//**********************************************************************************************************************
vector<string> SeqErrorCommand::setParameters(){
try {
- CommandParameter pquery("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pquery);
- CommandParameter preference("reference", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(preference);
- CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(pqfile);
- CommandParameter preport("report", "InputTypes", "", "", "none", "none", "QualReport",false,false); parameters.push_back(preport);
- CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
- CommandParameter pignorechimeras("ignorechimeras", "Boolean", "", "T", "", "", "",false,false); parameters.push_back(pignorechimeras);
- CommandParameter pthreshold("threshold", "Number", "", "1.0", "", "", "",false,false); parameters.push_back(pthreshold);
- CommandParameter pprocessors("processors", "Number", "", "1", "", "", "",false,false); parameters.push_back(pprocessors);
- CommandParameter psave("save", "Boolean", "", "F", "", "", "",false,false); parameters.push_back(psave);
- CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
- CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
+ CommandParameter pquery("fasta", "InputTypes", "", "", "none", "none", "none","errorType",false,true,true); parameters.push_back(pquery);
+ CommandParameter preference("reference", "InputTypes", "", "", "none", "none", "none","",false,true,true); parameters.push_back(preference);
+ CommandParameter pqfile("qfile", "InputTypes", "", "", "none", "none", "QualReport","",false,false); parameters.push_back(pqfile);
+ CommandParameter preport("report", "InputTypes", "", "", "none", "none", "QualReport","",false,false); parameters.push_back(preport);
+ CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none","",false,false); parameters.push_back(pname);
+ CommandParameter pignorechimeras("ignorechimeras", "Boolean", "", "T", "", "", "","",false,false); parameters.push_back(pignorechimeras);
+ CommandParameter pthreshold("threshold", "Number", "", "1.0", "", "", "","",false,false); parameters.push_back(pthreshold);
+ CommandParameter paligned("aligned", "Boolean", "T", "", "", "", "","",false,false); parameters.push_back(paligned);
+ CommandParameter pprocessors("processors", "Number", "", "1", "", "", "","",false,false,true); parameters.push_back(pprocessors);
+ CommandParameter psave("save", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(psave);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
vector<string> myArray;
for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
}
}
//**********************************************************************************************************************
+string SeqErrorCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "errorsummary") { pattern = "[filename],error.summary"; }
+ else if (type == "errorseq") { pattern = "[filename],error.seq"; }
+ else if (type == "errorquality") { pattern = "[filename],error.quality"; }
+ else if (type == "errorqualforward") { pattern = "[filename],error.qual.forward"; }
+ else if (type == "errorqualreverse") { pattern = "[filename],error.qual.reverse"; }
+ else if (type == "errorforward") { pattern = "[filename],error.seq.forward"; }
+ else if (type == "errorreverse") { pattern = "[filename],error.seq.reverse"; }
+ else if (type == "errorcount") { pattern = "[filename],error.count"; }
+ else if (type == "errormatrix") { pattern = "[filename],error.matrix"; }
+ else if (type == "errorchimera") { pattern = "[filename],error.chimera"; }
+ else if (type == "errorref-query") { pattern = "[filename],error.ref-query"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "getOutputPattern");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
SeqErrorCommand::SeqErrorCommand(){
try {
abort = true; calledHelp = true;
setParameters();
vector<string> tempOutNames;
- outputTypes["error.summary"] = tempOutNames;
- outputTypes["error.seq"] = tempOutNames;
- outputTypes["error.quality"] = tempOutNames;
- outputTypes["error.qual.forward"] = tempOutNames;
- outputTypes["error.qual.reverse"] = tempOutNames;
- outputTypes["error.forward"] = tempOutNames;
- outputTypes["error.reverse"] = tempOutNames;
- outputTypes["error.count"] = tempOutNames;
- outputTypes["error.matrix"] = tempOutNames;
+ outputTypes["errorsummary"] = tempOutNames;
