#include "reportfile.h"
#include "qualityscores.h"
#include "refchimeratest.h"
+#include "filterseqscommand.h"
+
//**********************************************************************************************************************
-vector<string> SeqErrorCommand::getValidParameters(){
+vector<string> SeqErrorCommand::setParameters(){
try {
- string Array[] = {"query", "reference", "name", "qfile", "report", "threshold", "inputdir", "outputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ 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);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "getValidParameters");
+ m->errorOut(e, "SeqErrorCommand", "setParameters");
exit(1);
}
}
//**********************************************************************************************************************
-SeqErrorCommand::SeqErrorCommand(){
+string SeqErrorCommand::getHelpString(){
try {
- abort = true; calledHelp = true;
- 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;
+ string helpString = "";
+ helpString += "The seq.error command reads a query alignment file and a reference alignment file and creates .....\n";
+ helpString += "The fasta parameter...\n";
+ helpString += "The reference parameter...\n";
+ helpString += "The qfile parameter...\n";
+ helpString += "The report parameter...\n";
+ helpString += "The name parameter...\n";
+ helpString += "The ignorechimeras parameter...\n";
+ helpString += "The threshold parameter...\n";
+ helpString += "The processors parameter...\n";
+ 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 += "Example seq.error(...).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
+ helpString += "For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
+ m->errorOut(e, "SeqErrorCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> SeqErrorCommand::getRequiredParameters(){
+string SeqErrorCommand::getOutputFileNameTag(string type, string inputName=""){
try {
- string Array[] = {"query","reference"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
+ string outputFileName = "";
+ 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 == "errorsummary") { outputFileName = "error.summary"; }
+ else if (type == "errorseq") { outputFileName = "error.seq"; }
+ else if (type == "errorquality") { outputFileName = "error.quality"; }
+ else if (type == "errorqualforward") { outputFileName = "error.qual.forward"; }
+ else if (type == "errorqualreverse") { outputFileName = "error.qual.reverse"; }
+ else if (type == "errorforward") { outputFileName = "error.seq.forward"; }
+ else if (type == "errorreverse") { outputFileName = "error.seq.reverse"; }
+ else if (type == "errorcount") { outputFileName = "error.count"; }
+ else if (type == "errormatrix") { outputFileName = "error.matrix"; }
+ else if (type == "errorchimera") { outputFileName = "error.chimera"; }
+ else if (type == "errorref-query") { outputFileName = "error.ref-query"; }
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true; }
+ }
+ return outputFileName;
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "getRequiredParameters");
+ m->errorOut(e, "SeqErrorCommand", "getOutputFileNameTag");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> SeqErrorCommand::getRequiredFiles(){
+SeqErrorCommand::SeqErrorCommand(){
try {
- vector<string> myArray;
- return myArray;
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> 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", "getRequiredFiles");
+ m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
exit(1);
}
}
try {
abort = false; calledHelp = false;
+ 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;}
else {
string temp;
-
- //valid paramters for this command
- string AlignArray[] = {"query", "reference", "name", "qfile", "report", "threshold", "inputdir", "ignorechimeras", "outputdir"};
-
- vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
//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 (inputDir == "not found"){ inputDir = ""; }
else {
string path;
- it = parameters.find("query");
+ it = parameters.find("fasta");
//user has given a template 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["query"] = inputDir + it->second; }
+ if (path == "") { parameters["fasta"] = inputDir + it->second; }
}
it = parameters.find("reference");
}
//check for required parameters
- queryFileName = validParameter.validFile(parameters, "query", true);
- if (queryFileName == "not found") { m->mothurOut("query is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
- else if (queryFileName == "not open") { abort = true; }
+ queryFileName = validParameter.validFile(parameters, "fasta", true);
+ if (queryFileName == "not found") {
+ queryFileName = m->getFastaFile();
+ 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") { queryFileName = ""; abort = true; }
+ else { m->setFastaFile(queryFileName); }
referenceFileName = validParameter.validFile(parameters, "reference", true);
if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
