//**********************************************************************************************************************
-
+vector<string> AlignCommand::getValidParameters(){
+ try {
+ string AlignArray[] = {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors","flip","threshold","outputdir","inputdir"};
+ vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getValidParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> AlignCommand::getRequiredParameters(){
+ try {
+ string AlignArray[] = {"template","candidate"};
+ vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getRequiredParameters");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+vector<string> AlignCommand::getRequiredFiles(){
+ try {
+ vector<string> myArray;
+ return myArray;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "getRequiredFiles");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+AlignCommand::AlignCommand(){
+ try {
+ abort = true;
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AlignCommand", "AlignCommand");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
AlignCommand::AlignCommand(string option) {
try {
-
abort = false;
//allow user to run help
for (it = parameters.begin(); it != parameters.end(); it++) {
if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
}
-
+
+ //initialize outputTypes
+ vector<string> tempOutNames;
+ outputTypes["fasta"] = tempOutNames;
+ outputTypes["alignreport"] = tempOutNames;
+ outputTypes["accnos"] = tempOutNames;
+
//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 = ""; }
//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["template"] = inputDir + it->second; }
}
candidateFileName = validParameter.validFile(parameters, "candidate", false);
if (candidateFileName == "not found") { m->mothurOut("candidate is a required parameter for the align.seqs command."); m->mothurOutEndLine(); abort = true; }
else {
- splitAtDash(candidateFileName, candidateFileNames);
+ m->splitAtDash(candidateFileName, candidateFileNames);
//go through files and make sure they are good, if not, then disregard them
for (int i = 0; i < candidateFileNames.size(); i++) {
+ //candidateFileNames[i] = m->getFullPathName(candidateFileNames[i]);
+
if (inputDir != "") {
- string path = hasPath(candidateFileNames[i]);
+ string path = m->hasPath(candidateFileNames[i]);
//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, "noerror");
-
+ ableToOpen = m->openInputFile(candidateFileNames[i], in, "noerror");
+ in.close();
+
//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]);
+ string tryPath = m->getDefaultPath() + m->getSimpleName(candidateFileNames[i]);
m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying default " + tryPath); m->mothurOutEndLine();
- ableToOpen = openInputFile(tryPath, in, "noerror");
+ ifstream in2;
+ ableToOpen = m->openInputFile(tryPath, in2, "noerror");
+ in2.close();
candidateFileNames[i] = tryPath;
}
}
- in.close();
+
+ //if you can't open it, try default location
+ if (ableToOpen == 1) {
+ if (m->getOutputDir() != "") { //default path is set
+ string tryPath = m->getOutputDir() + m->getSimpleName(candidateFileNames[i]);
+ m->mothurOut("Unable to open " + candidateFileNames[i] + ". Trying output directory " + tryPath); m->mothurOutEndLine();
+ ifstream in2;
+ ableToOpen = m->openInputFile(tryPath, in2, "noerror");
+ in2.close();
+ candidateFileNames[i] = tryPath;
+ }
+ }
+
+
if (ableToOpen == 1) {
m->mothurOut("Unable to open " + candidateFileNames[i] + ". It will be disregarded."); m->mothurOutEndLine();
convert(temp, processors);
temp = validParameter.validFile(parameters, "flip", false); if (temp == "not found"){ temp = "f"; }
- flip = isTrue(temp);
+ flip = m->isTrue(temp);
temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found"){ temp = "0.50"; }
convert(temp, threshold);
exit(1);
}
}
-
//**********************************************************************************************************************
AlignCommand::~AlignCommand(){
void AlignCommand::help(){
try {
m->mothurOut("The align.seqs command reads a file containing sequences and creates an alignment file and a report file.\n");
- m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend.\n");
