*/
#include "maligner.h"
+#include "kmerdb.hpp"
+#include "blastdb.hpp"
/***********************************************************************/
-Maligner::Maligner(vector<Sequence*> temp, int num, int match, int misMatch, float div, int minCov) :
- db(temp), numWanted(num), matchScore(match), misMatchPenalty(misMatch), minDivR(div), minCoverage(minCov) {}
+Maligner::Maligner(vector<Sequence*> temp, int num, int match, int misMatch, float div, int ms, int minCov, string mode, Database* dataLeft, Database* dataRight) :
+ db(temp), numWanted(num), matchScore(match), misMatchPenalty(misMatch), minDivR(div), minSimilarity(ms), minCoverage(minCov), searchMethod(mode), databaseLeft(dataLeft), databaseRight(dataRight) { m = MothurOut::getInstance(); }
/***********************************************************************/
-string Maligner::getResults(Sequence* q) {
+string Maligner::getResults(Sequence* q, DeCalculator* decalc) {
try {
outputResults.clear();
query = new Sequence(q->getName(), q->getAligned());
string chimera;
-
- decalc = new DeCalculator();
- //find closest seqs to query in template - returns copies of seqs so trim does not destroy - remember to deallocate
- refSeqs = decalc->findClosest(query, db, numWanted);
-
- refSeqs = minCoverageFilter(refSeqs);
+ if (searchMethod == "distance") {
+ //find closest seqs to query in template - returns copies of seqs so trim does not destroy - remember to deallocate
+ refSeqs = decalc->findClosest(query, db, numWanted, indexes);
+ }else if (searchMethod == "blast") {
+ refSeqs = getBlastSeqs(query, numWanted); //fills indexes
+ }else if (searchMethod == "kmer") {
+ refSeqs = getKmerSeqs(query, numWanted); //fills indexes
+ }else { m->mothurOut("not valid search."); exit(1); } //should never get here
+ if (m->control_pressed) { return chimera; }
+
+ refSeqs = minCoverageFilter(refSeqs);
+
if (refSeqs.size() < 2) {
for (int i = 0; i < refSeqs.size(); i++) { delete refSeqs[i]; }
percentIdenticalQueryChimera = 0.0;
}
int chimeraPenalty = computeChimeraPenalty();
-
+
+ //fills outputResults
+ chimera = chimeraMaligner(chimeraPenalty, decalc);
+
+ if (m->control_pressed) { return chimera; }
+
+ //free memory
+ delete query;
+
+ for (int i = 0; i < refSeqs.size(); i++) { delete refSeqs[i]; }
+
+ return chimera;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Maligner", "getResults");
+ exit(1);
+ }
+}
+/***********************************************************************/
+string Maligner::chimeraMaligner(int chimeraPenalty, DeCalculator* decalc) {
+ try {
+
+ string chimera;
+
//trims seqs to first non gap char in all seqs and last non gap char in all seqs
- decalc->trimSeqs(query, refSeqs);
+ spotMap = decalc->trimSeqs(query, refSeqs);
+ //you trimmed the whole sequence, skip
+ if (query->getAligned() == "") { return "no"; }
+
vector<Sequence*> temp = refSeqs;
temp.push_back(query);
verticalFilter(temp);
+//for (int i = 0; i < temp.size(); i++) { cout << temp[i]->getName() << '\n' << temp[i]->getAligned() << endl; } return "no";
vector< vector<score_struct> > matrix = buildScoreMatrix(query->getAligned().length(), refSeqs.size()); //builds and initializes