else if (referenceFileName == "not open") { abort = true; }
-
//check for optional parameters
namesFileName = validParameter.validFile(parameters, "name", true);
if(namesFileName == "not found"){ namesFileName = ""; }
+ else if (namesFileName == "not open") { namesFileName = ""; abort = true; }
+ else { m->setNameFile(namesFileName); }
qualFileName = validParameter.validFile(parameters, "qfile", true);
if(qualFileName == "not found"){ qualFileName = ""; }
-
+ else if (qualFileName == "not open") { qualFileName = ""; abort = true; }
+ else { m->setQualFile(qualFileName); }
+
reportFileName = validParameter.validFile(parameters, "report", true);
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.");
//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, "ignorechimeras", false); if (temp == "not found") { temp = "1"; }
- convert(temp, ignoreChimeras);
+ 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
+ referenceFileName = validParameter.validFile(parameters, "reference", true);
+ if (referenceFileName == "not found") {
+ //check for saved reference sequences
+ if (rdb->referenceSeqs.size() != 0) {
+ referenceFileName = "saved";
+ }else {
+ m->mothurOut("[ERROR]: You don't have any saved reference sequences and the reference parameter is a required.");
+ m->mothurOutEndLine();
+ abort = true;
+ }
+ }else if (referenceFileName == "not open") { abort = true; }
+ else { if (save) { rdb->setSavedReference(referenceFileName); } }
+
+
+ temp = validParameter.validFile(parameters, "ignorechimeras", false); if (temp == "not found") { temp = "T"; }
+ ignoreChimeras = m->isTrue(temp);
+
+ temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = m->getProcessors(); }
+ m->setProcessors(temp);
+ m->mothurConvert(temp, processors);
substitutionMatrix.resize(6);
- for(int i=0;i<6;i++){ substitutionMatrix[i].assign(6,0); }
+ 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) {
exit(1);
}
}
-
-//**********************************************************************************************************************
-
-void SeqErrorCommand::help(){
- try {
- m->mothurOut("The seq.error command reads a query alignment file and a reference alignment file and creates .....\n");
- m->mothurOut("Example seq.error(...).\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n");
- m->mothurOut("For more details please check out the wiki http://www.mothur.org/wiki/seq.error .\n\n");
- }
- catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "help");
- exit(1);
- }
-}
-
-//***************************************************************************************************************
-
-SeqErrorCommand::~SeqErrorCommand(){
-
-}
-
//***************************************************************************************************************
int SeqErrorCommand::execute(){
try{
if (abort == true) { if (calledHelp) { return 0; } return 2; }
-
- string errorSummaryFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.summary";
- m->openOutputFile(errorSummaryFileName, errorSummaryFile);
- outputNames.push_back(errorSummaryFileName); outputTypes["error.summary"].push_back(errorSummaryFileName);
- printErrorHeader();
- string errorSeqFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq";
- m->openOutputFile(errorSeqFileName, errorSeqFile);
- outputNames.push_back(errorSeqFileName); outputTypes["error.seq"].push_back(errorSeqFileName);
-
+ int start = time(NULL);
+ maxLength = 2000;
+ totalBases = 0;
+ totalMatches = 0;
+
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ string errorSummaryFileName = fileNameRoot + getOutputFileNameTag("errorsummary");
+ outputNames.push_back(errorSummaryFileName); outputTypes["errorsummary"].push_back(errorSummaryFileName);
+
+ string errorSeqFileName = fileNameRoot + getOutputFileNameTag("errorseq");
+ outputNames.push_back(errorSeqFileName); outputTypes["errorseq"].push_back(errorSeqFileName);
+
+ string errorChimeraFileName = fileNameRoot + getOutputFileNameTag("errorchimera");
+ outputNames.push_back(errorChimeraFileName); outputTypes["errorchimera"].push_back(errorChimeraFileName);
+
getReferences(); //read in reference sequences - make sure there's no ambiguous bases
- map<string, int> weights;
if(namesFileName != ""){ weights = getWeights(); }
- ifstream queryFile;
- m->openInputFile(queryFileName, queryFile);
+ vector<unsigned long long> fastaFilePos;
+ vector<unsigned long long> qFilePos;
+ vector<unsigned long long> reportFilePos;
- ifstream reportFile;
- ifstream qualFile;
+ setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
+
+ if (m->control_pressed) { return 0; }
+
+ 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(qualFileName == "") { qLines = lines; rLines = lines; } //fills with duds
+
+ int numSeqs = 0;
+#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 != ""){
+ printErrorQuality(qScoreErrorMap);
+ printQualityFR(qualForwardMap, qualReverseMap);
+ }
+
+ printErrorFRFile(errorForward, errorReverse);
+
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
