#include "distancedb.hpp"
/**************************************************************************************************/
-Classify::Classify(string tfile, string tempFile, string method, int kmerSize, float gapOpen, float gapExtend, float match, float misMatch) : taxFile(tfile), templateFile(tempFile) {
+void Classify::generateDatabaseAndNames(string tfile, string tempFile, string method, int kmerSize, float gapOpen, float gapExtend, float match, float misMatch) {
try {
- m = MothurOut::getInstance();
+ taxFile = tfile;
readTaxonomy(taxFile);
+ templateFile = tempFile;
+
int start = time(NULL);
int numSeqs = 0;
m->mothurOut("Generating search database... "); cout.flush();
#ifdef USE_MPI
- int pid;
- vector<long> positions;
+ int pid, processors;
+ vector<unsigned long int> positions;
+ int tag = 2001;
MPI_Status status;
MPI_File inMPI;
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
-
- char inFileName[tempFile.length()];
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
+
+ //char* inFileName = new char[tempFile.length()];
+ //memcpy(inFileName, tempFile.c_str(), tempFile.length());
+
+ char inFileName[1024];
strcpy(inFileName, tempFile.c_str());
-
+
MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
-
+ //delete inFileName;
+
if (pid == 0) { //only one process needs to scan file
- positions = setFilePosFasta(tempFile, numSeqs); //fills MPIPos, returns numSeqs
+ positions = m->setFilePosFasta(tempFile, numSeqs); //fills MPIPos, returns numSeqs
//send file positions to all processes
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&positions[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&numSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&positions[0], (numSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
+ }
}else{
- MPI_Bcast(&numSeqs, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
- positions.resize(numSeqs);
- MPI_Bcast(&positions[0], (numSeqs+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
+ MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ positions.resize(numSeqs+1);
+ MPI_Recv(&positions[0], (numSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
}
//create database
if(method == "kmer") { database = new KmerDB(tempFile, kmerSize); }
else if(method == "suffix") { database = new SuffixDB(numSeqs); }
- else if(method == "blast") { database = new BlastDB(gapOpen, gapExtend, match, misMatch); }
+ else if(method == "blast") { database = new BlastDB(tempFile.substr(0,tempFile.find_last_of(".")+1), gapOpen, gapExtend, match, misMatch); }
else if(method == "distance") { database = new DistanceDB(); }
else {
m->mothurOut(method + " is not a valid search option. I will run the command using kmer, ksize=8."); m->mothurOutEndLine();
for(int i=0;i<numSeqs;i++){
//read next sequence
int length = positions[i+1] - positions[i];
- char buf4[length];
+ char* buf4 = new char[length];
MPI_File_read_at(inMPI, positions[i], buf4, length, MPI_CHAR, &status);
string tempBuf = buf4;
if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
-
+ delete buf4;
istringstream iss (tempBuf,istringstream::in);
Sequence temp(iss);
database->generateDB();
MPI_File_close(&inMPI);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
//need to know number of template seqs for suffixdb
if (method == "suffix") {
ifstream inFASTA;
- openInputFile(tempFile, inFASTA);
- numSeqs = count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+ m->openInputFile(tempFile, inFASTA);
+ m->getNumSeqs(inFASTA, numSeqs);
inFASTA.close();
}
kmerDBName = tempFile.substr(0,tempFile.find_last_of(".")+1) + char('0'+ kmerSize) + "mer";
ifstream kmerFileTest(kmerDBName.c_str());
- if(kmerFileTest){ needToGenerate = false; }
+ if(kmerFileTest){
+ bool GoodFile = m->checkReleaseVersion(kmerFileTest, m->getVersion());
+ if (GoodFile) { needToGenerate = false; }
+ }
}
else if(method == "suffix") { database = new SuffixDB(numSeqs); }
- else if(method == "blast") { database = new BlastDB(gapOpen, gapExtend, match, misMatch); }
+ else if(method == "blast") { database = new BlastDB(tempFile.substr(0,tempFile.find_last_of(".")+1), gapOpen, gapExtend, match, misMatch); }
