*/
#include "seqerrorcommand.h"
+#include "reportfile.h"
+#include "qualityscores.h"
+#include "refchimeratest.h"
//**********************************************************************************************************************
vector<string> SeqErrorCommand::getValidParameters(){
try {
- string Array[] = {"query", "reference", "name", "threshold", "inputdir", "outputdir"};
+ string Array[] = {"query", "reference", "name", "qfile", "report", "threshold", "inputdir", "outputdir"};
vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
return myArray;
}
//**********************************************************************************************************************
SeqErrorCommand::SeqErrorCommand(){
try {
- //initialize outputTypes
+ abort = true; calledHelp = true;
vector<string> tempOutNames;
- outputTypes["error"] = tempOutNames;
- outputTypes["count"] = 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;
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "SeqErrorCommand");
SeqErrorCommand::SeqErrorCommand(string option) {
try {
- abort = false;
+ abort = false; calledHelp = false;
//allow user to run help
- if(option == "help") { help(); abort = true; }
+ if(option == "help") { help(); abort = true; calledHelp = true; }
else {
string temp;
//valid paramters for this command
- string AlignArray[] = {"query", "reference", "name", "threshold", "inputdir", "outputdir"};
-
-//need to implement name file option
+ string AlignArray[] = {"query", "reference", "name", "qfile", "report", "threshold", "inputdir", "ignorechimeras", "outputdir"};
vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
//initialize outputTypes
vector<string> tempOutNames;
- outputTypes["error"] = tempOutNames;
- outputTypes["count"] = 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;
+
//if the user changes the input directory command factory will send this info to us in the output parameter
string inputDir = validParameter.validFile(parameters, "inputdir", false);
}
it = parameters.find("name");
- //user has given a template file
+ //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["name"] = inputDir + it->second; }
}
+
+ it = parameters.find("qfile");
+ //user has given a quality score 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["qfile"] = inputDir + it->second; }
+ }
+
+ it = parameters.find("report");
+ //user has given a alignment report 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["report"] = inputDir + it->second; }
+ }
}
//check for required parameters
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; }
- //if the user changes the output directory command factory will send this info to us in the output parameter
+
+ //check for optional parameters
namesFileName = validParameter.validFile(parameters, "name", true);
if(namesFileName == "not found"){ namesFileName = ""; }
- cout << namesFileName << endl;
+
+ qualFileName = validParameter.validFile(parameters, "qfile", true);
+ if(qualFileName == "not found"){ qualFileName = ""; }
+
+ reportFileName = validParameter.validFile(parameters, "report", true);
+ if(reportFileName == "not found"){ reportFileName = ""; }
+
+ 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"){
// ...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);
-
- errorFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".errors";
- m->openOutputFile(errorFileName, errorFile);
- outputNames.push_back(errorFileName); outputTypes["error"].push_back(errorFileName);
- printErrorHeader();
+
+ temp = validParameter.validFile(parameters, "ignorechimeras", false); if (temp == "not found") { temp = "1"; }
+ convert(temp, ignoreChimeras);
+
+ substitutionMatrix.resize(6);
+ for(int i=0;i<6;i++){ substitutionMatrix[i].assign(6,0); }
}
}
catch(exception& e) {
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");
//***************************************************************************************************************
-SeqErrorCommand::~SeqErrorCommand(){ errorFile.close(); }
+SeqErrorCommand::~SeqErrorCommand(){
+
+}
//***************************************************************************************************************
int SeqErrorCommand::execute(){
try{
- if (abort == true) { return 0; }
+ 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);
getReferences(); //read in reference sequences - make sure there's no ambiguous bases
ifstream queryFile;
m->openInputFile(queryFileName, queryFile);
-
+
+ ifstream reportFile;
+ ifstream qualFile;
+
+ ReportFile report;
+ QualityScores quality;
