it = parameters.find("fasta");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["fasta"] = inputDir + it->second; }
}
it = parameters.find("hard");
//user has given a template file
if(it != parameters.end()){
- path = hasPath(it->second);
+ path = m->hasPath(it->second);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["hard"] = inputDir + it->second; }
}
fasta = validParameter.validFile(parameters, "fasta", false);
if (fasta == "not found") { m->mothurOut("fasta is a required parameter for the filter.seqs command."); m->mothurOutEndLine(); abort = true; }
else {
- splitAtDash(fasta, fastafileNames);
+ m->splitAtDash(fasta, fastafileNames);
//go through files and make sure they are good, if not, then disregard them
for (int i = 0; i < fastafileNames.size(); i++) {
if (inputDir != "") {
- string path = hasPath(fastafileNames[i]);
+ string path = m->hasPath(fastafileNames[i]);
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { fastafileNames[i] = inputDir + fastafileNames[i]; }
}
ifstream in;
- int ableToOpen = openInputFile(fastafileNames[i], in, "noerror");
+ int ableToOpen = m->openInputFile(fastafileNames[i], in, "noerror");
//if you can't open it, try default location
if (ableToOpen == 1) {
if (m->getDefaultPath() != "") { //default path is set
- string tryPath = m->getDefaultPath() + getSimpleName(fastafileNames[i]);
+ string tryPath = m->getDefaultPath() + m->getSimpleName(fastafileNames[i]);
m->mothurOut("Unable to open " + fastafileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
- ableToOpen = openInputFile(tryPath, in, "noerror");
+ ableToOpen = m->openInputFile(tryPath, in, "noerror");
fastafileNames[i] = tryPath;
}
}
fastafileNames.erase(fastafileNames.begin()+i);
i--;
}else{
- string simpleName = getSimpleName(fastafileNames[i]);
+ string simpleName = m->getSimpleName(fastafileNames[i]);
filterFileName += simpleName.substr(0, simpleName.find_first_of('.'));
}
in.close();
//if the user changes the output directory command factory will send this info to us in the output parameter
outputDir = validParameter.validFile(parameters, "outputdir", false); if (outputDir == "not found"){
outputDir = "";
- outputDir += hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
+ outputDir += m->hasPath(fastafileNames[0]); //if user entered a file with a path then preserve it
}
}
//check for optional parameter and set defaults
try {
m->mothurOut("The filter.seqs command reads a file containing sequences and creates a .filter and .filter.fasta file.\n");
- m->mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard and vertical. \n");
+ m->mothurOut("The filter.seqs command parameters are fasta, trump, soft, hard, processors and vertical. \n");
m->mothurOut("The fasta parameter is required. You may enter several fasta files to build the filter from and filter, by separating their names with -'s.\n");
m->mothurOut("For example: fasta=abrecovery.fasta-amazon.fasta \n");
m->mothurOut("The trump parameter .... The default is ...\n");
m->mothurOut("The soft parameter .... The default is ....\n");
m->mothurOut("The hard parameter allows you to enter a file containing the filter you want to use.\n");
m->mothurOut("The vertical parameter removes columns where all sequences contain a gap character. The default is T.\n");
+ m->mothurOut("The processors parameter allows you to specify the number of processors to use. The default is 1.\n");
m->mothurOut("The filter.seqs command should be in the following format: \n");
m->mothurOut("filter.seqs(fasta=yourFastaFile, trump=yourTrump) \n");
m->mothurOut("Example filter.seqs(fasta=abrecovery.fasta, trump=.).\n");
if (abort == true) { return 0; }
ifstream inFASTA;
- openInputFile(fastafileNames[0], inFASTA);
+ m->openInputFile(fastafileNames[0], inFASTA);
Sequence testSeq(inFASTA);
alignmentLength = testSeq.getAlignLength();
ofstream outFilter;
- string filterFile = outputDir + filterFileName + ".filter";
- openOutputFile(filterFile, outFilter);
+ //prevent giantic file name
+ string filterFile;
+ if (fastafileNames.size() > 3) { filterFile = outputDir + "merge.filter"; }
+ else { filterFile = outputDir + filterFileName + ".filter"; }
+
+ m->openOutputFile(filterFile, outFilter);
outFilter << filter << endl;
outFilter.close();
outputNames.push_back(filterFile);
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
- string filteredFasta = outputDir + getRootName(getSimpleName(fastafileNames[s])) + "filter.fasta";
+ string filteredFasta = outputDir + m->getRootName(m->getSimpleName(fastafileNames[s])) + "filter.fasta";
#ifdef USE_MPI
int pid, start, end, numSeqsPerProcessor, num;
int tag = 2001;
if (pid == 0) { //you are the root process
- MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
+ MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
numSeqs += num;
//send file positions to all processes
//wait on chidren
for(int i = 1; i < processors; i++) {
- char buf[4];
- MPI_Recv(buf, 4, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
+ char buf[5];
+ MPI_Recv(buf, 5, MPI_CHAR, i, tag, MPI_COMM_WORLD, &status);
}
}else { //you are a child process
if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPI); return 0; }
- char buf[4];
+ char buf[5];
strcpy(buf, "done");
//tell parent you are done.
