*/
#include "aligncommand.h"
-#include "sequence.hpp"
-
-#include "gotohoverlap.hpp"
-#include "needlemanoverlap.hpp"
-#include "blastalign.hpp"
-#include "noalign.hpp"
-
-#include "nast.hpp"
-#include "nastreport.hpp"
+#include "referencedb.h"
//**********************************************************************************************************************
vector<string> AlignCommand::setParameters(){
CommandParameter pgapextend("gapextend", "Number", "", "-1.0", "", "", "",false,false); parameters.push_back(pgapextend);
CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
CommandParameter pflip("flip", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(pflip);
+ CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
CommandParameter pthreshold("threshold", "Number", "", "0.50", "", "", "",false,false); parameters.push_back(pthreshold);
CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
helpString += "The flip parameter is used to specify whether or not you want mothur to try the reverse complement if a sequence falls below the threshold. The default is false.";
helpString += "The threshold is used to specify a cutoff at which an alignment is deemed 'bad' and the reverse complement may be tried. The default threshold is 0.50, meaning 50% of the bases are removed in the alignment.";
helpString += "If the flip parameter is set to true the reverse complement of the sequence is aligned and the better alignment is reported.";
+ helpString += "If the save parameter is set to true the reference sequences will be saved in memory, to clear them later you can use the clear.memory command. Default=f.";
helpString += "The default for the threshold parameter is 0.50, meaning at least 50% of the bases must remain or the sequence is reported as potentially reversed.";
helpString += "The align.seqs command should be in the following format:";
helpString += "align.seqs(reference=yourTemplateFile, fasta=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty)";
exit(1);
}
}
+//**********************************************************************************************************************
+string AlignCommand::getOutputFileNameTag(string type, string inputName=""){
+ try {
+ string tag = "";
+ map<string, vector<string> >::iterator it;
+
+ //is this a type this command creates
+ it = outputTypes.find(type);
+ if (it == outputTypes.end()) { m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+ else {
+ if (type == "fasta") { tag = "align"; }
+ else if (type == "alignreport") { tag = "align.report"; }
+ else if (type == "accnos") { tag = "flip.accnos"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return tag;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getOutputFileName");
+ exit(1);
+ }
+}
+
//**********************************************************************************************************************
AlignCommand::AlignCommand(){
try {
//**********************************************************************************************************************
AlignCommand::AlignCommand(string option) {
try {
- abort = false; calledHelp = false;
-
+ abort = false; calledHelp = false;
+ ReferenceDB* rdb = ReferenceDB::getInstance();
+
//allow user to run help
if(option == "help") { help(); abort = true; calledHelp = true;}
else if(option == "citation") { citation(); abort = true; calledHelp = true;}
}
}
- //check for required parameters
- templateFileName = validParameter.validFile(parameters, "reference", true);
-
- if (templateFileName == "not found") {
- m->mothurOut("reference is a required parameter for the align.seqs command.");
- m->mothurOutEndLine();
- abort = true;
- }else if (templateFileName == "not open") { abort = true; }
-
candidateFileName = validParameter.validFile(parameters, "fasta", false);
if (candidateFileName == "not found") {
//if there is a current fasta file, use it
//erase from file list
candidateFileNames.erase(candidateFileNames.begin()+i);
i--;
+ }else {
+ m->setFastaFile(candidateFileNames[i]);
}
}
}
// ...at some point should added some additional type checking...