+ outputTypes["errorseq"] = tempOutNames;
+ outputTypes["errorquality"] = tempOutNames;
+ outputTypes["errorqualforward"] = tempOutNames;
+ outputTypes["errorqualreverse"] = tempOutNames;
+ outputTypes["errorforward"] = tempOutNames;
+ outputTypes["errorreverse"] = tempOutNames;
+ outputTypes["errorcount"] = tempOutNames;
+ outputTypes["errormatrix"] = tempOutNames;
+ outputTypes["errorchimera"] = tempOutNames;
+ outputTypes["errorref-query"] = tempOutNames;
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
//initialize outputTypes
vector<string> tempOutNames;
- outputTypes["error.summary"] = tempOutNames;
- outputTypes["error.seq"] = tempOutNames;
- outputTypes["error.quality"] = tempOutNames;
- outputTypes["error.qual.forward"] = tempOutNames;
- outputTypes["error.qual.reverse"] = tempOutNames;
- outputTypes["error.forward"] = tempOutNames;
- outputTypes["error.reverse"] = tempOutNames;
- outputTypes["error.count"] = tempOutNames;
- outputTypes["error.matrix"] = tempOutNames;
+ outputTypes["errorsummary"] = tempOutNames;
+ outputTypes["errorseq"] = tempOutNames;
+ outputTypes["errorquality"] = tempOutNames;
+ outputTypes["errorqualforward"] = tempOutNames;
+ outputTypes["errorqualreverse"] = tempOutNames;
+ outputTypes["errorforward"] = tempOutNames;
+ outputTypes["errorreverse"] = tempOutNames;
+ outputTypes["errorcount"] = tempOutNames;
+ outputTypes["errormatrix"] = tempOutNames;
+ outputTypes["errorchimera"] = tempOutNames;
+ outputTypes["errorref-query"] = tempOutNames;
//if the user changes the input directory command factory will send this info to us in the output parameter
if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
else { m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
}
- else if (queryFileName == "not open") { abort = true; }
+ else if (queryFileName == "not open") { queryFileName = ""; abort = true; }
else { m->setFastaFile(queryFileName); }
referenceFileName = validParameter.validFile(parameters, "reference", true);
// ...at some point should added some additional type checking...
temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found") { temp = "1.00"; }
m->mothurConvert(temp, threshold);
+
+ temp = validParameter.validFile(parameters, "aligned", true); if (temp == "not found"){ temp = "t"; }
+ aligned = m->isTrue(temp);
+// rdb->aligned = aligned; #do we need these lines for aligned?
+// if (aligned) { //clear out old references
+// rdb->clearMemory();
+// }
temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
save = m->isTrue(temp);
substitutionMatrix.resize(6);
for(int i=0;i<6;i++){ substitutionMatrix[i].resize(6,0); }
+
+ if ((namesFileName == "") && (queryFileName != "")){
+ vector<string> files; files.push_back(queryFileName);
+ parser.getNameFile(files);
+ }
}
}
catch(exception& e) {
totalBases = 0;
totalMatches = 0;
- string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
- outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ map<string, string> variables;
+ variables["[filename]"] = fileNameRoot;
+ string errorSummaryFileName = getOutputFileName("errorsummary",variables);
+ outputNames.push_back(errorSummaryFileName); outputTypes["errorsummary"].push_back(errorSummaryFileName);
- string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
- outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
+ string errorSeqFileName = getOutputFileName("errorseq",variables);
+ outputNames.push_back(errorSeqFileName); outputTypes["errorseq"].push_back(errorSeqFileName);
- string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
- outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
+ string errorChimeraFileName = getOutputFileName("errorchimera",variables);
+ outputNames.push_back(errorChimeraFileName); outputTypes["errorchimera"].push_back(errorChimeraFileName);
getReferences(); //read in reference sequences - make sure there's no ambiguous bases
if(qualFileName == "") { qLines = lines; rLines = lines; } //fills with duds
int numSeqs = 0;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
if(processors == 1){
numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
}else{
if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
- string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
+ string errorCountFileName = getOutputFileName("errorcount",variables);
ofstream errorCountFile;
m->openOutputFile(errorCountFileName, errorCountFile);
- outputNames.push_back(errorCountFileName); outputTypes["error.count"].push_back(errorCountFileName);
