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
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);
+ 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;
MPI_Status status;
MPI_Comm_size(MPI_COMM_WORLD, &processors); //set processors to the number of mpi processes running
MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
- cout << pid << "is in create filter " << endl;
+
MPI_File outMPI;
MPI_File tempMPI;
MPI_File inMPI;
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 = 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 defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
- ifstream inFASTA;
- int numFastaSeqs;
- openInputFile(fastafileNames[s], inFASTA);
- getNumSeqs(inFASTA, numFastaSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numFastaSeqs));
-
+ int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
-
- driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
}else{
- setLines(fastafileNames[s]);
- createProcessesRunFilter(filter, fastafileNames[s]);
+ int numFastaSeqs = createProcessesRunFilter(filter, fastafileNames[s]);
+ numSeqs += numFastaSeqs;
rename((fastafileNames[s] + toString(processIDS[0]) + ".temp").c_str(), filteredFasta.c_str());
//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
- ifstream inFASTA;
- int numFastaSeqs;
- openInputFile(fastafileNames[s], inFASTA);
- getNumSeqs(inFASTA, numFastaSeqs);
- inFASTA.close();
-
- lines.push_back(new linePair(0, numFastaSeqs));
-
+ int numFastaSeqs = driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
-
- driverRunFilter(filter, filteredFasta, fastafileNames[s], lines[0]);
if (m->control_pressed) { return 1; }
#endif
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();
}
#endif
/**************************************************************************************/
-int FilterSeqsCommand::driverRunFilter(string F, string outputFilename, string inputFilename, linePair* line) {
+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(line->start);
-
- for(int i=0;i<line->num;i++){
+ in.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ while (!done) {
if (m->control_pressed) { in.close(); out.close(); return 0; }
- Sequence seq(in);
+ Sequence seq(in); m->gobble(in);
if (seq.getName() != "") {
string align = seq.getAligned();
string filterSeq = "";
}
out << '>' << seq.getName() << endl << filterSeq << endl;
- }
- gobble(in);
+ count++;
+ }
+
+ unsigned long int pos = in.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+
+ //report progress
+ if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
}
+ //report progress
+ if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
+
+
out.close();
in.close();
- return 0;
+ return count;
}
catch(exception& e) {
m->errorOut(e, "FilterSeqsCommand", "driverRunFilter");
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
- int exitCommand = 1;
+ int num = 0;
processIDS.clear();
//loop through and create all the processes you want
process++;
}else if (pid == 0){
string filteredFasta = filename + toString(getpid()) + ".temp";
- driverRunFilter(F, filteredFasta, filename, lines[process]);
+ num = driverRunFilter(F, filteredFasta, filename, lines[process]);
+
+ //pass numSeqs to parent
+ ofstream out;
+ string tempFile = filename + toString(getpid()) + ".num.temp";
+ m->openOutputFile(tempFile, out);
+ out << num << endl;
+ out.close();
+
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
for (int i=0;i<processors;i++) {
int temp = processIDS[i];
wait(&temp);
+ }
+
+ for (int i = 0; i < processIDS.size(); i++) {
+ ifstream in;
+ string tempFile = filename + toString(processIDS[i]) + ".num.temp";
+ m->openInputFile(tempFile, in);
+ if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
+ in.close(); remove(tempFile.c_str());
}
+
- return exitCommand;
+ return num;
#endif
}
catch(exception& e) {
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 = 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 defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
if(processors == 1){
- ifstream inFASTA;
- int numFastaSeqs;
- openInputFile(fastafileNames[s], inFASTA);
- getNumSeqs(inFASTA, numFastaSeqs);
- inFASTA.close();
-
+ int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
-
- lines.push_back(new linePair(0, numFastaSeqs));
-
- driverCreateFilter(F, fastafileNames[s], lines[0]);
}else{
- setLines(fastafileNames[s]);
- createProcessesCreateFilter(F, fastafileNames[s]);
+ int numFastaSeqs = createProcessesCreateFilter(F, fastafileNames[s]);
+ numSeqs += numFastaSeqs;
}
if (m->control_pressed) { return filterString; }
#else
- ifstream inFASTA;