+ vector<vector<int> > qualForwardMap;
+ vector<vector<int> > qualReverseMap;
+
+ if(qualFileName != "" && reportFileName != ""){
+ m->openInputFile(qualFileName, qualFile);
+ report = ReportFile(reportFile, reportFileName);
+
+ qualForwardMap.resize(1000);
+ qualReverseMap.resize(1000);
+ for(int i=0;i<1000;i++){
+ qualForwardMap[i].assign(100,0);
+ qualReverseMap[i].assign(100,0);
+ }
+ }
+
int totalBases = 0;
int totalMatches = 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);
+
+ 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);
+
+
+ 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);
+
+ int index = 0;
+ bool ignoreSeq = 0;
while(queryFile){
- Compare minCompare;
+ 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);
-
- for(int i=0;i<numRefs;i++){
- Compare currCompare = getErrors(query, referenceSeqs[i]);
-
- if(currCompare.errorRate < minCompare.errorRate){
- minCompare = currCompare;
- }
-
- }
+
+ int numParentSeqs = chimeraTest.analyzeQuery(query.getName(), query.getAligned());
+ int closestRefIndex = chimeraTest.getClosestRefIndex();
+ if(numParentSeqs > 1 && ignoreChimeras == 1) { ignoreSeq = 1; }
+ else { ignoreSeq = 0; }
+
+
+ Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
+
if(namesFileName != ""){
it = weights.find(query.getName());
minCompare.weight = it->second;
}
- else {
- minCompare.weight = 1;
+ else { minCompare.weight = 1; }
+
+ printErrorData(minCompare, numParentSeqs);
+
+ if(!ignoreSeq){
+ for(int i=0;i<minCompare.total;i++){
+ char letter = minCompare.sequence[i];
+ errorForward[letter][i] += minCompare.weight;
+ errorReverse[letter][minCompare.total-i-1] += minCompare.weight;
+ }
}
+
+ if(qualFileName != "" && reportFileName != ""){
+ report = ReportFile(reportFile);
+
+ int origLength = report.getQueryLength();
+ int startBase = report.getQueryStart();
+ int endBase = report.getQueryEnd();
- printErrorData(minCompare);
+ quality = QualityScores(qualFile, origLength);
- if(minCompare.errorRate < threshold){
+ 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){
totalBases += (minCompare.total * minCompare.weight);
totalMatches += minCompare.matches * minCompare.weight;
if(minCompare.mismatches > maxMismatch){
numSeqs++;
}
+ index++;
+ if(index % 1000 == 0){ cout << index << endl; }
}
queryFile.close();
+ errorSummaryFile.close();
+ errorSeqFile.close();
+
+ if(qualFileName != "" && reportFileName != ""){
+ printErrorQuality(qScoreErrorMap);
+ printQualityFR(qualForwardMap, qualReverseMap);
+ }
- int total = 0;
-
+ printErrorFRFile(errorForward, errorReverse);
- string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".count";
+ 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["count"].push_back(errorCountFileName);
-
+ 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";
-
+ 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;
}
started = 1;
if(q[i] == 'A'){
- if(r[i] == 'A'){ errors.AA++; errors.matches++; }
- if(r[i] == 'T'){ errors.AT++; }
- if(r[i] == 'G'){ errors.AG++; }
- if(r[i] == 'C'){ errors.AC++; }
- if(r[i] == '-'){ errors.Ai++; }
+ 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++; }
- if(r[i] == 'T'){ errors.TT++; errors.matches++; }
- if(r[i] == 'G'){ errors.TG++; }
- if(r[i] == 'C'){ errors.TC++; }
- if(r[i] == '-'){ errors.Ti++; }
+ 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++; }
- if(r[i] == 'T'){ errors.GT++; }
- if(r[i] == 'G'){ errors.GG++; errors.matches++; }
- if(r[i] == 'C'){ errors.GC++; }
- if(r[i] == '-'){ errors.Gi++; }
+ 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++; }
- if(r[i] == 'T'){ errors.CT++; }
- if(r[i] == 'G'){ errors.CG++; }
- if(r[i] == 'C'){ errors.CC++; errors.matches++; }
- if(r[i] == '-'){ errors.Ci++; }
+ 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++; }
- if(r[i] == 'T'){ errors.NT++; }
- if(r[i] == 'G'){ errors.NG++; }
- if(r[i] == 'C'){ errors.NC++; }
- if(r[i] == '-'){ errors.Ni++; }
+ 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++; }
- if(r[i] == 'T'){ errors.dT++; }
- if(r[i] == 'G'){ errors.dG++; }
- if(r[i] == 'C'){ errors.dC++; }
+ 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++;
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");
+// 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
void SeqErrorCommand::printErrorHeader(){
try {
- errorFile << "query\treference\tweight\t";
- errorFile << "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";
- errorFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\n";
+ 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";
