
198 catch(exception& e) {
\r
199 m->errorOut(e, "AlignmentDB", "findClosestSequence");
\r
203 /**************************************************************************************************/
\r