try {
abort = false;
-
+
//allow user to run help
if(option == "help") { help(); abort = true; }
OptionParser parser(option);
map<string, string> parameters = parser.getParameters();
- ValidParameters validParameter;
+ ValidParameters validParameter("align.seqs");
map<string, string>::iterator it;
//check to make sure all parameters are valid for command
//if the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { candidateFileNames[i] = inputDir + candidateFileNames[i]; }
}
-
+
int ableToOpen;
ifstream in;
- ableToOpen = openInputFile(candidateFileNames[i], in);
+
+ ableToOpen = openInputFile(candidateFileNames[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(candidateFileNames[i]);
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
+ ableToOpen = openInputFile(tryPath, in, "noerror");
+ candidateFileNames[i] = tryPath;
+ }
+ }
+ in.close();
+
if (ableToOpen == 1) {
- m->mothurOut(candidateFileNames[i] + " will be disregarded."); m->mothurOutEndLine();
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
//erase from file list
candidateFileNames.erase(candidateFileNames.begin()+i);
i--;
}
- in.close();
+
}
//make sure there is at least one valid file left
if (candidateFileNames.size() == 0) { m->mothurOut("no valid files."); m->mothurOutEndLine(); abort = true; }
}
-
+
//check for optional parameter and set defaults
// ...at some point should added some additional type checking...
string temp;
int AlignCommand::execute(){
try {
if (abort == true) { return 0; }
-
+
templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch);
int longestBase = templateDB->getLongestBase();
-
+
if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, longestBase); }
else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase); }
else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
vector<string> outputNames;
for (int s = 0; s < candidateFileNames.size(); s++) {
+ if (m->control_pressed) { return 0; }
+
m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
if (outputDir == "") { outputDir += hasPath(candidateFileNames[s]); }
int numFastaSeqs = 0;
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
int start = time(NULL);
+
+#ifdef USE_MPI
+ int pid, end, numSeqsPerProcessor;
+ int tag = 2001;
+ vector<unsigned long int> MPIPos;
+ MPIWroteAccnos = false;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors == 1){
- ifstream inFASTA;
- openInputFile(candidateFileNames[s], inFASTA);
- numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
+ MPI_Status status;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+ MPI_Comm_size(MPI_COMM_WORLD, &processors);
+
+ MPI_File inMPI;
+ MPI_File outMPIAlign;
+ MPI_File outMPIReport;
+ MPI_File outMPIAccnos;
+
+ int outMode=MPI_MODE_CREATE|MPI_MODE_WRONLY;
+ int inMode=MPI_MODE_RDONLY;
+
+ char outAlignFilename[1024];
+ strcpy(outAlignFilename, alignFileName.c_str());
+
+ char outReportFilename[1024];
+ strcpy(outReportFilename, reportFileName.c_str());
+
+ char outAccnosFilename[1024];
+ strcpy(outAccnosFilename, accnosFileName.c_str());
+
+ char inFileName[1024];
+ strcpy(inFileName, candidateFileNames[s].c_str());
- lines.push_back(new linePair(0, numFastaSeqs));
+ MPI_File_open(MPI_COMM_WORLD, inFileName, inMode, MPI_INFO_NULL, &inMPI); //comm, filename, mode, info, filepointer
+ MPI_File_open(MPI_COMM_WORLD, outAlignFilename, outMode, MPI_INFO_NULL, &outMPIAlign);
+ MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport);
+ MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
- driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }
+
+ if (pid == 0) { //you are the root process
+
+ MPIPos = setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
+
+ //send file positions to all processes
+ for(int i = 1; i < processors; i++) {
+ MPI_Send(&numFastaSeqs, 1, MPI_INT, i, tag, MPI_COMM_WORLD);
+ MPI_Send(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, i, tag, MPI_COMM_WORLD);
+ }
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numFastaSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
+
+ //align your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }
+