- ReportFile report;
- QualityScores quality;
- vector<vector<int> > qualForwardMap;
- vector<vector<int> > qualReverseMap;
+ string errorCountFileName = fileNameRoot + getOutputFileNameTag("errorcount");
+ ofstream errorCountFile;
+ m->openOutputFile(errorCountFileName, errorCountFile);
+ 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";
+ for(int i=0;i<misMatchCounts.size();i++){
+ m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
+ errorCountFile << i << '\t' << misMatchCounts[i] << endl;
+ }
+ errorCountFile.close();
+
+ if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
+
+ printSubMatrix();
+
+ string megAlignmentFileName = fileNameRoot + getOutputFileNameTag("errorref-query");
+ ofstream megAlignmentFile;
+ m->openOutputFile(megAlignmentFileName, megAlignmentFile);
+ 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;
+ }
- if(qualFileName != "" && reportFileName != ""){
- m->openInputFile(qualFileName, qualFile);
- report = ReportFile(reportFile, reportFileName);
+ m->mothurOut("It took " + toString(time(NULL) - start) + " secs to check " + toString(numSeqs) + " sequences.");
+ m->mothurOutEndLine();
+
+ m->mothurOutEndLine();
+ m->mothurOut("Output File Names: "); m->mothurOutEndLine();
+ for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
+ m->mothurOutEndLine();
+
+ return 0;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "execute");
+ 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) || (__linux__) || (__unix__) || (__unix)
+
+ //loop through and create all the processes you want
+ while (process != processors) {
+ int pid = fork();
- qualForwardMap.resize(1000);
- qualReverseMap.resize(1000);
- for(int i=0;i<1000;i++){
- qualForwardMap[i].assign(100,0);
- qualReverseMap[i].assign(100,0);
- }
+ if (pid > 0) {
+ 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){
+
+ num = driver(filename, qFileName, rFileName, summaryFileName + toString(getpid()) + ".temp", errorOutputFileName+ toString(getpid()) + ".temp", chimeraOutputFileName + toString(getpid()) + ".temp", lines[process], qLines[process], rLines[process]);
+
+ //pass groupCounts to parent
+ ofstream out;
+ string tempFile = filename + toString(getpid()) + ".info.temp";
+ m->openOutputFile(tempFile, out);
+
+ //output totalBases and totalMatches
+ out << num << '\t' << totalBases << '\t' << totalMatches << endl << endl;
+
+ //output substitutionMatrix
+ for(int i = 0; i < substitutionMatrix.size(); i++) {
+ for (int j = 0; j < substitutionMatrix[i].size(); j++) {
+ out << substitutionMatrix[i][j] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+ //output qScoreErrorMap
+ for (it = qScoreErrorMap.begin(); it != qScoreErrorMap.end(); it++) {
+ vector<int> thisScoreErrorMap = it->second;
+ out << it->first << '\t';
+ for (int i = 0; i < thisScoreErrorMap.size(); i++) {
+ out << thisScoreErrorMap[i] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+ //output qualForwardMap
+ for(int i = 0; i < qualForwardMap.size(); i++) {
+ for (int j = 0; j < qualForwardMap[i].size(); j++) {
+ out << qualForwardMap[i][j] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+ //output qualReverseMap
+ for(int i = 0; i < qualReverseMap.size(); i++) {
+ for (int j = 0; j < qualReverseMap[i].size(); j++) {
+ out << qualReverseMap[i][j] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+
+ //output errorForward
+ for (it = errorForward.begin(); it != errorForward.end(); it++) {
+ vector<int> thisErrorForward = it->second;
+ out << it->first << '\t';
+ for (int i = 0; i < thisErrorForward.size(); i++) {
+ out << thisErrorForward[i] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+ //output errorReverse
+ for (it = errorReverse.begin(); it != errorReverse.end(); it++) {
+ vector<int> thisErrorReverse = it->second;
+ out << it->first << '\t';
+ for (int i = 0; i < thisErrorReverse.size(); i++) {
+ out << thisErrorReverse[i] << '\t';
+ }
+ out << endl;
+ }
+ out << endl;
+
+ //output misMatchCounts
+ out << misMatchCounts.size() << endl;
+ for (int j = 0; j < misMatchCounts.size(); j++) {
+ out << misMatchCounts[j] << '\t';
+ }
+ out << endl;
+
+
+ //output megaAlignVector
+ for (int j = 0; j < megaAlignVector.size(); j++) {
+ out << megaAlignVector[j] << endl;
+ }
+ out << endl;
+
+ out.close();
+
+ 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);
+ }
}
- int totalBases = 0;
- int totalMatches = 0;
+ //do my part
+ num = driver(filename, qFileName, rFileName, summaryFileName, errorOutputFileName, chimeraOutputFileName, lines[0], qLines[0], rLines[0]);
- vector<int> misMatchCounts(11, 0);
+ //force parent to wait until all the processes are done
+ for (int i=0;i<processIDS.size();i++) {
+ int temp = processIDS[i];
+ wait(&temp);
+ }
+
+ //append files
+ for(int i=0;i<processIDS.size();i++){
+
+ m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
+
+ m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
+ m->mothurRemove((summaryFileName + toString(processIDS[i]) + ".temp"));
+ m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
