#include "reportfile.h"
#include "qualityscores.h"
#include "refchimeratest.h"
+#include "myPerseus.h"
#include "filterseqscommand.h"
//**********************************************************************************************************************
+
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", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pname);
+ CommandParameter pcount("count", "InputTypes", "", "", "namecount", "none", "none","",false,false,true); parameters.push_back(pcount);
+ 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); }
exit(1);
}
}
+
//**********************************************************************************************************************
+
string SeqErrorCommand::getHelpString(){
try {
string helpString = "";
helpString += "The reference parameter...\n";
helpString += "The qfile parameter...\n";
helpString += "The report parameter...\n";
- helpString += "The name parameter...\n";
+ helpString += "The name parameter allows you to provide a name file associated with the fasta file.\n";
+ helpString += "The count parameter allows you to provide a count file associated with the fasta file.\n";
helpString += "The ignorechimeras parameter...\n";
helpString += "The threshold parameter...\n";
helpString += "The processors parameter...\n";
exit(1);
}
}
+
+//**********************************************************************************************************************
+
+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");
exit(1);
}
}
+
//***************************************************************************************************************
SeqErrorCommand::SeqErrorCommand(string option) {
//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 the user has not given a path then, add inputdir. else leave path alone.
if (path == "") { parameters["name"] = inputDir + it->second; }
}
+
+ it = parameters.find("count");
+ //user has given a names file
+ if(it != parameters.end()){
+ path = m->hasPath(it->second);
+ //if the user has not given a path then, add inputdir. else leave path alone.
+ if (path == "") { parameters["count"] = inputDir + it->second; }
+ }
it = parameters.find("qfile");
//user has given a quality score file
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);
if(namesFileName == "not found"){ namesFileName = ""; }
else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
else { m->setNameFile(namesFileName); }
+
+ //check for optional parameters
+ countfile = validParameter.validFile(parameters, "count", true);
+ if(countfile == "not found"){ countfile = ""; }
+ else if (countfile == "not open") { countfile = ""; abort = true; }
+ else { m->setCountTableFile(countfile); }
qualFileName = validParameter.validFile(parameters, "qfile", true);
if(qualFileName == "not found"){ qualFileName = ""; }
if(reportFileName == "not found"){ reportFileName = ""; }
else if (reportFileName == "not open") { reportFileName = ""; abort = true; }
- if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
- m->mothurOut("if you use either a qual file or a report file, you have to have both.");
- m->mothurOutEndLine();
- abort = true;
- }
-
outputDir = validParameter.validFile(parameters, "outputdir", false);
if (outputDir == "not found"){
outputDir = "";
outputDir += m->hasPath(queryFileName); //if user entered a file with a path then preserve it
}
+ if ((countfile != "") && (namesFileName != "")) { m->mothurOut("You must enter ONLY ONE of the following: count or name."); m->mothurOutEndLine(); abort = true; }
+
//check for optional parameter and set defaults
// ...at some point should added some additional type checking...
temp = validParameter.validFile(parameters, "threshold", false); if (temp == "not found") { temp = "1.00"; }
- convert(temp, threshold);
+ m->mothurConvert(temp, threshold);
+
temp = validParameter.validFile(parameters, "save", false); if (temp == "not found"){ temp = "f"; }
save = m->isTrue(temp);
temp = validParameter.validFile(parameters, "ignorechimeras", false); if (temp == "not found") { temp = "T"; }
ignoreChimeras = m->isTrue(temp);
+ temp = validParameter.validFile(parameters, "aligned", false); if (temp == "not found"){ temp = "t"; }
+ aligned = m->isTrue(temp);
+
temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
m->setProcessors(temp);
- convert(temp, processors);
+ m->mothurConvert(temp, processors);
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);
+ }
+
+ if(aligned == true){
+ if((reportFileName != "" && qualFileName == "") || (reportFileName == "" && qualFileName != "")){
+ m->mothurOut("if you use either a qual file or a report file, you have to have both.");
+ m->mothurOutEndLine();
+ abort = true;
+ }
+ }
+ else{
+ if(reportFileName != ""){
+ m->mothurOut("we are ignoring the report file if your sequences are not aligned. we will check that the sequences in your fasta and and qual file are the same length.");