+ m->mothurOut("The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen, gapextend and processors.\n");
m->mothurOut("The template and candidate parameters are required. You may enter multiple fasta files by separating their names with dashes. ie. fasta=abrecovery.fasta-amzon.fasta \n");
m->mothurOut("The search parameter allows you to specify the method to find most similar template. Your options are: suffix, kmer and blast. The default is kmer.\n");
m->mothurOut("The align parameter allows you to specify the alignment method to use. Your options are: gotoh, needleman, blast and noalign. The default is needleman.\n");
m->mothurOutEndLine();
alignment = new NeedlemanOverlap(gapOpen, match, misMatch, longestBase);
}
- vector<string> outputNames;
for (int s = 0; s < candidateFileNames.size(); s++) {
- if (m->control_pressed) { return 0; }
+ if (m->control_pressed) { outputTypes.clear(); return 0; }
m->mothurOut("Aligning sequences from " + candidateFileNames[s] + " ..." ); m->mothurOutEndLine();
- if (outputDir == "") { outputDir += hasPath(candidateFileNames[s]); }
- string alignFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align";
- string reportFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "align.report";
- string accnosFileName = outputDir + getRootName(getSimpleName(candidateFileNames[s])) + "flip.accnos";
+ 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";
bool hasAccnos = true;
int numFastaSeqs = 0;
int start = time(NULL);
#ifdef USE_MPI
- int pid, end, numSeqsPerProcessor;
+ int pid, numSeqsPerProcessor;
int tag = 2001;
vector<unsigned long int> MPIPos;
MPIWroteAccnos = false;
MPI_File_open(MPI_COMM_WORLD, outReportFilename, outMode, MPI_INFO_NULL, &outMPIReport);
MPI_File_open(MPI_COMM_WORLD, outAccnosFilename, outMode, MPI_INFO_NULL, &outMPIAccnos);
- if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); return 0; }
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
if (pid == 0) { //you are the root process
- MPIPos = setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
+ MPIPos = m->setFilePosFasta(candidateFileNames[s], numFastaSeqs); //fills MPIPos, returns numSeqs
//send file positions to all processes
for(int i = 1; i < processors; i++) {
//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; }
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
for (int i = 1; i < processors; i++) {
bool tempResult;
//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; }
+ if (m->control_pressed) { MPI_File_close(&inMPI); MPI_File_close(&outMPIAlign); MPI_File_close(&outMPIReport); MPI_File_close(&outMPIAccnos); outputTypes.clear(); return 0; }
MPI_Send(&MPIWroteAccnos, 1, MPI_INT, 0, tag, MPI_COMM_WORLD);
}
}
#else
- vector<unsigned long int> positions = divideFile(candidateFileNames[s], processors);
-
+
+ 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(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; }
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); outputTypes.clear(); return 0; }
//delete accnos file if its blank else report to user
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
+ if (m->isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
else {
m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
if (!flip) {
vector<string> nonBlankAccnosFiles;
//delete blank accnos files generated with multiple processes
for(int i=0;i<processors;i++){
- if (!(isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
+ if (!(m->isBlank(accnosFileName + toString(processIDS[i]) + ".temp"))) {
nonBlankAccnosFiles.push_back(accnosFileName + toString(processIDS[i]) + ".temp");
}else { remove((accnosFileName + toString(processIDS[i]) + ".temp").c_str()); }
}
m->mothurOutEndLine();
}else{ hasAccnos = false; }
- if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); return 0; }
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); outputTypes.clear(); return 0; }
}
#else
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; }
+ if (m->control_pressed) { remove(accnosFileName.c_str()); remove(alignFileName.c_str()); remove(reportFileName.c_str()); outputTypes.clear(); return 0; }
//delete accnos file if its blank else report to user