+ errorSummaryFile << "insertions\tdeletions\tsubstitutions\tambig\tmatches\tmismatches\ttotal\terror\tnumparents\n";
- errorFile << setprecision(6);
- errorFile.setf(ios::fixed);
+ errorSummaryFile << setprecision(6);
+ errorSummaryFile.setf(ios::fixed);
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "printErrorHeader");
//***************************************************************************************************************
-void SeqErrorCommand::printErrorData(Compare error){
+void SeqErrorCommand::printErrorData(Compare error, int numParentSeqs){
try {
- errorFile << error.queryName << '\t' << error.refName << '\t' << error.weight << '\t';
- errorFile << error.AA << '\t' << error.AT << '\t' << error.AG << '\t' << error.AC << '\t';
- errorFile << error.TA << '\t' << error.TT << '\t' << error.TG << '\t' << error.TC << '\t';
- errorFile << error.GA << '\t' << error.GT << '\t' << error.GG << '\t' << error.GC << '\t';
- errorFile << error.CA << '\t' << error.CT << '\t' << error.CG << '\t' << error.CC << '\t';
- errorFile << error.NA << '\t' << error.NT << '\t' << error.NG << '\t' << error.NC << '\t';
- errorFile << error.Ai << '\t' << error.Ti << '\t' << error.Gi << '\t' << error.Ci << '\t' << error.Ni << '\t' ;
- errorFile << error.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
-
- errorFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t'; //insertions
- errorFile << error.dA + error.dT + error.dG + error.dC << '\t'; //deletions
- errorFile << 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
- errorFile << error.NA + error.NT + error.NG + error.NC + error.Ni << '\t'; //ambiguities
- errorFile << error.matches << '\t' << error.mismatches << '\t' << error.total << '\t' << error.errorRate << endl;
+ 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.dA << '\t' << error.dT << '\t' << error.dG << '\t' << error.dC << '\t';
+
+ errorSummaryFile << error.Ai + error.Ti + error.Gi + error.Ci << '\t'; //insertions
+ errorSummaryFile << error.dA + error.dT + error.dG + error.dC << '\t'; //deletions
+ 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][g] += error.weight * error.GA;
+ 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;
+ substitutionMatrix[t][c] += error.weight * error.CT;
+ substitutionMatrix[t][gap] += error.weight * error.dT;
+ substitutionMatrix[t][n] += error.weight * error.NT;
+
+ substitutionMatrix[g][a] += error.weight * error.AG;
+ substitutionMatrix[g][t] += error.weight * error.TG;
+ substitutionMatrix[g][g] += error.weight * error.GG;
+ substitutionMatrix[g][c] += error.weight * error.CG;
+ substitutionMatrix[g][gap] += error.weight * error.dG;
+ substitutionMatrix[g][n] += error.weight * error.NG;
+
+ substitutionMatrix[c][a] += error.weight * error.AC;
+ substitutionMatrix[c][t] += error.weight * error.TC;
+ substitutionMatrix[c][g] += error.weight * error.GC;
+ substitutionMatrix[c][c] += error.weight * error.CC;
+ substitutionMatrix[c][gap] += error.weight * error.dC;
+ substitutionMatrix[c][n] += error.weight * error.NC;
+
+ substitutionMatrix[gap][a] += error.weight * error.Ai;
+ substitutionMatrix[gap][t] += error.weight * error.Ti;
+ substitutionMatrix[gap][g] += error.weight * error.Gi;
+ substitutionMatrix[gap][c] += error.weight * error.Ci;
+ substitutionMatrix[gap][n] += error.weight * error.Ni;
+ }
}
catch(exception& e) {
m->errorOut(e, "SeqErrorCommand", "printErrorData");
//***************************************************************************************************************
+void SeqErrorCommand::printSubMatrix(){
+ try {
+ string subMatrixFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.matrix";
+ ofstream subMatrixFile;
+ m->openOutputFile(subMatrixFileName, subMatrixFile);
+ outputNames.push_back(subMatrixFileName); outputTypes["error.matrix"].push_back(subMatrixFileName);
+ vector<string> bases(6);
+ bases[0] = "A";
+ bases[1] = "T";
+ bases[2] = "G";
+ bases[3] = "C";
+ bases[4] = "Gap";
+ bases[5] = "N";
+ vector<int> refSums(5,1);
+
+ for(int i=0;i<5;i++){
+ subMatrixFile << "\tr" << bases[i];
+
+ for(int j=0;j<6;j++){
+ refSums[i] += substitutionMatrix[i][j];
+ }
+ }
+ subMatrixFile << endl;
+
+ for(int i=0;i<6;i++){
+ subMatrixFile << 'q' << bases[i];
+ for(int j=0;j<5;j++){
+ subMatrixFile << '\t' << substitutionMatrix[j][i];
+ }
+ subMatrixFile << endl;
+ }
+
+ subMatrixFile << "total";
+ for(int i=0;i<5;i++){
+ subMatrixFile << '\t' << refSums[i];
+ }
+ subMatrixFile << endl;
+ subMatrixFile.close();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "printSubMatrix");