+ m->mothurOutEndLine();
+ }
+ }
+
+
}
}
catch(exception& e) {
exit(1);
}
}
+
//***************************************************************************************************************
int SeqErrorCommand::execute(){
if (abort == true) { if (calledHelp) { return 0; } return 2; }
int start = time(NULL);
- maxLength = 2000;
+ maxLength = 5000;
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(namesFileName != ""){ weights = getWeights(); }
+ if(namesFileName != "") { weights = getWeights(); }
+ else if (countfile != "") {
+ CountTable ct;
+ ct.readTable(countfile, false);
+ weights = ct.getNameMap();
+ }
vector<unsigned long long> fastaFilePos;
vector<unsigned long long> qFilePos;
for (int i = 0; i < (fastaFilePos.size()-1); i++) {
lines.push_back(linePair(fastaFilePos[i], fastaFilePos[(i+1)]));
if (qualFileName != "") { qLines.push_back(linePair(qFilePos[i], qFilePos[(i+1)])); }
- if (reportFileName != "") { rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)])); }
+ if (reportFileName != "" && aligned == true) { rLines.push_back(linePair(reportFilePos[i], reportFilePos[(i+1)])); }
}
if(qualFileName == "") { qLines = lines; rLines = lines; } //fills with duds
-
+ if(aligned == false){ rLines = lines; }
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{
numSeqs = createProcesses(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName);
}
#else
numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
#endif
-
- if(qualFileName != "" && reportFileName != ""){
+
+ if(qualFileName != ""){
printErrorQuality(qScoreErrorMap);
printQualityFR(qualForwardMap, qualReverseMap);
}
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";
}
errorCountFile.close();
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
+// if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
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;
megAlignmentFile << megaAlignVector[i] << endl;
}
+ megAlignmentFile.close();
+
m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
m->mothurOutEndLine();
+ //set fasta file as new current fastafile
+ string current = "";
+ itTypes = outputTypes.find("errorseq");
+ if (itTypes != outputTypes.end()) {
+ if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
+ }
+
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
exit(1);
}
}
+
//**********************************************************************************************************************
+
int SeqErrorCommand::createProcesses(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName) {
try {
int process = 1;
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) {
int misMatchSize;
in >> misMatchSize; m->gobble(in);
if (misMatchSize > misMatchCounts.size()) { misMatchCounts.resize(misMatchSize, 0); }
- for (int j = 0; j < misMatchCounts.size(); j++) {
+ for (int j = 0; j < misMatchSize; j++) {
in >> tempNum; misMatchCounts[j] += tempNum;
}
m->gobble(in);
exit(1);
}
}
+
//**********************************************************************************************************************
+
int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
try {
int numSeqs = 0;
map<string, int>::iterator it;
- qScoreErrorMap['m'].assign(41, 0);
- qScoreErrorMap['s'].assign(41, 0);
- qScoreErrorMap['i'].assign(41, 0);
- qScoreErrorMap['a'].assign(41, 0);
+ qScoreErrorMap['m'].assign(101, 0);
+ qScoreErrorMap['s'].assign(101, 0);
+ qScoreErrorMap['i'].assign(101, 0);
+ qScoreErrorMap['a'].assign(101, 0);
errorForward['m'].assign(maxLength,0);
errorForward['s'].assign(maxLength,0);
//open inputfiles and go to beginning place for this processor
ifstream queryFile;
m->openInputFile(filename, queryFile);
+
queryFile.seekg(line.start);
ifstream reportFile;
ifstream qualFile;
- if(qFileName != "" && rFileName != ""){
+ if((qFileName != "" && rFileName != "" && aligned)){
m->openInputFile(qFileName, qualFile);
qualFile.seekg(qline.start);
qualForwardMap.resize(maxLength);
qualReverseMap.resize(maxLength);
for(int i=0;i<maxLength;i++){
- qualForwardMap[i].assign(41,0);
- qualReverseMap[i].assign(41,0);
+ qualForwardMap[i].assign(101,0);
+ qualReverseMap[i].assign(101,0);
}
}
-
+ else if(qFileName != "" && !aligned){
+
+ m->openInputFile(qFileName, qualFile);
+ qualFile.seekg(qline.start);
+
+ qualForwardMap.resize(maxLength);
+ qualReverseMap.resize(maxLength);
+ for(int i=0;i<maxLength;i++){
+ qualForwardMap[i].assign(101,0);
+ qualReverseMap[i].assign(101,0);
+ }
+ }
+
ofstream outChimeraReport;
m->openOutputFile(chimeraOutputFileName, outChimeraReport);
- RefChimeraTest chimeraTest(referenceSeqs);
- if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
+
+
+ RefChimeraTest chimeraTest;
+
+ chimeraTest = RefChimeraTest(referenceSeqs, aligned);
+ if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
+
ofstream errorSummaryFile;