+ for (int i = 1; i < processors; i++) {
+ bool tempResult;
+ MPI_Recv(&tempResult, 1, MPI_INT, i, tag, MPI_COMM_WORLD, &status);
+ if (tempResult != 0) { MPIWroteAccnos = true; }
+ }
+ }else{ //you are a child process
+ MPI_Recv(&numFastaSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
+ MPIPos.resize(numFastaSeqs+1);
+ MPI_Recv(&MPIPos[0], (numFastaSeqs+1), MPI_LONG, 0, tag, MPI_COMM_WORLD, &status);
+
+
+ //figure out how many sequences you have to align
+ numSeqsPerProcessor = numFastaSeqs / processors;
+ int startIndex = pid * numSeqsPerProcessor;
+ if(pid == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - pid * numSeqsPerProcessor; }
+
+
+ //align your part
+ driverMPI(startIndex, numSeqsPerProcessor, inMPI, outMPIAlign, outMPIReport, outMPIAccnos, MPIPos);
+
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }
+
+ MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
+ }
+
+ //close files
+ MPI_File_close(&inMPI);
+ MPI_File_close(&outMPIAlign);
+ MPI_File_close(&outMPIReport);
+ MPI_File_close(&outMPIAccnos);
+
+ //delete accnos file if blank
+ if (pid == 0) {
+ //delete accnos file if its blank else report to user
+ if (MPIWroteAccnos) {
+ m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
+ if (!flip) {
+ m->mothurOut(" If you set the flip parameter to true mothur will try aligning the reverse compliment as well.");
+ }else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }
+ m->mothurOutEndLine();
+ }else {
+ //MPI_Info info;
+ //MPI_File_delete(outAccnosFilename, info);
+ hasAccnos = false;
+ remove(accnosFileName.c_str());
+ }
+ }
+
+#else
+ vector<unsigned long int> positions = divideFile(candidateFileNames[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){
+ numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); return 0; }
//delete accnos file if its blank else report to user
if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
}else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }
m->mothurOutEndLine();
}
- }
- else{
- vector<int> positions;
+ }else{
processIDS.resize(0);
- ifstream inFASTA;
- openInputFile(candidateFileNames[s], inFASTA);
-
- string input;
- while(!inFASTA.eof()){
- input = getline(inFASTA);
- if (input.length() != 0) {
- if(input[0] == '>'){ long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
- }
- }
- inFASTA.close();
-
- numFastaSeqs = positions.size();
-
- int numSeqsPerProcessor = numFastaSeqs / processors;
-
- for (int i = 0; i < processors; i++) {
- long int startPos = positions[ i * numSeqsPerProcessor ];
- if(i == processors - 1){
- numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor;
- }
- lines.push_back(new linePair(startPos, numSeqsPerProcessor));
- }
-
- createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+ numFastaSeqs = createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str());
rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str());
}else{ m->mothurOut(" If the reverse compliment proved to be better it was reported."); }
m->mothurOutEndLine();
}else{ hasAccnos = false; }
+
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); return 0; }
}
-#else
- ifstream inFASTA;
- openInputFile(candidateFileNames[s], inFASTA);
- numFastaSeqs=count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
- inFASTA.close();
-
- lines.push_back(new linePair(0, numFastaSeqs));
+ #else
+ numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
- driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); return 0; }
//delete accnos file if its blank else report to user
if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
m->mothurOutEndLine();
}
-#endif
-
+ #endif
+
+#endif
+
+
+ #ifdef USE_MPI
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid);
+
+ if (pid == 0) { //only one process should output to screen
+ #endif
+
outputNames.push_back(alignFileName);
outputNames.push_back(reportFileName);
if (hasAccnos) { outputNames.push_back(accnosFileName); }
-
+
+ #ifdef USE_MPI
+ }
+ #endif
+
m->mothurOut("It took " + toString(time(NULL) - start) + " secs to align " + toString(numFastaSeqs) + " sequences.");
m->mothurOutEndLine();