+ 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";
+ ofstream errorForwardFile;
+ m->openOutputFile(errorForwardFileName, errorForwardFile);
+ outputNames.push_back(errorForwardFileName); outputTypes["error.forward"].push_back(errorForwardFileName);
+
+ errorForwardFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
+ for(int i=0;i<1000;i++){
+ float match = (float)errorForward['m'][i];
+ float subst = (float)errorForward['s'][i];
+ float insert = (float)errorForward['i'][i];
+ float del = (float)errorForward['d'][i];
+ float amb = (float)errorForward['a'][i];
+ float total = match + subst + insert + del + amb;
+ if(total == 0){ break; }
+ errorForwardFile << i+1 << '\t' << total << '\t' << match/total << '\t' << subst/total << '\t' << insert/total << '\t' << del/total << '\t' << amb/total << endl;
+ }
+ errorForwardFile.close();
+
+ string errorReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.seq.reverse";
+ ofstream errorReverseFile;
+ m->openOutputFile(errorReverseFileName, errorReverseFile);
+ outputNames.push_back(errorReverseFileName); outputTypes["error.reverse"].push_back(errorReverseFileName);
+
+ errorReverseFile << "position\ttotalseqs\tmatch\tsubstitution\tinsertion\tdeletion\tambiguous" << endl;
+ for(int i=0;i<1000;i++){
+ float match = (float)errorReverse['m'][i];
+ float subst = (float)errorReverse['s'][i];
+ float insert = (float)errorReverse['i'][i];
+ float del = (float)errorReverse['d'][i];
+ float amb = (float)errorReverse['a'][i];
+ float total = match + subst + insert + del + amb;
+ if(total == 0){ break; }
+ errorReverseFile << i+1 << '\t' << total << '\t' << match/total << '\t' << subst/total << '\t' << insert/total << '\t' << del/total << '\t' << amb/total << endl;
+ }
+ errorReverseFile.close();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
+ exit(1);
+ }
+}
+
+//***************************************************************************************************************
+
+void SeqErrorCommand::printErrorQuality(map<char, vector<int> > qScoreErrorMap){
+ try{
+
+ string errorQualityFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.quality";
+ ofstream errorQualityFile;
+ m->openOutputFile(errorQualityFileName, errorQualityFile);
+ outputNames.push_back(errorQualityFileName); outputTypes["error.quality"].push_back(errorQualityFileName);
+
+ errorQualityFile << "qscore\tmatches\tsubstitutions\tinsertions\tambiguous" << endl;
+ for(int i=0;i<41;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");
+ exit(1);
+ }
+}
+
+
+//***************************************************************************************************************
+
+void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector<vector<int> > qualReverseMap){
+ try{
+ int lastRow = 0;
+ int lastColumn = 0;
+ for(int i=0;i<qualForwardMap.size();i++){
+ for(int j=0;j<qualForwardMap[i].size();j++){
+ if(qualForwardMap[i][j] != 0){
+ if(lastRow < i) { lastRow = i+2; }
+ if(lastColumn < j) { lastColumn = j+2; }
+ }
+ }
+ }
+ string qualityForwardFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.forward";
+ ofstream qualityForwardFile;
+ m->openOutputFile(qualityForwardFileName, qualityForwardFile);
+ outputNames.push_back(qualityForwardFileName); outputTypes["error.qual.forward"].push_back(qualityForwardFileName);
+ for(int i=0;i<lastColumn;i++){ qualityForwardFile << '\t' << i; } qualityForwardFile << endl;
+ for(int i=0;i<lastRow;i++){
+ qualityForwardFile << i+1;
+ for(int j=0;j<lastColumn;j++){
+ qualityForwardFile << '\t' << qualForwardMap[i][j];
+ }
+
+ qualityForwardFile << endl;
+ }
+ qualityForwardFile.close();
+
+
+ string qualityReverseFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.qual.reverse";
+ ofstream qualityReverseFile;
+ m->openOutputFile(qualityReverseFileName, qualityReverseFile);
+ outputNames.push_back(qualityReverseFileName); outputTypes["error.qual.reverse"].push_back(qualityReverseFileName);
+
+ for(int i=0;i<lastColumn;i++){ qualityReverseFile << '\t' << i; } qualityReverseFile << endl;
+ for(int i=0;i<lastRow;i++){
+
+ qualityReverseFile << i+1;
+ for(int j=0;j<lastColumn;j++){
+ qualityReverseFile << '\t' << qualReverseMap[i][j];
+ }
+ qualityReverseFile << endl;
+ }
+ qualityReverseFile.close();
+ }
+ catch(exception& e) {
+ m->errorOut(e, "SeqErrorCommand", "printErrorFRFile");
+ exit(1);
+ }
+}
+
+
+//***************************************************************************************************************
projects / mothur.git / blobdiff