+ m->mothurRemove((errorOutputFileName + toString(processIDS[i]) + ".temp"));
+ m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
+ m->mothurRemove((chimeraOutputFileName + toString(processIDS[i]) + ".temp"));
+
+ ifstream in;
+ string tempFile = filename + toString(processIDS[i]) + ".info.temp";
+ m->openInputFile(tempFile, in);
+
+ //input totalBases and totalMatches
+ int tempBases, tempMatches, tempNumSeqs;
+ in >> tempNumSeqs >> tempBases >> tempMatches; m->gobble(in);
+ totalBases += tempBases; totalMatches += tempMatches; num += tempNumSeqs;
+
+ //input substitutionMatrix
+ int tempNum;
+ for(int i = 0; i < substitutionMatrix.size(); i++) {
+ for (int j = 0; j < substitutionMatrix[i].size(); j++) {
+ in >> tempNum; substitutionMatrix[i][j] += tempNum;
+ }
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input qScoreErrorMap
+ char first;
+ for (int i = 0; i < qScoreErrorMap.size(); i++) {
+ in >> first;
+ vector<int> thisScoreErrorMap = qScoreErrorMap[first];
+
+ for (int i = 0; i < thisScoreErrorMap.size(); i++) {
+ in >> tempNum; thisScoreErrorMap[i] += tempNum;
+ }
+ qScoreErrorMap[first] = thisScoreErrorMap;
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input qualForwardMap
+ for(int i = 0; i < qualForwardMap.size(); i++) {
+ for (int j = 0; j < qualForwardMap[i].size(); j++) {
+ in >> tempNum; qualForwardMap[i][j] += tempNum;
+ }
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input qualReverseMap
+ for(int i = 0; i < qualReverseMap.size(); i++) {
+ for (int j = 0; j < qualReverseMap[i].size(); j++) {
+ in >> tempNum; qualReverseMap[i][j] += tempNum;
+ }
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input errorForward
+ for (int i = 0; i < errorForward.size(); i++) {
+ in >> first;
+ vector<int> thisErrorForward = errorForward[first];
+
+ for (int i = 0; i < thisErrorForward.size(); i++) {
+ in >> tempNum; thisErrorForward[i] += tempNum;
+ }
+ errorForward[first] = thisErrorForward;
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input errorReverse
+ for (int i = 0; i < errorReverse.size(); i++) {
+ in >> first;
+ vector<int> thisErrorReverse = errorReverse[first];
+
+ for (int i = 0; i < thisErrorReverse.size(); i++) {
+ in >> tempNum; thisErrorReverse[i] += tempNum;
+ }
+ errorReverse[first] = thisErrorReverse;
+ m->gobble(in);
+ }
+ m->gobble(in);
+
+ //input misMatchCounts
+ int misMatchSize;
+ in >> misMatchSize; m->gobble(in);
+ if (misMatchSize > misMatchCounts.size()) { misMatchCounts.resize(misMatchSize, 0); }
+ for (int j = 0; j < misMatchCounts.size(); j++) {
+ in >> tempNum; misMatchCounts[j] += tempNum;
+ }
+ m->gobble(in);
+
+ //input megaAlignVector
+ string thisLine;
+ for (int j = 0; j < megaAlignVector.size(); j++) {
+ thisLine = m->getline(in); m->gobble(in); megaAlignVector[j] += thisLine + '\n';
+ }
+ m->gobble(in);
+
+ in.close(); m->mothurRemove(tempFile);
+
+ }
+#endif
+ return num;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "createProcesses");
+ exit(1);
+ }
+}
+//**********************************************************************************************************************
+int SeqErrorCommand::driver(string filename, string qFileName, string rFileName, string summaryFileName, string errorOutputFileName, string chimeraOutputFileName, linePair line, linePair qline, linePair rline) {
+
+ try {
+ ReportFile report;
+ QualityScores quality;
+
+ misMatchCounts.resize(11, 0);
int maxMismatch = 0;
int numSeqs = 0;
map<string, int>::iterator it;
- map<char, vector<int> > qScoreErrorMap;
qScoreErrorMap['m'].assign(41, 0);
qScoreErrorMap['s'].assign(41, 0);
qScoreErrorMap['i'].assign(41, 0);
qScoreErrorMap['a'].assign(41, 0);
- map<char, vector<int> > errorForward;
- errorForward['m'].assign(1000,0);
- errorForward['s'].assign(1000,0);
- errorForward['i'].assign(1000,0);
- errorForward['d'].assign(1000,0);
- errorForward['a'].assign(1000,0);
+ errorForward['m'].assign(maxLength,0);
+ errorForward['s'].assign(maxLength,0);
+ errorForward['i'].assign(maxLength,0);
+ errorForward['d'].assign(maxLength,0);
+ errorForward['a'].assign(maxLength,0);
- map<char, vector<int> > errorReverse;
- errorReverse['m'].assign(1000,0);
- errorReverse['s'].assign(1000,0);
- errorReverse['i'].assign(1000,0);
- errorReverse['d'].assign(1000,0);
- errorReverse['a'].assign(1000,0);
+ errorReverse['m'].assign(maxLength,0);
+ errorReverse['s'].assign(maxLength,0);
+ errorReverse['i'].assign(maxLength,0);
+ errorReverse['d'].assign(maxLength,0);
+ errorReverse['a'].assign(maxLength,0);
-
- string errorChimeraFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.chimera";
- RefChimeraTest chimeraTest(referenceSeqs, errorChimeraFileName);
- outputNames.push_back(errorChimeraFileName); outputTypes["error.chimera"].push_back(errorChimeraFileName);
+ //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 != ""){
+ m->openInputFile(qFileName, qualFile);
+ qualFile.seekg(qline.start);
+
+ //gobble headers