m->mothurOutEndLine();
//**********************************************************************************************************************
-int AlignCommand::driver(linePair* line, string alignFName, string reportFName, string accnosFName, string filename){
+int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
try {
ofstream alignmentFile;
openOutputFile(alignFName, alignmentFile);
ifstream inFASTA;
openInputFile(filename, inFASTA);
- inFASTA.seekg(line->start);
-
- for(int i=0;i<line->numSeqs;i++){
-
+ inFASTA.seekg(filePos->start);
+
+ bool done = false;
+ int count = 0;
+
+ while (!done) {
+
+ if (m->control_pressed) { return 0; }
+
Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA);
+
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * threshold;
//if there is a possibility that this sequence should be reversed
if (candidateSeq->getNumBases() < numBasesNeeded) {
- string wasBetter = "";
+ string wasBetter = "";
//if the user wants you to try the reverse
if (flip) {
//get reverse compliment
delete nast;
nast = nast2;
needToDeleteCopy = true;
+ wasBetter = "\treverse complement produced a better alignment, so mothur used the reverse complement.";
}else{
- wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";
+ wasBetter = "\treverse complement did NOT produce a better alignment so it was not used, please check sequence.";
delete nast2;
delete copy;
}
report.print();
delete nast;
if (needToDeleteCopy) { delete copy; }
+
+ count++;
}
delete candidateSeq;
+ unsigned long int pos = inFASTA.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->numSeqs) % 100 != 0){ m->mothurOut(toString(line->numSeqs)); m->mothurOutEndLine(); }
+ if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
alignmentFile.close();
inFASTA.close();
accnosFile.close();
- return 1;
+ return count;
}
catch(exception& e) {
m->errorOut(e, "AlignCommand", "driver");
exit(1);
}
}
+//**********************************************************************************************************************
+#ifdef USE_MPI
+int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<unsigned long int>& MPIPos){
+ try {
+ string outputString = "";
+ MPI_Status statusReport;
+ MPI_Status statusAlign;
+ MPI_Status statusAccnos;
+ MPI_Status status;
+ int pid;
+ MPI_Comm_rank(MPI_COMM_WORLD, &pid); //find out who we are
+
+ NastReport report;
+
+ if (pid == 0) {
+ outputString = report.getHeaders();
+ int length = outputString.length();
+
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write_shared(reportFile, buf, length, MPI_CHAR, &statusReport);
+
+ delete buf;
+ }
+
+ for(int i=0;i<num;i++){
+
+ if (m->control_pressed) { return 0; }
+
+ //read next sequence
+ int length = MPIPos[start+i+1] - MPIPos[start+i];
+
+ char* buf4 = new char[length];
+ //memcpy(buf4, outputString.c_str(), length);
+
+ MPI_File_read_at(inMPI, MPIPos[start+i], buf4, length, MPI_CHAR, &status);
+
+ string tempBuf = buf4;
+ delete buf4;
+
+ if (tempBuf.length() > length) { tempBuf = tempBuf.substr(0, length); }
+
+ istringstream iss (tempBuf,istringstream::in);
+
+ Sequence* candidateSeq = new Sequence(iss);
+
+ int origNumBases = candidateSeq->getNumBases();
+ string originalUnaligned = candidateSeq->getUnaligned();
+ int numBasesNeeded = origNumBases * threshold;
+
+ if (candidateSeq->getName() != "") { //incase there is a commented sequence at the end of a file
+ if (candidateSeq->getUnaligned().length() > alignment->getnRows()) {
+ alignment->resize(candidateSeq->getUnaligned().length()+1);
+ }
+
+ Sequence temp = templateDB->findClosestSequence(candidateSeq);
+ Sequence* templateSeq = &temp;
+
+ float searchScore = templateDB->getSearchScore();
+
+ Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
+ Sequence* copy;
+
+ Nast* nast2;
+ bool needToDeleteCopy = false; //this is needed in case you have you enter the ifs below
+ //since nast does not make a copy of hte sequence passed, and it is used by the reporter below
+ //you can't delete the copy sequence til after you report, but you may choose not to create it in the first place
+ //so this bool tells you if you need to delete it