- MPI_Send(buf, 4, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
+ MPI_Send(buf, 5, MPI_CHAR, 0, tag, MPI_COMM_WORLD);
}
MPI_File_close(&outMPI);
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- vector<unsigned long int> positions = divideFile(fastafileNames[s], processors);
+ vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
for (int i = 0; i < (positions.size()-1); i++) {
lines.push_back(new linePair(positions[i], positions[(i+1)]));
//append fasta files
for(int i=1;i<processors;i++){
- appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
+ m->appendFiles((fastafileNames[s] + toString(processIDS[i]) + ".temp"), filteredFasta);
remove((fastafileNames[s] + toString(processIDS[i]) + ".temp").c_str());
}
}
if (m->control_pressed) { return 1; }
#else
- numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
+ int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
if (m->control_pressed) { return 1; }
istringstream iss (tempBuf,istringstream::in);
delete buf4;
- Sequence seq(iss); gobble(iss);
+ Sequence seq(iss); m->gobble(iss);
if (seq.getName() != "") {
string align = seq.getAligned();
int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* filePos) {
try {
ofstream out;
- openOutputFile(outputFilename, out);
+ m->openOutputFile(outputFilename, out);
ifstream in;
- openInputFile(inputFilename, in);
+ m->openInputFile(inputFilename, in);
in.seekg(filePos->start);
if (m->control_pressed) { in.close(); out.close(); return 0; }
- Sequence seq(in); gobble(in);
+ Sequence seq(in); m->gobble(in);
if (seq.getName() != "") {
string align = seq.getAligned();
string filterSeq = "";
count++;
}
- unsigned long int pos = in.tellg();
- if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ unsigned long int pos = in.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (in.eof()) { break; }
+ #endif
//report progress
if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
//pass numSeqs to parent
ofstream out;
string tempFile = filename + toString(getpid()) + ".num.temp";
- openOutputFile(tempFile, out);
+ m->openOutputFile(tempFile, out);
out << num << endl;
out.close();
for (int i = 0; i < processIDS.size(); i++) {
ifstream in;
string tempFile = filename + toString(processIDS[i]) + ".num.temp";
- openInputFile(tempFile, in);
+ m->openInputFile(tempFile, in);
if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
in.close(); remove(tempFile.c_str());
}
F.setLength(alignmentLength);
- if(trump != '*' || isTrue(vertical) || soft != 0){
+ if(trump != '*' || m->isTrue(vertical) || soft != 0){
F.initialize();
}
else { F.setFilter(string(alignmentLength, '1')); }
numSeqs = 0;
- if(trump != '*' || isTrue(vertical) || soft != 0){
+ if(trump != '*' || m->isTrue(vertical) || soft != 0){
for (int s = 0; s < fastafileNames.size(); s++) {
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
if (m->control_pressed) { MPI_File_close(&inMPI); return 0; }
if (pid == 0) { //you are the root process
- MPIPos = setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
+ MPIPos = m->setFilePosFasta(fastafileNames[s], num); //fills MPIPos, returns numSeqs
numSeqs += num;
//send file positions to all processes
MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
#else
- vector<unsigned long int> positions = divideFile(fastafileNames[s], processors);
+ vector<unsigned long int> positions = m->divideFile(fastafileNames[s], processors);
for (int i = 0; i < (positions.size()-1); i++) {
lines.push_back(new linePair(positions[i], positions[(i+1)]));
if (m->control_pressed) { return filterString; }
#else
- numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
+ int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
if (m->control_pressed) { return filterString; }
#endif
MPI_Comm_rank(MPI_COMM_WORLD, &pid);
- if(trump != '*' || isTrue(vertical) || soft != 0){
+ if(trump != '*' || m->isTrue(vertical) || soft != 0){
if (pid == 0) { //only one process should output the filter
if (pid == 0) { //only one process should output the filter
#endif
+
F.setNumSeqs(numSeqs);