- if (isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
+ if (m->isBlank(accnosFileName)) { remove(accnosFileName.c_str()); hasAccnos = false; }
else {
m->mothurOut("Some of you sequences generated alignments that eliminated too many bases, a list is provided in " + accnosFileName + ".");
if (!flip) {
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); }
+ outputNames.push_back(alignFileName); outputTypes["fasta"].push_back(alignFileName);
+ outputNames.push_back(reportFileName); outputTypes["alignreport"].push_back(reportFileName);
+ if (hasAccnos) { outputNames.push_back(accnosFileName); outputTypes["accnos"].push_back(accnosFileName); }
#ifdef USE_MPI
}
int AlignCommand::driver(linePair* filePos, string alignFName, string reportFName, string accnosFName, string filename){
try {
ofstream alignmentFile;
- openOutputFile(alignFName, alignmentFile);
+ m->openOutputFile(alignFName, alignmentFile);
ofstream accnosFile;
- openOutputFile(accnosFName, accnosFile);
+ m->openOutputFile(accnosFName, accnosFile);
NastReport report(reportFName);
ifstream inFASTA;
- openInputFile(filename, inFASTA);
+ m->openInputFile(filename, inFASTA);
inFASTA.seekg(filePos->start);
if (m->control_pressed) { return 0; }
- Sequence* candidateSeq = new Sequence(inFASTA); gobble(inFASTA);
-
+ Sequence* candidateSeq = new Sequence(inFASTA); m->gobble(inFASTA);
+ report.setCandidate(candidateSeq);
+
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * threshold;
Sequence temp = templateDB->findClosestSequence(candidateSeq);
Sequence* templateSeq = &temp;
-
+
float searchScore = templateDB->getSearchScore();
Nast* nast = new Nast(alignment, candidateSeq, templateSeq);
+
Sequence* copy;
Nast* nast2;
string wasBetter = "";
//if the user wants you to try the reverse
if (flip) {
+
//get reverse compliment
copy = new Sequence(candidateSeq->getName(), originalUnaligned);
copy->reverseComplement();
accnosFile << candidateSeq->getName() << wasBetter << endl;
}
- report.setCandidate(candidateSeq);
report.setTemplate(templateSeq);
report.setSearchParameters(search, searchScore);
report.setAlignmentParameters(align, alignment);
}
delete candidateSeq;
- unsigned long int pos = inFASTA.tellg();
- if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ unsigned long int pos = inFASTA.tellg();
+ if ((pos == -1) || (pos >= filePos->end)) { break; }
+ #else
+ if (inFASTA.eof()) { break; }
+ #endif
//report progress
if((count) % 100 == 0){ m->mothurOut(toString(count)); m->mothurOutEndLine(); }
istringstream iss (tempBuf,istringstream::in);
Sequence* candidateSeq = new Sequence(iss);
-
+ report.setCandidate(candidateSeq);
+
int origNumBases = candidateSeq->getNumBases();
string originalUnaligned = candidateSeq->getUnaligned();
int numBasesNeeded = origNumBases * threshold;
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.";
delete nast2;
MPIWroteAccnos = true;
}
- report.setCandidate(candidateSeq);
report.setTemplate(templateSeq);
report.setSearchParameters(search, searchScore);
report.setAlignmentParameters(align, alignment);
//pass numSeqs to parent
ofstream out;
- string tempFile = toString(getpid()) + ".temp";
- openOutputFile(tempFile, out);
+ string tempFile = alignFileName + 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); }
+ }else {
+ m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
+ for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
+ exit(0);
+ }
}
//force parent to wait until all the processes are done
for (int i = 0; i < processIDS.size(); i++) {
ifstream in;
- string tempFile = toString(processIDS[i]) + ".temp";
- openInputFile(tempFile, 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());
}
ofstream output;
ifstream input;
- openOutputFileAppend(filename, output);
- openInputFile(temp, input);
+ m->openOutputFileAppend(filename, output);
+ m->openInputFile(temp, input);
while(char c = input.get()){
if(input.eof()) { break; }
ofstream output;
ifstream input;
- openOutputFileAppend(filename, output);
- openInputFile(temp, input);
+ m->openOutputFileAppend(filename, output);
+ m->openInputFile(temp, input);
while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line
projects / mothur.git / blobdiff