string temp;
temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; }
- convert(temp, kmerSize);
+ m->mothurConvert(temp, kmerSize);
temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }
- convert(temp, match);
+ m->mothurConvert(temp, match);
temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }
- convert(temp, misMatch);
+ m->mothurConvert(temp, misMatch);
temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; }
- convert(temp, gapOpen);
+ m->mothurConvert(temp, gapOpen);
temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; }
- convert(temp, gapExtend);
+ m->mothurConvert(temp, gapExtend);
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
- convert(temp, processors);
+ m->mothurConvert(temp, processors);
temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; }
flip = m->isTrue(temp);
+ temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
+ save = m->isTrue(temp);
+ rdb->save = save;
+ if (save) { //clear out old references
+ rdb->clearMemory();
+ }
+
+ //this has to go after save so that if the user sets save=t and provides no reference we abort
+ templateFileName = validParameter.validFile(parameters, "reference", true);
+ if (templateFileName == "not found") {
+ //check for saved reference sequences
+ if (rdb->referenceSeqs.size() != 0) {
+ templateFileName = "saved";
+ }else {
+ m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required for the align.seqs command.");
+ m->mothurOutEndLine();
+ abort = true;
+ }
+ }else if (templateFileName == "not open") { abort = true; }
+ else { if (save) { rdb->setSavedReference(templateFileName); } }
+
temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; }
- convert(temp, threshold);
+ m->mothurConvert(temp, threshold);
search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }
if ((search != "suffix") && (search != "kmer") && (search != "blast")) { m->mothurOut("invalid search option: choices are kmer, suffix or blast."); m->mothurOutEndLine(); abort=true; }
if (abort == false) {
for (int i = 0; i < lines.size(); i++) { delete lines[i]; } lines.clear();
delete templateDB;
- delete alignment;
}
}
//**********************************************************************************************************************
try {
if (abort == true) { if (calledHelp) { return 0; } return 2; }
- 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); }
- else if(align == "noalign") { alignment = new NoAlign(); }
- else {
- m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
- m->mothurOutEndLine();
- alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
- }
+ templateDB = new AlignmentDB(templateFileName, search, kmerSize, gapOpen, gapExtend, match, misMatch, rand());
for (int s = 0; s < candidateFileNames.size(); s++) {
if (m->control_pressed) { outputTypes.clear(); return 0; }
m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
if (outputDir == "") { outputDir += m->hasPath(candidateFileNames[s]); }
- string alignFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "align";
- string reportFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "align.report";
- string accnosFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + "flip.accnos";
+ string alignFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + getOutputFileNameTag("fasta");
+ string reportFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + getOutputFileNameTag("alignreport");
+ string accnosFileName = outputDir + m->getRootName(m->getSimpleName(candidateFileNames[s])) + getOutputFileNameTag("accnos");
bool hasAccnos = true;
int numFastaSeqs = 0;
#ifdef USE_MPI
int pid, numSeqsPerProcessor;
int tag = 2001;
- vector<unsigned long int> MPIPos;
+ vector<unsigned long long> MPIPos;
MPIWroteAccnos = false;
MPI_Status status;
//MPI_Info info;
//MPI_File_delete(outAccnosFilename, info);
hasAccnos = false;
- remove(accnosFileName.c_str());
+ m->mothurRemove(accnosFileName);
}
}
#else
- vector<unsigned long int> positions = m->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)
+ vector<unsigned long long> positions;
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ positions = m->divideFile(candidateFileNames[s], processors);
+ for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(new linePair(positions[i], positions[(i+1)])); }
+ #else
+ if (processors == 1) {
+ lines.push_back(new linePair(0, 1000));
+ }else {
+ positions = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs);
+ if (positions.size() < processors) { processors = positions.size(); }
+
+ //figure out how many sequences you have to process
+ int numSeqsPerProcessor = numFastaSeqs / processors;
+ for (int i = 0; i < processors; i++) {
+ int startIndex = i * numSeqsPerProcessor;
+ if(i == (processors - 1)){ numSeqsPerProcessor = numFastaSeqs - i * numSeqsPerProcessor; }
+ lines.push_back(new linePair(positions[startIndex], numSeqsPerProcessor));
+ }
+ }
+ #endif
+
if(processors == 1){
numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
}else{
numFastaSeqs = createProcesses(alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
}
- #else
- numFastaSeqs = driver(lines[0], alignFileName, reportFileName, accnosFileName, candidateFileNames[s]);
- #endif
-
- if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); outputTypes.clear(); return 0; }
+
+ if (m->control_pressed) { m->mothurRemove(accnosFileName); m->mothurRemove(alignFileName); m->mothurRemove(reportFileName); outputTypes.clear(); return 0; }
//delete accnos file if its blank else report to user
- if (m->isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