- fillScoreMatrix(matrix, refSeqs, chimeraPenalty);
+ if (m->control_pressed) { return chimera; }
+ fillScoreMatrix(matrix, refSeqs, chimeraPenalty);
+
vector<score_struct> path = extractHighestPath(matrix);
- vector<trace_struct> trace = mapTraceRegionsToAlignment(path, refSeqs);
+ if (m->control_pressed) { return chimera; }
+ vector<trace_struct> trace = mapTraceRegionsToAlignment(path, refSeqs);
+
if (trace.size() > 1) { chimera = "yes"; }
else { chimera = "no"; }
int traceStart = path[0].col;
- int traceEnd = path[path.size()-1].col;
-
+ int traceEnd = path[path.size()-1].col;
string queryInRange = query->getAligned();
queryInRange = queryInRange.substr(traceStart, (traceEnd-traceStart+1));
string chimeraSeq = constructChimericSeq(trace, refSeqs);
-
+
percentIdenticalQueryChimera = computePercentID(queryInRange, chimeraSeq);
- delete decalc;
+ if (m->control_pressed) { return chimera; }
//save output results
for (int i = 0; i < trace.size(); i++) {
results temp;
temp.parent = refSeqs[seqIndex]->getName();
+ temp.parentAligned = db[indexes[seqIndex]]->getAligned();
+ temp.nastRegionStart = spotMap[regionStart];
+ temp.nastRegionEnd = spotMap[regionEnd];
temp.regionStart = regionStart;
temp.regionEnd = regionEnd;
outputResults.push_back(temp);
}
-
- //free memory
- delete query;
- for (int i = 0; i < refSeqs.size(); i++) { delete refSeqs[i]; }
-
+
return chimera;
}
catch(exception& e) {
- errorOut(e, "Maligner", "getResults");
+ m->errorOut(e, "Maligner", "chimeraMaligner");
exit(1);
}
}
//if coverage above minimum
if (coverage > minCoverage) {
newRefs.push_back(ref[i]);
+ }else {
+ delete ref[i];
}
}
return newRefs;
}
catch(exception& e) {
- errorOut(e, "Maligner", "minCoverageFilter");
+ m->errorOut(e, "Maligner", "minCoverageFilter");
exit(1);
}
}
}
catch(exception& e) {
- errorOut(e, "Maligner", "computeChimeraPenalty");
+ m->errorOut(e, "Maligner", "computeChimeraPenalty");
exit(1);
}
}
if(gaps[i] == seqs.size()) { filterString[i] = '0'; numColRemoved++; }
}
+ map<int, int> newMap;
//for each sequence
for (int i = 0; i < seqs.size(); i++) {
string seqAligned = seqs[i]->getAligned();
string newAligned = "";
+ int count = 0;
for (int j = 0; j < seqAligned.length(); j++) {
//if this spot is not a gap
- if (filterString[j] == '1') { newAligned += seqAligned[j]; }
+ if (filterString[j] == '1') {
+ newAligned += seqAligned[j];
+ newMap[count] = spotMap[j];
+ count++;
+ }
}
seqs[i]->setAligned(newAligned);
}
-
+ spotMap = newMap;
}
catch(exception& e) {
- errorOut(e, "Maligner", "verticalFilter");
+ m->errorOut(e, "Maligner", "verticalFilter");
exit(1);
}
}
return m;
}
catch(exception& e) {
- errorOut(e, "Maligner", "buildScoreMatrix");
+ m->errorOut(e, "Maligner", "buildScoreMatrix");
exit(1);
}
}
//***************************************************************************************************************
-void Maligner::fillScoreMatrix(vector<vector<score_struct> >& m, vector<Sequence*> seqs, int penalty) {
+void Maligner::fillScoreMatrix(vector<vector<score_struct> >& ms, vector<Sequence*> seqs, int penalty) {
try{
//get matrix dimensions
//are you both gaps?