- if(isTrue(vertical) == 1) { F.doVertical(); }
+ if(m->isTrue(vertical) == 1) { F.doVertical(); }
if(soft != 0) { F.doSoft(); }
filterString = F.getFilter();
try {
ifstream in;
- openInputFile(filename, in);
+ m->openInputFile(filename, in);
in.seekg(filePos->start);
if (m->control_pressed) { in.close(); return 1; }
- Sequence seq(in); gobble(in);
+ Sequence seq(in); m->gobble(in);
if (seq.getName() != "") {
if (seq.getAligned().length() != alignmentLength) { m->mothurOut("Sequences are not all the same length, please correct."); m->mothurOutEndLine(); m->control_pressed = true; }
if(trump != '*') { F.doTrump(seq); }
- if(isTrue(vertical) || soft != 0) { F.getFreqs(seq); }
+ if(m->isTrue(vertical) || soft != 0) { F.getFreqs(seq); }
cout.flush();
count++;
}
- unsigned long int pos = in.tellg();
- if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ unsigned long int pos = in.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (in.eof()) { break; }
+ #endif
//report progress
if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
if (seq.getAligned().length() != alignmentLength) { cout << "Alignment length is " << alignmentLength << " and sequence " << seq.getName() << " has length " << seq.getAligned().length() << ", please correct." << endl; exit(1); }
if(trump != '*'){ F.doTrump(seq); }
- if(isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
+ if(m->isTrue(vertical) || soft != 0){ F.getFreqs(seq); }
cout.flush();
//report progress
int FilterSeqsCommand::createProcessesCreateFilter(Filters& F, string filename) {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- int process = 0;
+ int process = 1;
int num = 0;
processIDS.clear();
processIDS.push_back(pid); //create map from line number to pid so you can append files in correct order later
process++;
}else if (pid == 0){
+ //reset child's filter counts to 0;
+ F.a.clear(); F.a.resize(alignmentLength, 0);
+ F.t.clear(); F.t.resize(alignmentLength, 0);
+ F.g.clear(); F.g.resize(alignmentLength, 0);
+ F.c.clear(); F.c.resize(alignmentLength, 0);
+ F.gap.clear(); F.gap.resize(alignmentLength, 0);
+
num = driverCreateFilter(F, filename, lines[process]);
//write out filter counts to file
filename += toString(getpid()) + "filterValues.temp";
ofstream out;
- openOutputFile(filename, out);
+ m->openOutputFile(filename, out);
out << num << endl;
out << F.getFilter() << endl;
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
+ //parent do your part
+ num = driverCreateFilter(F, filename, lines[0]);
+
//force parent to wait until all the processes are done
- for (int i=0;i<processors;i++) {
+ for (int i=0;i<(processors-1);i++) {
int temp = processIDS[i];
wait(&temp);
}
- //parent reads in and combine Filter info
+ //parent reads in and combines Filter info
for (int i = 0; i < processIDS.size(); i++) {
string tempFilename = filename + toString(processIDS[i]) + "filterValues.temp";
ifstream in;
- openInputFile(tempFilename, in);
+ m->openInputFile(tempFilename, in);
int temp, tempNum;
string tempFilterString;
- in >> tempNum; gobble(in); num += tempNum;
+ in >> tempNum; m->gobble(in); num += tempNum;
in >> tempFilterString;
F.mergeFilter(tempFilterString);
- for (int k = 0; k < alignmentLength; k++) { in >> temp; F.a[k] += temp; } gobble(in);
- for (int k = 0; k < alignmentLength; k++) { in >> temp; F.t[k] += temp; } gobble(in);
- for (int k = 0; k < alignmentLength; k++) { in >> temp; F.g[k] += temp; } gobble(in);
- for (int k = 0; k < alignmentLength; k++) { in >> temp; F.c[k] += temp; } gobble(in);
- for (int k = 0; k < alignmentLength; k++) { in >> temp; F.gap[k] += temp; } gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.a[k] += temp; } m->gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.t[k] += temp; } m->gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.g[k] += temp; } m->gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.c[k] += temp; } m->gobble(in);
+ for (int k = 0; k < alignmentLength; k++) { in >> temp; F.gap[k] += temp; } m->gobble(in);
in.close();
remove(tempFilename.c_str());
projects / mothur.git / blobdiff