AlignCommand::AlignCommand(string option){
try {
-// globaldata = GlobalData::getInstance();
+ // globaldata = GlobalData::getInstance();
abort = false;
//allow user to run help
if(option == "help") { help(); abort = true; }
else {
-
+
//valid paramters for this command
string AlignArray[] = {"template","candidate","search","ksize","align","match","mismatch","gapopen","gapextend", "processors"};
vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
map<string, string> parameters = parser.getParameters();
ValidParameters validParameter;
-
+
//check to make sure all parameters are valid for command
for (map<string, string>::iterator it = parameters.begin(); it != parameters.end(); it++) {
if (validParameter.isValidParameter(it->first, myArray, it->second) != true) { abort = true; }
templateFileName = validParameter.validFile(parameters, "template", true);
if (templateFileName == "not found") { cout << "template is a required parameter for the align.seqs command." << endl; abort = true; }
else if (templateFileName == "not open") { abort = true; }
-
+
candidateFileName = validParameter.validFile(parameters, "candidate", true);
if (candidateFileName == "not found") { cout << "candidate is a required parameter for the align.seqs command." << endl; abort = true; }
else if (candidateFileName == "not open") { abort = true; }
string temp;
temp = validParameter.validFile(parameters, "ksize", false); if (temp == "not found"){ temp = "8"; }
convert(temp, kmerSize);
-
+
temp = validParameter.validFile(parameters, "match", false); if (temp == "not found"){ temp = "1.0"; }
convert(temp, match);
-
+
temp = validParameter.validFile(parameters, "mismatch", false); if (temp == "not found"){ temp = "-1.0"; }
convert(temp, misMatch);
-
- temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-1.0"; }
+
+ temp = validParameter.validFile(parameters, "gapopen", false); if (temp == "not found"){ temp = "-2.0"; }
convert(temp, gapOpen);
-
- temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-2.0"; }
+
+ temp = validParameter.validFile(parameters, "gapextend", false); if (temp == "not found"){ temp = "-1.0"; }
convert(temp, gapExtend);
-
+
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
convert(temp, processors);
-
+
search = validParameter.validFile(parameters, "search", false); if (search == "not found"){ search = "kmer"; }
align = validParameter.validFile(parameters, "align", false); if (align == "not found"){ align = "needleman"; }
}
-
+
}
catch(exception& e) {
cout << "Standard Error: " << e.what() << " has occurred in the AlignCommand class Function AlignCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
void AlignCommand::help(){
try {
cout << "The align.seqs command reads a file containing sequences and creates an alignment file and a report file." << "\n";
- cout << "The align.seqs command parameters are fasta, candidate, search, ksize, align, match, mismatch, gapopen and gapextend. " << "\n";
- cout << "The fasta and candidate parameters are required." << "\n";
+ cout << "The align.seqs command parameters are template, candidate, search, ksize, align, match, mismatch, gapopen and gapextend. " << "\n";
+ cout << "The template and candidate parameters are required." << "\n";
cout << "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";
cout << "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";
cout << "The ksize parameter allows you to specify the kmer size for finding most similar template to candidate. The default is 7." << "\n";
cout << "The gapopen parameter allows you to specify the penalty for opening a gap in an alignment. The default is -1.0." << "\n";
cout << "The gapextend parameter allows you to specify the penalty for extending a gap in an alignment. The default is -2.0." << "\n";
cout << "The align.seqs command should be in the following format: " << "\n";
- cout << "align.seqs(fasta=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) " << "\n";
- cout << "Example align.seqs(candidate=candidate.fasta, fasta=core.filtered, align=kmer, search=gotoh, ksize=8, match=2.0, mismatch=3.0, gapopen=-2.0, gapextend=-1.0)" << "\n";
- cout << "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFastaFile)." << "\n" << "\n";