+ if (rline.start == 0) { report = ReportFile(reportFile, rFileName); }
+ else{
+ m->openInputFile(rFileName, reportFile);
+ reportFile.seekg(rline.start);
+ }
+
+ qualForwardMap.resize(maxLength);
+ qualReverseMap.resize(maxLength);
+ for(int i=0;i<maxLength;i++){
+ qualForwardMap[i].assign(41,0);
+ qualReverseMap[i].assign(41,0);
+ }
+ }
+
+ ofstream outChimeraReport;
+ m->openOutputFile(chimeraOutputFileName, outChimeraReport);
+ RefChimeraTest chimeraTest(referenceSeqs);
+ if (line.start == 0) { chimeraTest.printHeader(outChimeraReport); }
+
+ ofstream errorSummaryFile;
+ m->openOutputFile(summaryFileName, errorSummaryFile);
+ if (line.start == 0) { printErrorHeader(errorSummaryFile); }
+
+ ofstream errorSeqFile;
+ m->openOutputFile(errorOutputFileName, errorSeqFile);
+
+ megaAlignVector.resize(numRefs, "");
int index = 0;
bool ignoreSeq = 0;
- while(queryFile){
+ 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; }
- if (m->control_pressed) { errorSummaryFile.close(); errorSeqFile.close(); for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
Sequence query(queryFile);
-
- int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned());
+
+ int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned(), outChimeraReport);
int closestRefIndex = chimeraTest.getClosestRefIndex();
-
+
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());
minCompare.weight = it->second;
}
- else { minCompare.weight = 1; }
-
- printErrorData(minCompare, numParentSeqs);
-
+ else{ minCompare.weight = 1; }
+
+ printErrorData(minCompare, numParentSeqs, errorSummaryFile, errorSeqFile);
+
if(!ignoreSeq){
- for(int i=0;i<minCompare.total;i++){
+
+ for(int i=0;i<minCompare.sequence.length();i++){
char letter = minCompare.sequence[i];
- errorForward[letter][i] += minCompare.weight;
- errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
+
+ if(letter != 'r'){
+ errorForward[letter][i] += minCompare.weight;
+ errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
+ }
}
}
if(qualFileName != "" && reportFileName != ""){
report = ReportFile(reportFile);
- int origLength = report.getQueryLength();
+ // int origLength = report.getQueryLength();
int startBase = report.getQueryStart();
int endBase = report.getQueryEnd();
-
+
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);
+
+// cout << endl;
}
}
}
misMatchCounts[minCompare.mismatches] += minCompare.weight;
numSeqs++;
+
+ megaAlignVector[closestRefIndex] += query.getInlineSeq() + '\n';
}
index++;
- if(index % 1000 == 0){ cout << index << endl; }
+
+ #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'); }
}
queryFile.close();
+ if(qFileName != "" && rFileName != ""){ reportFile.close(); qualFile.close(); }
errorSummaryFile.close();
errorSeqFile.close();
-
- if(qualFileName != "" && reportFileName != ""){
- printErrorQuality(qScoreErrorMap);
- printQualityFR(qualForwardMap, qualReverseMap);
- }
- printErrorFRFile(errorForward, errorReverse);
+ //report progress
+ if(index % 100 != 0){ m->mothurOut(toString(index) + '\n'); }
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
- string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
- ofstream errorCountFile;
- m->openOutputFile(errorCountFileName, errorCountFile);
- outputNames.push_back(errorCountFileName); outputTypes["error.count"].push_back(errorCountFileName);
- m->mothurOut("Overall error rate:\t" + toString((double)(totalBases - totalMatches) / (double)totalBases) + "\n\n");
- m->mothurOut("Errors\tSequences\n");
- errorCountFile << "Errors\tSequences\n";
- for(int i=0;i<misMatchCounts.size();i++){
- m->mothurOut(toString(i) + '\t' + toString(misMatchCounts[i]) + '\n');
- errorCountFile << i << '\t' << misMatchCounts[i] << endl;
- }
- errorCountFile.close();
-
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
- printSubMatrix();
-
- m->mothurOutEndLine();
- m->mothurOut("Output File Names: "); m->mothurOutEndLine();
- for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
- m->mothurOutEndLine();
-
- return 0;
+ return index;
}
catch(exception& e) {
- m->errorOut(e, "SeqErrorCommand", "execute");
+ m->errorOut(e, "SeqErrorCommand", "driver");
exit(1);
}
}
-
//***************************************************************************************************************
void SeqErrorCommand::getReferences(){
try {
+ int numAmbigSeqs = 0;
- ifstream referenceFile;
- m->openInputFile(referenceFileName, referenceFile);
+ int maxStartPos = 0;
+ int minEndPos = 100000;
- int numAmbigSeqs = 0;
+ if (referenceFileName == "saved") {
+ int start = time(NULL);
+ m->mothurOutEndLine(); m->mothurOut("Using sequences from " + rdb->getSavedReference() + " that are saved in memory."); m->mothurOutEndLine();
+
+ for (int i = 0; i < rdb->referenceSeqs.size(); i++) {
+ int numAmbigs = rdb->referenceSeqs[i].getAmbigBases();
+ if(numAmbigs > 0){ numAmbigSeqs++; }
+
+ // int startPos = rdb->referenceSeqs[i].getStartPos();
+ // if(startPos > maxStartPos) { maxStartPos = startPos; }
+ //
+ // int endPos = rdb->referenceSeqs[i].getEndPos();