+ if (m->isBlank(accnosFileName)) { m->mothurRemove(accnosFileName); hasAccnos = false; }
else {
m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
if (!flip) {
}
//**********************************************************************************************************************
-
int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
try {
ofstream alignmentFile;
bool done = false;
int count = 0;
+
+ //moved this into driver to avoid deep copies in windows paralellized version
+ Alignment* alignment;
+ 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); }
+ else if(align == "noalign") { alignment = new NoAlign(); }
+ else {
+ m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
+ m->mothurOutEndLine();
+ alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
+ }
while (!done) {
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { break; }
Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
report.setCandidate(candidateSeq);
Sequence temp = templateDB->findClosestSequence(candidateSeq);
Sequence* templateSeq = &temp;
-
+
float searchScore = templateDB->getSearchScore();
Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
}
delete candidateSeq;
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- unsigned long int pos = inFASTA.tellg();
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+ unsigned long long pos = inFASTA.tellg();
if ((pos == -1) || (pos >= filePos->end)) { break; }
#else
if (inFASTA.eof()) { break; }
//report progress
if((count) % 100 != 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
+ delete alignment;
alignmentFile.close();
inFASTA.close();
accnosFile.close();
}
//**********************************************************************************************************************
#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){
+int AlignCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_File& alignFile, MPI_File& reportFile, MPI_File& accnosFile, vector<unsigned long long>& MPIPos){
try {
string outputString = "";
MPI_Status statusReport;
delete buf;
}
+ Alignment* alignment;
+ 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); }
+ else if(align == "noalign") { alignment = new NoAlign(); }
+ else {
+ m->mothurOut(align + " is not a valid alignment option. I will run the command using needleman.");
+ m->mothurOutEndLine();
+ alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
+ }
+
+
for(int i=0;i<num;i++){
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { delete alignment; return 0; }
//read next sequence
int length = MPIPos[start+i+1] - MPIPos[start+i];
int AlignCommand::createProcesses(string alignFileName, string reportFileName, string accnosFName, string filename) {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ int num = 0;
processIDS.resize(0);
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
int process = 1;
- int num = 0;
- // processIDS.resize(0);
//loop through and create all the processes you want
while (process != processors) {
vector<string> nonBlankAccnosFiles;
if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
- else { remove(accnosFName.c_str()); } //remove so other files can be renamed to it
+ else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
for (int i = 0; i < processIDS.size(); i++) {
ifstream in;
string tempFile = alignFileName + 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());
+ in.close(); m->mothurRemove(tempFile);
- appendAlignFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
- remove((alignFileName + toString(processIDS[i]) + ".temp").c_str());
+ m->appendFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
+ m->mothurRemove((alignFileName + toString(processIDS[i]) + ".temp"));
appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
- remove((reportFileName + toString(processIDS[i]) + ".temp").c_str());
+ m->mothurRemove((reportFileName + toString(processIDS[i]) + ".temp"));
if (!(m->isBlank(accnosFName + toString(processIDS[i]) + ".temp"))) {
nonBlankAccnosFiles.push_back(accnosFName + toString(processIDS[i]) + ".temp");
- }else { remove((accnosFName + toString(processIDS[i]) + ".temp").c_str()); }
+ }else { m->mothurRemove((accnosFName + toString(processIDS[i]) + ".temp")); }
}
rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
- appendAlignFiles(nonBlankAccnosFiles[h], accnosFName);
- remove(nonBlankAccnosFiles[h].c_str());
+ m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
+ m->mothurRemove(nonBlankAccnosFiles[h]);
}
}else { //recreate the accnosfile if needed
ofstream out;
m->openOutputFile(accnosFName, out);
out.close();
}
+#else
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
+ //Windows version shared memory, so be careful when passing variables through the alignData struct.
+ //Above fork() will clone, so memory is separate, but that's not the case with windows,
+ //////////////////////////////////////////////////////////////////////////////////////////////////////
- return num;
-#endif
- }
- catch(exception& e) {
- m->errorOut(e, "AlignCommand", "createProcesses");
- exit(1);
- }
-}
-/**************************************************************************************************/
-
-void AlignCommand::appendAlignFiles(string temp, string filename) {
- try{
+ vector<alignData*> pDataArray;
+ DWORD dwThreadIdArray[processors-1];
+ HANDLE hThreadArray[processors-1];
- ofstream output;
- ifstream input;
- m->openOutputFileAppend(filename, output);
- m->openInputFile(temp, input);
+ //Create processor worker threads.