- int numFastaSeqs;
- openInputFile(fastafileNames[s], inFASTA);
- getNumSeqs(inFASTA, numFastaSeqs);
- inFASTA.close();
-
+ int numFastaSeqs = driverCreateFilter(F, fastafileNames[s], lines[0]);
numSeqs += numFastaSeqs;
-
- lines.push_back(new linePair(0, numFastaSeqs));
-
- driverCreateFilter(F, fastafileNames[s], lines[0]);
if (m->control_pressed) { return filterString; }
#endif
#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();
#ifdef USE_MPI
}
}
/**************************************************************************************/
-int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* line) {
+int FilterSeqsCommand::driverCreateFilter(Filters& F, string filename, linePair* filePos) {
try {
ifstream in;
- openInputFile(filename, in);
+ m->openInputFile(filename, in);
- in.seekg(line->start);
-
- for(int i=0;i<line->num;i++){
+ in.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ while (!done) {
if (m->control_pressed) { in.close(); return 1; }
- Sequence seq(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(trump != '*') { F.doTrump(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; }
+
//report progress
- if((i+1) % 100 == 0){ m->mothurOut(toString(i+1)); m->mothurOutEndLine(); }
+ if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
}
-
//report progress
- if((line->num) % 100 != 0){ m->mothurOut(toString(line->num)); m->mothurOutEndLine(); }
-
+ if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
in.close();
- return 0;
+ return count;
}
catch(exception& e) {
m->errorOut(e, "FilterSeqsCommand", "driverCreateFilter");
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
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
- int exitCommand = 1;
+ int num = 0;
processIDS.clear();
//loop through and create all the processes you want
while (process != processors) {
- int pid = vfork();
+ int pid = fork();
if (pid > 0) {
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){
- driverCreateFilter(F, filename, lines[process]);
+ //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;
+ m->openOutputFile(filename, out);
+
+ out << num << endl;
+ out << F.getFilter() << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.a[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.t[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.g[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.c[k] << '\t'; } out << endl;
+ for (int k = 0; k < alignmentLength; k++) { out << F.gap[k] << '\t'; } out << endl;
+
+ //cout << F.getFilter() << endl;
+ out.close();
+
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
wait(&temp);
}
- return exitCommand;
-#endif
- }
- catch(exception& e) {
- m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
- exit(1);
- }
-}
-/**************************************************************************************************/
-
-int FilterSeqsCommand::setLines(string filename) {
- try {
-
- vector<unsigned long int> positions;
- bufferSizes.clear();
-
- ifstream inFASTA;
- openInputFile(filename, inFASTA);
+ //parent reads in and combine Filter info
+ for (int i = 0; i < processIDS.size(); i++) {
+ string tempFilename = filename + toString(processIDS[i]) + "filterValues.temp";
+ ifstream in;
+ m->openInputFile(tempFilename, in);
- string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
+ int temp, tempNum;
+ string tempFilterString;
- if (input.length() != 0) {
- if(input[0] == '>'){ unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
- }
- }
- inFASTA.close();
-
- int numFastaSeqs = positions.size();
-
- FILE * pFile;
- unsigned long int size;
-
- //get num bytes in file
- pFile = fopen (filename.c_str(),"rb");
- if (pFile==NULL) perror ("Error opening file");
- else{
- fseek (pFile, 0, SEEK_END);
- size=ftell (pFile);
- fclose (pFile);
- }
-
- numSeqs += numFastaSeqs;
-
- int numSeqsPerProcessor = numFastaSeqs / processors;
-
- for (int i = 0; i < processors; i++) {
+ in >> tempNum; m->gobble(in); num += tempNum;
- unsigned long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
- bufferSizes.push_back(size - startPos);
- }else{
- unsigned long int myEnd = positions[ (i+1) * numSeqsPerProcessor ];
- bufferSizes.push_back(myEnd-startPos);
- }
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
+ in >> tempFilterString;
+ F.mergeFilter(tempFilterString);
+
+ 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());
}
- return 0;
+ return num;
+#endif
}
catch(exception& e) {
- m->errorOut(e, "FilterSeqsCommand", "setLines");
+ m->errorOut(e, "FilterSeqsCommand", "createProcessesCreateFilter");
exit(1);
}
}
projects / mothur.git / blobdiff