+ // if(endPos < minEndPos) { minEndPos = endPos; }
+
+ 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();
- while(referenceFile){
- Sequence currentSeq(referenceFile);
- int numAmbigs = currentSeq.getAmbigBases();
- if(numAmbigs > 0){ numAmbigSeqs++; }
- referenceSeqs.push_back(currentSeq);
- m->gobble(referenceFile);
+ }else {
+ int start = time(NULL);
+
+ ifstream referenceFile;
+ m->openInputFile(referenceFileName, referenceFile);
+
+ while(referenceFile){
+ Sequence currentSeq(referenceFile);
+ int numAmbigs = currentSeq.getAmbigBases();
+ if(numAmbigs > 0){ numAmbigSeqs++; }
+
+ // int startPos = currentSeq.getStartPos();
+ // if(startPos > maxStartPos) { maxStartPos = startPos; }
+ //
+ // int endPos = currentSeq.getEndPos();
+ // if(endPos < minEndPos) { minEndPos = endPos; }
+ referenceSeqs.push_back(currentSeq);
+
+ if (rdb->save) { rdb->referenceSeqs.push_back(currentSeq); }
+
+ m->gobble(referenceFile);
+ }
+ referenceFile.close();
+
+ m->mothurOut("It took " + toString(time(NULL) - start) + " to read " + toString(referenceSeqs.size()) + " sequences.");m->mothurOutEndLine();
+ }
+
+ numRefs = referenceSeqs.size();
+
+ for(int i=0;i<numRefs;i++){
+ referenceSeqs[i].padToPos(maxStartPos);
+ referenceSeqs[i].padFromPos(minEndPos);
}
- referenceFile.close();
if(numAmbigSeqs != 0){
m->mothurOut("Warning: " + toString(numAmbigSeqs) + " reference sequences have ambiguous bases, these bases will be ignored\n");
- }
+ }
- numRefs = referenceSeqs.size();
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "getReferences");
//***************************************************************************************************************
-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++){
- if(q[i] != 'N' && q[i] != '.' && r[i] != '.' && (q[i] != '-' || r[i] != '-')){ // no missing data and no double gaps
- started = 1;
-
- if(q[i] == 'A'){
- if(r[i] == 'A'){ errors.AA++; errors.matches++; errors.sequence += 'm'; }
- if(r[i] == 'T'){ errors.AT++; errors.sequence += 's'; }
- if(r[i] == 'G'){ errors.AG++; errors.sequence += 's'; }
- if(r[i] == 'C'){ errors.AC++; errors.sequence += 's'; }
- if(r[i] == '-'){ errors.Ai++; errors.sequence += 'i'; }
- }
- else if(q[i] == 'T'){
- if(r[i] == 'A'){ errors.TA++; errors.sequence += 's'; }
- if(r[i] == 'T'){ errors.TT++; errors.matches++; errors.sequence += 'm'; }
- if(r[i] == 'G'){ errors.TG++; errors.sequence += 's'; }
- if(r[i] == 'C'){ errors.TC++; errors.sequence += 's'; }
- if(r[i] == '-'){ errors.Ti++; errors.sequence += 'i'; }
- }
- else if(q[i] == 'G'){
- if(r[i] == 'A'){ errors.GA++; errors.sequence += 's'; }
- if(r[i] == 'T'){ errors.GT++; errors.sequence += 's'; }
- if(r[i] == 'G'){ errors.GG++; errors.matches++; errors.sequence += 'm'; }
- if(r[i] == 'C'){ errors.GC++; errors.sequence += 's'; }
- if(r[i] == '-'){ errors.Gi++; errors.sequence += 'i'; }
- }
- else if(q[i] == 'C'){
- if(r[i] == 'A'){ errors.CA++; errors.sequence += 's'; }
- if(r[i] == 'T'){ errors.CT++; errors.sequence += 's'; }
- if(r[i] == 'G'){ errors.CG++; errors.sequence += 's'; }
- if(r[i] == 'C'){ errors.CC++; errors.matches++; errors.sequence += 'm'; }
- if(r[i] == '-'){ errors.Ci++; errors.sequence += '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;
+
+ if(q[i] == 'A'){
+ if(r[i] == 'A'){ errors.AA++; errors.matches++; errors.sequence += 'm'; }
+ if(r[i] == 'T'){ errors.AT++; errors.sequence += 's'; }
+ if(r[i] == 'G'){ errors.AG++; errors.sequence += 's'; }
+ if(r[i] == 'C'){ errors.AC++; errors.sequence += 's'; }
+ if(r[i] == '-'){ errors.Ai++; errors.sequence += 'i'; }
+ }
+ else if(q[i] == 'T'){
+ if(r[i] == 'A'){ errors.TA++; errors.sequence += 's'; }
+ if(r[i] == 'T'){ errors.TT++; errors.matches++; errors.sequence += 'm'; }
+ if(r[i] == 'G'){ errors.TG++; errors.sequence += 's'; }
+ if(r[i] == 'C'){ errors.TC++; errors.sequence += 's'; }
+ if(r[i] == '-'){ errors.Ti++; errors.sequence += 'i'; }
+ }
+ else if(q[i] == 'G'){
+ if(r[i] == 'A'){ errors.GA++; errors.sequence += 's'; }
+ if(r[i] == 'T'){ errors.GT++; errors.sequence += 's'; }
+ if(r[i] == 'G'){ errors.GG++; errors.matches++; errors.sequence += 'm'; }
+ if(r[i] == 'C'){ errors.GC++; errors.sequence += 's'; }
+ if(r[i] == '-'){ errors.Gi++; errors.sequence += 'i'; }
+ }
+ else if(q[i] == 'C'){
+ if(r[i] == 'A'){ errors.CA++; errors.sequence += 's'; }
+ if(r[i] == 'T'){ errors.CT++; errors.sequence += 's'; }
+ if(r[i] == 'G'){ errors.CG++; errors.sequence += 's'; }
+ if(r[i] == 'C'){ errors.CC++; errors.matches++; errors.sequence += 'm'; }
+ if(r[i] == '-'){ errors.Ci++; errors.sequence += 'i'; }
+ }
+ else if(q[i] == 'N'){
+ if(r[i] == 'A'){ errors.NA++; errors.sequence += 'a'; }
+ if(r[i] == 'T'){ errors.NT++; errors.sequence += 'a'; }
+ if(r[i] == 'G'){ errors.NG++; errors.sequence += 'a'; }
+ if(r[i] == 'C'){ errors.NC++; errors.sequence += 'a'; }
+ if(r[i] == '-'){ errors.Ni++; errors.sequence += 'a'; }
+ }
+ else if(q[i] == '-' && r[i] != '-'){
+ if(r[i] == 'A'){ errors.dA++; errors.sequence += 'd'; }