+ cout << "align.seqs(template=yourTemplateFile, candidate=yourCandidateFile, align=yourAlignmentMethod, search=yourSearchmethod, ksize=yourKmerSize, match=yourMatchBonus, mismatch=yourMismatchpenalty, gapopen=yourGapopenPenalty, gapextend=yourGapExtendPenalty) " << "\n";
+ cout << "Example align.seqs(candidate=candidate.fasta, template=core.filtered, align=kmer, search=gotoh, ksize=8, match=2.0, mismatch=3.0, gapopen=-2.0, gapextend=-1.0)" << "\n";
+ cout << "Note: No spaces between parameter labels (i.e. candidate), '=' and parameters (i.e.yourFastaFile)." << "\n" << "\n";
}
catch(exception& e) {
cout << "Standard Error: " << e.what() << " has occurred in the AlignCommand class Function help. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
int AlignCommand::execute(){
try {
if (abort == true) { return 0; }
-
+
srand( (unsigned)time( NULL ) ); //needed to assign names to temporary files
-
+
if(search == "kmer") { templateDB = new KmerDB(templateFileName, kmerSize); }
else if(search == "suffix") { templateDB = new SuffixDB(templateFileName); }
else if(search == "blast") { templateDB = new BlastDB(templateFileName, gapOpen, gapExtend, match, misMatch); }
cout << search << " is not a valid search option. I will run the command using kmer, ksize=8." << endl; kmerSize = 8;
templateDB = new KmerDB(templateFileName, kmerSize);
}
-
+
if(align == "gotoh") { alignment = new GotohOverlap(gapOpen, gapExtend, match, misMatch, 3000); }
else if(align == "needleman") { alignment = new NeedlemanOverlap(gapOpen, match, misMatch, 3000); }
else if(align == "blast") { alignment = new BlastAlignment(gapOpen, gapExtend, match, misMatch); }
}
else{
vector<int> positions;
+ processIDS.resize(0);
ifstream inFASTA;
openInputFile(candidateFileName, inFASTA);
-
+
while(!inFASTA.eof()){
char c = inFASTA.get();
if(c == '>'){ positions.push_back(inFASTA.tellg()); }
lines.push_back(new linePair(startPos, numSeqsPerProcessor));
}
createProcesses(alignFileName, reportFileName);
-
- //append and remove temp files
- map<int, int>::iterator it = processIDS.begin();
- rename((alignFileName + toString(it->second) + ".temp").c_str(), alignFileName.c_str());
- rename((reportFileName + toString(it->second) + ".temp").c_str(), alignFileName.c_str());
- it++;
+ rename((alignFileName + toString(processIDS[0]) + ".temp").c_str(), alignFileName.c_str());
+ rename((reportFileName + toString(processIDS[0]) + ".temp").c_str(), reportFileName.c_str());
- for (; it != processIDS.end(); it++) {
- appendAlignFiles((alignFileName + toString(it->second) + ".temp"), alignFileName);
- remove((alignFileName + toString(it->second) + ".temp").c_str());
-
- appendReportFiles((reportFileName + toString(it->second) + ".temp"), reportFileName);
- remove((reportFileName + toString(it->second) + ".temp").c_str());
+ for(int i=1;i<processors;i++){
+ appendAlignFiles((alignFileName + toString(processIDS[i]) + ".temp"), alignFileName);
+ remove((alignFileName + toString(processIDS[i]) + ".temp").c_str());
+
+ appendReportFiles((reportFileName + toString(processIDS[i]) + ".temp"), reportFileName);
+ remove((reportFileName + toString(processIDS[i]) + ".temp").c_str());
}
}
#else
driver(lines[0], alignFileName, reportFileName);
#endif
-
+
cout << "It took " << time(NULL) - start << " secs to align " << numFastaSeqs << " sequences" << endl;
cout << endl;
-
-
-
+
+
+
return 0;
}
catch(exception& e) {
delete candidateSeq;
}
-
+
+ alignmentFile.close();
+ inFASTA.close();
+
return 1;
}
catch(exception& e) {
try {
#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
int process = 0;
- processIDS.clear();
+ // processIDS.resize(0);
//loop through and create all the processes you want
while (process != processors) {
int pid = fork();
-
+
if (pid > 0) {
- processIDS[lines[process]->numSeqs] = pid; //create map from line number to pid so you can append files in correct order later
+ 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");
}
//force parent to wait until all the processes are done
- for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
- int temp = it->second;
+ for (int i=0;i<processors;i++) {
+ int temp = processIDS[i];
wait(&temp);
}
#endif
ifstream input;
openInputFile(temp, input);
while (!input.eof()) { char c = input.get(); if (c == 10 || c == 13){ break; } } // get header line
-
+
string line;
//read input file and write to output file
while(input.eof() != true) {
projects / mothur.git / blobdiff