+ outputNames.push_back(errorCountFileName); outputTypes["errorcount"].push_back(errorCountFileName);
m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n");
m->mothurOut("Errors\tSequences\n");
errorCountFile << "Errors\tSequences\n";
printSubMatrix();
- string megAlignmentFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.ref-query";
+ string megAlignmentFileName = getOutputFileName("errorref-query",variables);
ofstream megAlignmentFile;
m->openOutputFile(megAlignmentFileName, megAlignmentFile);
- outputNames.push_back(megAlignmentFileName); outputTypes["error.ref-query"].push_back(megAlignmentFileName);
+ outputNames.push_back(megAlignmentFileName); outputTypes["errorref-query"].push_back(megAlignmentFileName);
for(int i=0;i<numRefs;i++){
megAlignmentFile << referenceSeqs[i].getInlineSeq() << endl;
processIDS.clear();
map<char, vector<int> >::iterator it;
int num = 0;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//loop through and create all the processes you want
while (process != processors) {
ofstream outChimeraReport;
m->openOutputFile(chimeraOutputFileName, outChimeraReport);
- RefChimeraTest chimeraTest(referenceSeqs);
- if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
+
+ RefChimeraTest chimeraTest(referenceSeqs, aligned);
+
+ if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
ofstream errorSummaryFile;
m->openOutputFile(summaryFileName, errorSummaryFile);
if(numParentSeqs > 1 && ignoreChimeras == 1) { ignoreSeq = 1; }
else { ignoreSeq = 0; }
- Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
+ Compare minCompare;
+ getErrors(query, referenceSeqs[closestRefIndex], minCompare);
if(namesFileName != ""){
it = weights.find(query.getName());
index++;
- #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+ #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
unsigned long long pos = queryFile.tellg();
if ((pos == -1) || (pos >= line.end)) { break; }
#else
if (queryFile.eof()) { break; }
#endif
- if(index % 100 == 0){ m->mothurOut(toString(index) + '\n'); }
+ if(index % 100 == 0){ m->mothurOut(toString(index)); m->mothurOutEndLine(); }
}
queryFile.close();
if(qFileName != "" && rFileName != ""){ reportFile.close(); qualFile.close(); }
errorSeqFile.close();
//report progress
- if(index % 100 != 0){ m->mothurOut(toString(index) + '\n'); }
+ if(index % 100 != 0){ m->mothurOut(toString(index)); m->mothurOutEndLine(); }
return index;
}
//
// int endPos = rdb->referenceSeqs[i].getEndPos();
// if(endPos < minEndPos) { minEndPos = endPos; }
+ if (rdb->referenceSeqs[i].getNumBases() == 0) {
+ m->mothurOut("[WARNING]: " + rdb->referenceSeqs[i].getName() + " is blank, ignoring.");m->mothurOutEndLine();
+ }else {
+ referenceSeqs.push_back(rdb->referenceSeqs[i]);
+ }
- referenceSeqs.push_back(rdb->referenceSeqs[i]);
}
referenceFileName = rdb->getSavedReference();
- m->mothurOut("It took " + toString(time(NULL) - start) + " to load " + toString(rdb->referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
+ m->mothurOut("It took " + toString(time(NULL) - start) + " to load " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
}else {
int start = time(NULL);
//
// int endPos = currentSeq.getEndPos();
// if(endPos < minEndPos) { minEndPos = endPos; }
- referenceSeqs.push_back(currentSeq);
-
- if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
+ if (currentSeq.getNumBases() == 0) {
+ m->mothurOut("[WARNING]: " + currentSeq.getName() + " is blank, ignoring.");m->mothurOutEndLine();
+ }else {
+ referenceSeqs.push_back(currentSeq);
+ if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
+ }
m->gobble(referenceFile);
}
for(int i=0;i<numRefs;i++){
referenceSeqs[i].padToPos(maxStartPos);
referenceSeqs[i].padFromPos(minEndPos);
- }
+ }
if(numAmbigSeqs != 0){
m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
//***************************************************************************************************************
-Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
+int SeqErrorCommand::getErrors(Sequence query, Sequence reference, Compare& errors){
try {
if(query.getAlignLength() != reference.getAlignLength()){
m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
string r = reference.getAligned();
int started = 0;
- Compare errors;
+ //Compare errors;
for(int i=0;i<alignLength;i++){
// cout << r[i] << '\t' << q[i] << '\t';
errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
errors.queryName = query.getName();
errors.refName = reference.getName();
-
- return errors;
+ //return errors;
+ return 0;
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "getErrors");
void SeqErrorCommand::printSubMatrix(){
try {
- string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ map<string, string> variables;
+ variables["[filename]"] = fileNameRoot;
+ string subMatrixFileName = getOutputFileName("errormatrix",variables);
ofstream subMatrixFile;
m->openOutputFile(subMatrixFileName, subMatrixFile);
- outputNames.push_back(subMatrixFileName); outputTypes["error.matrix"].push_back(subMatrixFileName);
+ outputNames.push_back(subMatrixFileName); outputTypes["errormatrix"].push_back(subMatrixFileName);
vector<string> bases(6);
bases[0] = "A";
bases[1] = "T";
void SeqErrorCommand::printErrorFRFile(map<char, vector<int> > errorForward, map<char, vector<int> > errorReverse){
try{
- string errorForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.forward";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ map<string, string> variables;
+ variables["[filename]"] = fileNameRoot;
+ string errorForwardFileName = getOutputFileName("errorforward",variables);
ofstream errorForwardFile;
m->openOutputFile(errorForwardFileName, errorForwardFile);
- outputNames.push_back(errorForwardFileName); outputTypes["error.forward"].push_back(errorForwardFileName);
+ outputNames.push_back(errorForwardFileName); outputTypes["errorforward"].push_back(errorForwardFileName);
errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
for(int i=0;i<maxLength;i++){
}
errorForwardFile.close();
- string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
+ string errorReverseFileName = getOutputFileName("errorreverse",variables);
ofstream errorReverseFile;
m->openOutputFile(errorReverseFileName, errorReverseFile);
- outputNames.push_back(errorReverseFileName); outputTypes["error.reverse"].push_back(errorReverseFileName);
+ outputNames.push_back(errorReverseFileName); outputTypes["errorreverse"].push_back(errorReverseFileName);
errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
for(int i=0;i<maxLength;i++){
void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
try{
-
- string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ map<string, string> variables;
+ variables["[filename]"] = fileNameRoot;
+ string errorQualityFileName = getOutputFileName("errorquality",variables);
ofstream errorQualityFile;
m->openOutputFile(errorQualityFileName, errorQualityFile);
- outputNames.push_back(errorQualityFileName); outputTypes["error.quality"].push_back(errorQualityFileName);
+ outputNames.push_back(errorQualityFileName); outputTypes["errorquality"].push_back(errorQualityFileName);
errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
for(int i=0;i<41;i++){
}
}
}
-
- string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ map<string, string> variables;
+ variables["[filename]"] = fileNameRoot;
+ string qualityForwardFileName = getOutputFileName("errorqualforward",variables);
ofstream qualityForwardFile;
m->openOutputFile(qualityForwardFileName, qualityForwardFile);
- outputNames.push_back(qualityForwardFileName); outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
+ outputNames.push_back(qualityForwardFileName); outputTypes["errorqualforward"].push_back(qualityForwardFileName);
for(int i=0;i<numColumns;i++){ qualityForwardFile << '\t' << i; } qualityForwardFile << endl;
qualityForwardFile.close();
- string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
+ string qualityReverseFileName = getOutputFileName("errorqualreverse",variables);
ofstream qualityReverseFile;
m->openOutputFile(qualityReverseFileName, qualityReverseFile);
- outputNames.push_back(qualityReverseFileName); outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
+ outputNames.push_back(qualityReverseFileName); outputTypes["errorqualreverse"].push_back(qualityReverseFileName);
for(int i=0;i<numColumns;i++){ qualityReverseFile << '\t' << i; } qualityReverseFile << endl;
for(int i=0;i<numRows;i++){
int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long long>& fastaFilePos, vector<unsigned long long>& qfileFilePos, vector<unsigned long long>& rfileFilePos) {
try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
//set file positions for fasta file
fastaFilePos = m->divideFile(filename, processors);
projects / mothur.git / blobdiff