+ if(r[i] == 'T'){ errors.dT++; errors.sequence += 'd'; }
+ if(r[i] == 'G'){ errors.dG++; errors.sequence += 'd'; }
+ if(r[i] == 'C'){ errors.dC++; errors.sequence += 'd'; }
+ }
+ errors.total++;
}
- else if(q[i] == 'N'){
- if(r[i] == 'A'){ errors.NA++; errors.sequence += 'a'; }
- if(r[i] == 'T'){ errors.NT++; errors.sequence += 'a'; }
- if(r[i] == 'G'){ errors.NG++; errors.sequence += 'a'; }
- if(r[i] == 'C'){ errors.NC++; errors.sequence += 'a'; }
- if(r[i] == '-'){ errors.Ni++; errors.sequence += 'a'; }
+ else{
+
+ if(q[i] == '-'){
+ errors.sequence += 'd'; errors.total++;
+ }
+ else{
+ errors.sequence += 'r';
+ }
}
- else if(q[i] == '-' && r[i] != '-'){
- if(r[i] == 'A'){ errors.dA++; errors.sequence += 'd'; }
- if(r[i] == 'T'){ errors.dT++; errors.sequence += 'd'; }
- if(r[i] == 'G'){ errors.dG++; errors.sequence += 'd'; }
- if(r[i] == 'C'){ errors.dC++; errors.sequence += 'd'; }
- }
- errors.total++;
-
}
+
else if(q[i] == '.' && r[i] != '.'){ // reference extends beyond query
if(started == 1){ break; }
}
else if(q[i] != '.' && r[i] == '.'){ // query extends beyond reference
-// m->mothurOut("Warning: " + toString(query.getName()) + " extend beyond " + toString(reference.getName()) + ". Ignoring the extra bases in the query\n");
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;
errors.queryName = query.getName();
errors.refName = reference.getName();
- return errors;
+ //return errors;
+ return 0;
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "getErrors");
//***************************************************************************************************************
-void SeqErrorCommand::printErrorHeader(){
+void SeqErrorCommand::printErrorHeader(ofstream& errorSummaryFile){
try {
errorSummaryFile << "query\treference\tweight\t";
errorSummaryFile << "AA\tAT\tAG\tAC\tTA\tTT\tTG\tTC\tGA\tGT\tGG\tGC\tCA\tCT\tCG\tCC\tNA\tNT\tNG\tNC\tAi\tTi\tGi\tCi\tNi\tdA\tdT\tdG\tdC\t";
//***************************************************************************************************************
-void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs){
+void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs, ofstream& errorSummaryFile, ofstream& errorSeqFile){
try {
+
errorSummaryFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
errorSummaryFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
errorSummaryFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
errorSummaryFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
errorSummaryFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
errorSummaryFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
- errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t' ;
+ errorSummaryFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t';
errorSummaryFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t'; //insertions
errorSummaryFile << error.mismatches - (error.Ai + error.Ti + error.Gi + error.Ci) - (error.dA + error.dT + error.dG + error.dC) - (error.NA + error.NT + error.NG + error.NC + error.Ni) << '\t'; //substitutions
errorSummaryFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t'; //ambiguities
errorSummaryFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << '\t' << numParentSeqs << endl;
-
+
errorSeqFile << '>' << error.queryName << "\tref:" << error.refName << '\n' << error.sequence << endl;
-
int a=0; int t=1; int g=2; int c=3;
int gap=4; int n=5;
+
if(numParentSeqs == 1 || ignoreChimeras == 0){
substitutionMatrix[a][a] += error.weight * error.AA;
substitutionMatrix[a][t] += error.weight * error.TA;
substitutionMatrix[a][c] += error.weight * error.CA;
substitutionMatrix[a][gap] += error.weight * error.dA;
substitutionMatrix[a][n] += error.weight * error.NA;
-
+
substitutionMatrix[t][a] += error.weight * error.AT;
substitutionMatrix[t][t] += error.weight * error.TT;
substitutionMatrix[t][g] += error.weight * error.GT;
void SeqErrorCommand::printSubMatrix(){
try {
- string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ string subMatrixFileName = fileNameRoot + getOutputFileNameTag("errormatrix");
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));
+ string errorForwardFileName = fileNameRoot + getOutputFileNameTag("errorforward");
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<1000;i++){
+ for(int i=0;i<maxLength;i++){
float match = (float)errorForward['m'][i];
float subst = (float)errorForward['s'][i];
float insert = (float)errorForward['i'][i];
}
errorForwardFile.close();
- string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
+ string errorReverseFileName = fileNameRoot + getOutputFileNameTag("errorreverse");
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<1000;i++){
+ for(int i=0;i<maxLength;i++){
float match = (float)errorReverse['m'][i];
float subst = (float)errorReverse['s'][i];
float insert = (float)errorReverse['i'][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));
+ string errorQualityFileName = fileNameRoot + getOutputFileNameTag("errorquality");