+
+ //if there is a possibility that this sequence should be reversed
+ if (candidateSeq->getNumBases() < numBasesNeeded) {
+
+ string wasBetter = "";
+ //if the user wants you to try the reverse
+ if (flip) {
+ //get reverse compliment
+ copy = new Sequence(candidateSeq->getName(), originalUnaligned);
+ copy->reverseComplement();
+
+ //rerun alignment
+ Sequence temp2 = templateDB->findClosestSequence(copy);
+ Sequence* templateSeq2 = &temp2;
+
+ searchScore = templateDB->getSearchScore();
+
+ nast2 = new Nast(alignment, copy, templateSeq2);
+
+ //check if any better
+ if (copy->getNumBases() > candidateSeq->getNumBases()) {
+ candidateSeq->setAligned(copy->getAligned()); //use reverse compliments alignment since its better
+ templateSeq = templateSeq2;
+ delete nast;
+ nast = nast2;
+ needToDeleteCopy = true;
+ }else{
+ wasBetter = "\treverse complement did NOT produce a better alignment, please check sequence.";
+ delete nast2;
+ delete copy;
+ }
+ }
+
+ //create accnos file with names
+ outputString = candidateSeq->getName() + wasBetter + "\n";
+
+ //send results to parent
+ int length = outputString.length();
+
+ char* buf = new char[length];
+ memcpy(buf, outputString.c_str(), length);
+
+ MPI_File_write_shared(accnosFile, buf, length, MPI_CHAR, &statusAccnos);
+ delete buf;
+ MPIWroteAccnos = true;
+ }
+
+ report.setCandidate(candidateSeq);
+ report.setTemplate(templateSeq);
+ report.setSearchParameters(search, searchScore);
+ report.setAlignmentParameters(align, alignment);
+ report.setNastParameters(*nast);
+
+ outputString = ">" + candidateSeq->getName() + "\n" + candidateSeq->getAligned() + "\n";
+
+ //send results to parent
+ int length = outputString.length();
+ char* buf2 = new char[length];
+ memcpy(buf2, outputString.c_str(), length);
+
+ MPI_File_write_shared(alignFile, buf2, length, MPI_CHAR, &statusAlign);
+
+ delete buf2;
+
+ outputString = report.getReport();
+
+ //send results to parent
+ length = outputString.length();
+ char* buf3 = new char[length];
+ memcpy(buf3, outputString.c_str(), length);
+
+ MPI_File_write_shared(reportFile, buf3, length, MPI_CHAR, &statusReport);
+
+ delete buf3;
+ delete nast;
+ if (needToDeleteCopy) { delete copy; }
+ }
+ delete candidateSeq;
+
+ //report progress
+ if((i+1) % 100 == 0){ cout << (toString(i+1)) << endl; }
+ }
+ //report progress
+ if((num) % 100 != 0){ cout << (toString(num)) << endl; }
+
+ return 1;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "driverMPI");
+ exit(1);
+ }
+}
+#endif
/**************************************************************************************************/
-void AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {
+int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
+ int num = 0;
// processIDS.resize(0);
//loop through and create all the processes you want
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){
- driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename);
+ num = driver(lines[process], alignFileName + toString(getpid()) + ".temp", reportFileName + toString(getpid()) + ".temp", accnosFName + toString(getpid()) + ".temp", filename);
+
+ //pass numSeqs to parent
+ ofstream out;
+ string tempFile = alignFileName + toString(getpid()) + ".num.temp";
+ openOutputFile(tempFile, out);
+ out << num << endl;
+ out.close();
+
exit(0);
}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
}
int temp = processIDS[i];
wait(&temp);
}
+
+ for (int i = 0; i < processIDS.size(); i++) {
+ ifstream in;
+ string tempFile = alignFileName + toString(processIDS[i]) + ".num.temp";
+ openInputFile(tempFile, in);
+ if (!in.eof()) { int tempNum = 0; in >> tempNum; num += tempNum; }
+ in.close(); remove(tempFile.c_str());
+ }
+
+ return num;
#endif
}
catch(exception& e) {
exit(1);
}
}
-
/**************************************************************************************************/
void AlignCommand::appendAlignFiles(string temp, string filename) {
exit(1);
}
}
-
//**********************************************************************************************************************
projects / mothur.git / blobdiff