m->openOutputFile(summaryFileName, errorSummaryFile);
ofstream errorSeqFile;
m->openOutputFile(errorOutputFileName, errorSeqFile);
- megaAlignVector.resize(numRefs, "");
+ megaAlignVector.assign(numRefs, "");
int index = 0;
bool ignoreSeq = 0;
bool moreSeqs = 1;
while (moreSeqs) {
-
- if (m->control_pressed) { queryFile.close(); if(qFileName != "" && rFileName != ""){ reportFile.close(); qualFile.close(); } outChimeraReport.close(); errorSummaryFile.close();errorSeqFile.close(); return 0; }
-
+
Sequence query(queryFile);
-
- int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
- int closestRefIndex = chimeraTest.getClosestRefIndex();
-
+ Sequence reference;
+ int numParentSeqs = -1;
+ int closestRefIndex = -1;
+
+ string querySeq = query.getAligned();
+ if (!aligned) { querySeq = query.getUnaligned(); }
+
+ numParentSeqs = chimeraTest.analyzeQuery(query.getName(), querySeq, outChimeraReport);
+
+ closestRefIndex = chimeraTest.getClosestRefIndex();
+
+ reference = referenceSeqs[closestRefIndex];
+
+ reference.setAligned(chimeraTest.getClosestRefAlignment());
+ query.setAligned(chimeraTest.getQueryAlignment());
+
if(numParentSeqs > 1 && ignoreChimeras == 1) { ignoreSeq = 1; }
- else { ignoreSeq = 0; }
-
- Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
+ else { ignoreSeq = 0; }
+
+ Compare minCompare;
+
+ getErrors(query, reference, minCompare);
- if(namesFileName != ""){
+ if((namesFileName != "") || (countfile != "")){
it = weights.find(query.getName());
minCompare.weight = it->second;
}
else{ minCompare.weight = 1; }
+
printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
if(!ignoreSeq){
-
for(int i=0;i<minCompare.sequence.length();i++){
char letter = minCompare.sequence[i];
-
if(letter != 'r'){
errorForward[letter][i] += minCompare.weight;
- errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
+ errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
}
- }
+ }
}
- if(qualFileName != "" && reportFileName != ""){
+ if(aligned && qualFileName != "" && reportFileName != ""){
report = ReportFile(reportFile);
// int origLength = report.getQueryLength();
quality = QualityScores(qualFile);
if(!ignoreSeq){
-// cout << query.getName() << '\t';
-
quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
+ }
+ }
+ else if(aligned == false && qualFileName != ""){
-// cout << endl;
+ quality = QualityScores(qualFile);
+ int qualityLength = quality.getLength();
+
+ if(qualityLength != query.getNumBases()){ cout << "warning - quality and fasta sequence files do not match at " << query.getName() << '\t' << qualityLength <<'\t' << query.getNumBases() << endl; }
+
+ int startBase = 1;
+ int endBase = qualityLength;
+
+ if(!ignoreSeq){
+ quality.updateQScoreErrorMap(qScoreErrorMap, minCompare.sequence, startBase, endBase, minCompare.weight);
+ quality.updateForwardMap(qualForwardMap, startBase, endBase, minCompare.weight);
+ quality.updateReverseMap(qualReverseMap, startBase, endBase, minCompare.weight);
}
- }
-
- if(minCompare.errorRate < threshold && !ignoreSeq){
+ }
+
+ if(minCompare.errorRate <= threshold && !ignoreSeq){
totalBases += (minCompare.total * minCompare.weight);
totalMatches += minCompare.matches * minCompare.weight;
if(minCompare.mismatches > maxMismatch){
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->mothurOutJustToScreen(toString(index)+"\n"); }
}
queryFile.close();
- if(qFileName != "" && rFileName != ""){ reportFile.close(); qualFile.close(); }
- errorSummaryFile.close();
+ outChimeraReport.close();
+ errorSummaryFile.close();
errorSeqFile.close();
-
+
+ if(qFileName != "" && rFileName != "") { reportFile.close(); qualFile.close(); }
+ else if(qFileName != "" && aligned == false){ qualFile.close(); }
+
//report progress
- if(index % 100 != 0){ m->mothurOut(toString(index) + '\n'); }
+ m->mothurOutJustToScreen(toString(index)+"\n");
return index;
}
exit(1);
}
}
+
//***************************************************************************************************************
void SeqErrorCommand::getReferences(){
//
// 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;
+
+ errors.sequence = "";
for(int i=0;i<alignLength;i++){
-// cout << r[i] << '\t' << q[i] << '\t';
+
if(q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){ // no missing data and no double gaps
if(r[i] != 'N'){
started = 1;
errors.sequence += 'r';
}
}
- }
-
+ }
else if(q[i] == '.' && r[i] != '.'){ // reference extends beyond query
if(started == 1){ break; }
}
else if(q[i] == '.' && r[i] == '.'){ // both are missing data
if(started == 1){ break; }
}
-// cout << errors.sequence[errors.sequence.length()-1] << endl;
}
-// cout << errors.sequence << endl;
+
errors.mismatches = errors.total-errors.matches;
- errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total;
+ if(errors.total != 0){ errors.errorRate = (double)(errors.total-errors.matches) / (double)errors.total; }