else if(method == "distance") { database = new DistanceDB(); }
else {
m->mothurOut(method + " is not a valid search option. I will run the command using kmer, ksize=8.");
if (needToGenerate) {
ifstream fastaFile;
- openInputFile(tempFile, fastaFile);
+ m->openInputFile(tempFile, fastaFile);
while (!fastaFile.eof()) {
Sequence temp(fastaFile);
- gobble(fastaFile);
+ m->gobble(fastaFile);
names.push_back(temp.getName());
-
+
database->addSequence(temp);
}
fastaFile.close();
}else if ((method == "kmer") && (!needToGenerate)) {
ifstream kmerFileTest(kmerDBName.c_str());
database->readKmerDB(kmerFileTest);
-
+
ifstream fastaFile;
- openInputFile(tempFile, fastaFile);
+ m->openInputFile(tempFile, fastaFile);
while (!fastaFile.eof()) {
Sequence temp(fastaFile);
- gobble(fastaFile);
-
+ m->gobble(fastaFile);
+
names.push_back(temp.getName());
}
fastaFile.close();
}
-#endif
+#endif
+
database->setNumSeqs(names.size());
+ //sanity check
+ bool okay = phyloTree->ErrorCheck(names);
+
+ if (!okay) { m->control_pressed = true; }
+
m->mothurOut("DONE."); m->mothurOutEndLine();
m->mothurOut("It took " + toString(time(NULL) - start) + " seconds generate search database. "); m->mothurOutEndLine();
}
catch(exception& e) {
- m->errorOut(e, "Classify", "Classify");
+ m->errorOut(e, "Classify", "generateDatabaseAndNames");
exit(1);
}
}
/**************************************************************************************************/
+Classify::Classify() { m = MothurOut::getInstance(); database = NULL; }
+/**************************************************************************************************/
-void Classify::readTaxonomy(string file) {
+int Classify::readTaxonomy(string file) {
try {
phyloTree = new PhyloTree();
m->mothurOut("Reading in the " + file + " taxonomy...\t"); cout.flush();
#ifdef USE_MPI
- int pid, num;
- vector<long> positions;
+ int pid, num, processors;
+ vector<unsigned long int> positions;
+ int tag = 2001;
MPI_Status status;
MPI_File inMPI;
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
+
+ //char* inFileName = new char[file.length()];
+ //memcpy(inFileName, file.c_str(), file.length());
- char inFileName[file.length()];
+ char inFileName[1024];
strcpy(inFileName, file.c_str());
-
+
MPI_File_open(MPI_COMM_WORLD, inFileName, MPI_MODE_RDONLY, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
-
+ //delete inFileName;
+
if (pid == 0) {
- positions = setFilePosEachLine(file, num);
+ positions = m->setFilePosEachLine(file, num);
//send file positions to all processes
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //send numSeqs
- MPI_Bcast(&positions[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //send file pos
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&num, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&positions[0], (num+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
+ }
}else{
- MPI_Bcast(&num, 1, MPI_INT, 0, MPI_COMM_WORLD); //get numSeqs
- positions.resize(num);
- MPI_Bcast(&positions[0], (num+1), MPI_LONG, 0, MPI_COMM_WORLD); //get file positions
+ MPI_Recv(&num, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ positions.resize(num+1);
+ MPI_Recv(&positions[0], (num+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
}
//read file
for(int i=0;i<num;i++){
//read next sequence
int length = positions[i+1] - positions[i];
- char buf4[length];
+ char* buf4 = new char[length];
MPI_File_read_at(inMPI, positions[i], buf4, length, MPI_CHAR, &status);
string tempBuf = buf4;
if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
-
+ delete buf4;
+
istringstream iss (tempBuf,istringstream::in);
iss >> name >> taxInfo;
taxonomy[name] = taxInfo;
}
MPI_File_close(&inMPI);
+ MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
ifstream inTax;
- openInputFile(file, inTax);
+ m->openInputFile(file, inTax);
//read template seqs and save
while (!inTax.eof()) {
phyloTree->addSeqToTree(name, taxInfo);
- gobble(inTax);
+ m->gobble(inTax);
}
inTax.close();
#endif
phyloTree->assignHeirarchyIDs(0);
+ phyloTree->setUp(file);
+
m->mothurOut("DONE.");
m->mothurOutEndLine(); cout.flush();
+
+ return phyloTree->getNumSeqs();
}
catch(exception& e) {
projects / mothur.git / blobdiff