+ for( int i=0; i<processors-1; i++ ){
+ //copy templateDb
+ //AlignmentDB* tempDB = new AlignmentDB(*templateDB);
+
+ // Allocate memory for thread data.
+ string extension = "";
+ if (i != 0) { extension = toString(i) + ".temp"; }
+
+ alignData* tempalign = new alignData(templateFileName, (alignFileName + extension), (reportFileName + extension), (accnosFName + extension), filename, align, search, kmerSize, m, lines[i]->start, lines[i]->end, flip, match, misMatch, gapOpen, gapExtend, threshold, i);
+ pDataArray.push_back(tempalign);
+ processIDS.push_back(i);
+
+ //MySeqSumThreadFunction is in header. It must be global or static to work with the threads.
+ //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
+ hThreadArray[i] = CreateThread(NULL, 0, MyAlignThreadFunction, pDataArray[i], 0, &dwThreadIdArray[i]);
+ }
- while(char c = input.get()){
- if(input.eof()) { break; }
- else { output << c; }
+ //need to check for line ending error
+ ifstream inFASTA;
+ m->openInputFile(filename, inFASTA);
+ inFASTA.seekg(lines[processors-1]->start-1);
+ char c = inFASTA.peek();
+
+ if (c != '>') { //we need to move back
+ lines[processors-1]->start--;
}
- input.close();
- output.close();
+ //using the main process as a worker saves time and memory
+ //do my part - do last piece because windows is looking for eof
+ num = driver(lines[processors-1], (alignFileName + toString(processors-1) + ".temp"), (reportFileName + toString(processors-1) + ".temp"), (accnosFName + toString(processors-1) + ".temp"), filename);
+
+ //Wait until all threads have terminated.
+ WaitForMultipleObjects(processors-1, hThreadArray, TRUE, INFINITE);
+
+ //Close all thread handles and free memory allocations.
+ for(int i=0; i < pDataArray.size(); i++){
+ num += pDataArray[i]->count;
+ CloseHandle(hThreadArray[i]);
+ delete pDataArray[i];
+ }
+
+ vector<string> nonBlankAccnosFiles;
+ if (!(m->isBlank(accnosFName))) { nonBlankAccnosFiles.push_back(accnosFName); }
+ else { m->mothurRemove(accnosFName); } //remove so other files can be renamed to it
+
+ for (int i = 1; i < processors; i++) {
+ m->appendFiles((alignFileName + toString(i) + ".temp"), alignFileName);
+ m->mothurRemove((alignFileName + toString(i) + ".temp"));
+
+ appendReportFiles((reportFileName + toString(i) + ".temp"), reportFileName);
+ m->mothurRemove((reportFileName + toString(i) + ".temp"));
+
+ if (!(m->isBlank(accnosFName + toString(i) + ".temp"))) {
+ nonBlankAccnosFiles.push_back(accnosFName + toString(i) + ".temp");
+ }else { m->mothurRemove((accnosFName + toString(i) + ".temp")); }
+ }
+
+ //append accnos files
+ if (nonBlankAccnosFiles.size() != 0) {
+ rename(nonBlankAccnosFiles[0].c_str(), accnosFName.c_str());
+
+ for (int h=1; h < nonBlankAccnosFiles.size(); h++) {
+ m->appendFiles(nonBlankAccnosFiles[h], accnosFName);
+ m->mothurRemove(nonBlankAccnosFiles[h]);
+ }
+ }else { //recreate the accnosfile if needed
+ ofstream out;
+ m->openOutputFile(accnosFName, out);
+ out.close();
+ }
+#endif
+
+ return num;
}
catch(exception& e) {
- m->errorOut(e, "AlignCommand", "appendAlignFiles");
+ m->errorOut(e, "AlignCommand", "createProcesses");
exit(1);
}
}
while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line
- while(char c = input.get()){
- if(input.eof()) { break; }
- else { output << c; }
- }
+ char buffer[4096];
+ while (!input.eof()) {
+ input.read(buffer, 4096);
+ output.write(buffer, input.gcount());
+ }
input.close();
output.close();