+ else{ errors.errorRate = 0; }
+
errors.queryName = query.getName();
errors.refName = reference.getName();
-
- return errors;
+
+ return 0;
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "getErrors");
nameCountMap[seqName] = m->getNumNames(redundantSeqs);
m->gobble(nameFile);
}
+
+ nameFile.close();
+
return nameCountMap;
}
-
//***************************************************************************************************************
void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
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";
exit(1);
}
}
+
//***************************************************************************************************************
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++){
+ for(int i=0;i<101;i++){
errorQualityFile << i << '\t' << qScoreErrorMap['m'][i] << '\t' << qScoreErrorMap['s'][i] << '\t' << qScoreErrorMap['i'][i] << '\t'<< qScoreErrorMap['a'][i] << endl;
}
errorQualityFile.close();
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
+ m->errorOut(e, "SeqErrorCommand", "printErrorQuality");
exit(1);
}
}
-
//***************************************************************************************************************
void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
}
}
}
-
- 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++){
qualityReverseFile.close();
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
+ m->errorOut(e, "SeqErrorCommand", "printQualityFR");
exit(1);
}
}
+
/**************************************************************************************************/
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);
string sname = ""; nameStream >> sname;
sname = sname.substr(1);
+
+ m->checkName(sname);
map<string, int>::iterator it = firstSeqNames.find(sname);
qfileFilePos.push_back(size);
- //seach for filePos of each first name in the rfile and save in rfileFilePos
- string junk;
- ifstream inR;
- m->openInputFile(rfilename, inR);
-
- //read column headers
- for (int i = 0; i < 16; i++) {
- if (!inR.eof()) { inR >> junk; }
- else { break; }
- }
-
- while(!inR.eof()){
-
- if (m->control_pressed) { inR.close(); return processors; }
-
- input = m->getline(inR);
-
- if (input.length() != 0) {
-
- istringstream nameStream(input);
- string sname = ""; nameStream >> sname;
-
- map<string, int>::iterator it = firstSeqNamesReport.find(sname);
-
- if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
- unsigned long long pos = inR.tellg();
- rfileFilePos.push_back(pos - input.length() - 1);
- firstSeqNamesReport.erase(it);
- }
- }
-
- if (firstSeqNamesReport.size() == 0) { break; }
- m->gobble(inR);
- }
- inR.close();
-
- if (firstSeqNamesReport.size() != 0) {
- for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
- m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
- }
- m->control_pressed = true;
- return processors;
- }
-
- //get last file position of qfile
- FILE * rFile;
- unsigned long long sizeR;
-
- //get num bytes in file
- rFile = fopen (rfilename.c_str(),"rb");
- if (rFile==NULL) perror ("Error opening file");
- else{
- fseek (rFile, 0, SEEK_END);
- sizeR=ftell (rFile);
- fclose (rFile);
+ if(aligned){
+ //seach for filePos of each first name in the rfile and save in rfileFilePos
+ string junk;
+ ifstream inR;
+
+ m->openInputFile(rfilename, inR);
+
+ //read column headers
+ for (int i = 0; i < 16; i++) {
+ if (!inR.eof()) { inR >> junk; }
+ else { break; }
+ }
+
+ while(!inR.eof()){
+
+ input = m->getline(inR);
+
+ if (input.length() != 0) {
+
+ istringstream nameStream(input);
+ string sname = ""; nameStream >> sname;
+
+ m->checkName(sname);
+
+ map<string, int>::iterator it = firstSeqNamesReport.find(sname);
+
+ if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
+ unsigned long long pos = inR.tellg();
+ rfileFilePos.push_back(pos - input.length() - 1);
+ firstSeqNamesReport.erase(it);
+ }
+ }
+
+ if (firstSeqNamesReport.size() == 0) { break; }
+ m->gobble(inR);
+ }
+ inR.close();
+
+ if (firstSeqNamesReport.size() != 0) {
+ for (map<string, int>::iterator it = firstSeqNamesReport.begin(); it != firstSeqNamesReport.end(); it++) {
+ m->mothurOut(it->first + " is in your fasta file and not in your report file, aborting."); m->mothurOutEndLine();
+ }
+ m->control_pressed = true;
+ return processors;
+ }
+
+ //get last file position of qfile
+ FILE * rFile;
+ unsigned long long sizeR;
+
+ //get num bytes in file
+ rFile = fopen (rfilename.c_str(),"rb");
+ if (rFile==NULL) perror ("Error opening file");
+ else{
+ fseek (rFile, 0, SEEK_END);
+ sizeR=ftell (rFile);
+ fclose (rFile);
+ }
+
+ rfileFilePos.push_back(sizeR);
}
-
- rfileFilePos.push_back(sizeR);
-
return processors;
#else
exit(1);
}
}
+
//***************************************************************************************************************
projects / mothur.git / blobdiff