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++){
//***************************************************************************************************************
void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
- try{
-
- int lastRow = 0;
- int lastColumn = 0;
+ try{
+ int numRows = 0;
+ int numColumns = qualForwardMap[0].size();
for(int i=0;i<qualForwardMap.size();i++){
- for(int j=0;j<qualForwardMap[i].size();j++){
+ for(int j=0;j<numColumns;j++){
if(qualForwardMap[i][j] != 0){
- if(lastRow < i) { lastRow = i+2; }
- if(lastColumn < j) { lastColumn = j+2; }
+ if(numRows < i) { numRows = i+20; }
}
}
}
-
- string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
+ string fileNameRoot = outputDir + m->getRootName(m->getSimpleName(queryFileName));
+ string qualityForwardFileName = fileNameRoot + getOutputFileNameTag("errorqualforward");
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<lastColumn;i++){ qualityForwardFile << '\t' << i; } qualityForwardFile << endl;
+ for(int i=0;i<numColumns;i++){ qualityForwardFile << '\t' << i; } qualityForwardFile << endl;
- for(int i=0;i<lastRow;i++){
+ for(int i=0;i<numRows;i++){
qualityForwardFile << i+1;
- for(int j=0;j<lastColumn;j++){
+ for(int j=0;j<numColumns;j++){
qualityForwardFile << '\t' << qualForwardMap[i][j];
}
qualityForwardFile.close();
- string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
+ string qualityReverseFileName = fileNameRoot + getOutputFileNameTag("errorqualreverse");
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<lastColumn;i++){ qualityReverseFile << '\t' << i; } qualityReverseFile << endl;
- for(int i=0;i<lastRow;i++){
+ for(int i=0;i<numColumns;i++){ qualityReverseFile << '\t' << i; } qualityReverseFile << endl;
+ for(int i=0;i<numRows;i++){
qualityReverseFile << i+1;
- for(int j=0;j<lastColumn;j++){
+ for(int j=0;j<numColumns;j++){
qualityReverseFile << '\t' << qualReverseMap[i][j];
}
qualityReverseFile << endl;
m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
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) || (__linux__) || (__unix__) || (__unix)
+ //set file positions for fasta file
+ fastaFilePos = m->divideFile(filename, processors);
+
+ if (qfilename == "") { return processors; }
+
+ //get name of first sequence in each chunk
+ map<string, int> firstSeqNames;
+ for (int i = 0; i < (fastaFilePos.size()-1); i++) {
+ ifstream in;
+ m->openInputFile(filename, in);
+ in.seekg(fastaFilePos[i]);
+
+ Sequence temp(in);
+ firstSeqNames[temp.getName()] = i;
+
+ in.close();
+ }
+
+ //make copy to use below
+ map<string, int> firstSeqNamesReport = firstSeqNames;
+
+ //seach for filePos of each first name in the qfile and save in qfileFilePos
+ ifstream inQual;
+ m->openInputFile(qfilename, inQual);
+
+ string input;
+ while(!inQual.eof()){
+ input = m->getline(inQual);
+
+ if (input.length() != 0) {
+ if(input[0] == '>'){ //this is a sequence name line
+ istringstream nameStream(input);
+
+ string sname = ""; nameStream >> sname;
+ sname = sname.substr(1);
+
+ map<string, int>::iterator it = firstSeqNames.find(sname);
+
+ if(it != firstSeqNames.end()) { //this is the start of a new chunk
+ unsigned long long pos = inQual.tellg();
+ qfileFilePos.push_back(pos - input.length() - 1);
+ firstSeqNames.erase(it);
+ }
+ }
+ }
+
+ if (firstSeqNames.size() == 0) { break; }
+ }
+ inQual.close();
+
+ if (firstSeqNames.size() != 0) {
+ for (map<string, int>::iterator it = firstSeqNames.begin(); it != firstSeqNames.end(); it++) {
+ m->mothurOut(it->first + " is in your fasta file and not in your quality file, aborting."); m->mothurOutEndLine();
+ }
+ m->control_pressed = true;
+ return processors;
+ }
+
+ //get last file position of qfile
+ FILE * pFile;
+ unsigned long long size;
+
+ //get num bytes in file
+ pFile = fopen (qfilename.c_str(),"rb");
+ if (pFile==NULL) perror ("Error opening file");
+ else{
+ fseek (pFile, 0, SEEK_END);
+ size=ftell (pFile);
+ fclose (pFile);
+ }
+
+ 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);
+ }
+
+ rfileFilePos.push_back(sizeR);
+
+ return processors;
+
+#else
+
+ fastaFilePos.push_back(0); qfileFilePos.push_back(0);
+ //get last file position of fastafile
+ FILE * pFile;
+ unsigned long long size;
+
+ //get num bytes in file
+ pFile = fopen (filename.c_str(),"rb");
+ if (pFile==NULL) perror ("Error opening file");
+ else{
+ fseek (pFile, 0, SEEK_END);
+ size=ftell (pFile);
+ fclose (pFile);
+ }
+ fastaFilePos.push_back(size);
+
+ //get last file position of fastafile
+ FILE * qFile;
+
+ //get num bytes in file
+ qFile = fopen (qfilename.c_str(),"rb");
+ if (qFile==NULL) perror ("Error opening file");
+ else{
+ fseek (qFile, 0, SEEK_END);
+ size=ftell (qFile);
+ fclose (qFile);
+ }
+ qfileFilePos.push_back(size);
+
+ return 1;
+
+#endif
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "setLines");
+ exit(1);
+ }
+}
//***************************************************************************************************************
projects / mothur.git / blobdiff