if ((!isalpha(queryAligned[0])) && (!isalpha(subjectAligned[0]))) {
- m[i][0].score = 0;
+ ms[i][0].score = 0;
}else if (queryAligned[0] == subjectAligned[0]) {
- m[i][0].score = matchScore;
+ ms[i][0].score = matchScore;
}else{
- m[i][0].score = 0;
+ ms[i][0].score = 0;
}
}
//compute score based on previous columns scores
for (int prevIndex = 0; prevIndex < numRows; prevIndex++) { //iterate through rows
- int sumScore = matchMisMatchScore + m[prevIndex][j-1].score;
+ int sumScore = matchMisMatchScore + ms[prevIndex][j-1].score;
//you are not at yourself
if (prevIndex != i) { sumScore += penalty; }
if (sumScore < 0) { sumScore = 0; }
- if (sumScore > m[i][j].score) {
- m[i][j].score = sumScore;
- m[i][j].prev = prevIndex;
+ if (sumScore > ms[i][j].score) {
+ ms[i][j].score = sumScore;
+ ms[i][j].prev = prevIndex;
}
}
}
}
catch(exception& e) {
- errorOut(e, "Maligner", "fillScoreMatrix");
+ m->errorOut(e, "Maligner", "fillScoreMatrix");
exit(1);
}
}
//***************************************************************************************************************
-vector<score_struct> Maligner::extractHighestPath(vector<vector<score_struct> > m) {
+vector<score_struct> Maligner::extractHighestPath(vector<vector<score_struct> > ms) {
try {
//get matrix dimensions
int numCols = query->getAligned().length();
- int numRows = m.size();
+ int numRows = ms.size();
//find highest score scoring matrix
for (int i = 0; i < numRows; i++) {
for (int j = 0; j < numCols; j++) {
- if (m[i][j].score > highestScore) {
- highestScore = m[i][j].score;
- highestStruct = m[i][j];
+ if (ms[i][j].score > highestScore) {
+ highestScore = ms[i][j].score;
+ highestStruct = ms[i][j];
}
}
}
int score = highestStruct.score;
while (pos >= 0 && score > 0) {
- score_struct temp = m[rowIndex][pos];
+ score_struct temp = ms[rowIndex][pos];
score = temp.score;
if (score > 0) { path.push_back(temp); }
}
catch(exception& e) {
- errorOut(e, "Maligner", "extractHighestPath");
+ m->errorOut(e, "Maligner", "extractHighestPath");
exit(1);
}
}
}
catch(exception& e) {
- errorOut(e, "Maligner", "mapTraceRegionsToAlignment");
+ m->errorOut(e, "Maligner", "mapTraceRegionsToAlignment");
exit(1);
}
}
return chimera;
}
catch(exception& e) {
- errorOut(e, "Maligner", "constructChimericSeq");
+ m->errorOut(e, "Maligner", "constructChimericSeq");
exit(1);
}
}
try {
if (queryAlign.length() != chimera.length()) {
- mothurOut("Error, alignment strings are of different lengths: "); mothurOutEndLine();
- mothurOut(toString(queryAlign.length())); mothurOutEndLine(); mothurOutEndLine(); mothurOutEndLine(); mothurOutEndLine();
- mothurOut(toString(chimera.length())); mothurOutEndLine();
+ m->mothurOut("Error, alignment strings are of different lengths: "); m->mothurOutEndLine();
+ m->mothurOut(toString(queryAlign.length())); m->mothurOutEndLine(); m->mothurOutEndLine(); m->mothurOutEndLine(); m->mothurOutEndLine();
+ m->mothurOut(toString(chimera.length())); m->mothurOutEndLine();
return -1.0;
}
}
catch(exception& e) {
- errorOut(e, "Maligner", "computePercentID");
+ m->errorOut(e, "Maligner", "computePercentID");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+vector<Sequence*> Maligner::getBlastSeqs(Sequence* q, int num) {
+ try {
+ indexes.clear();
+ vector<Sequence*> refResults;
+
+ //get parts of query
+ string queryUnAligned = q->getUnaligned();
+ string leftQuery = queryUnAligned.substr(0, int(queryUnAligned.length() * 0.33)); //first 1/3 of the sequence
+ string rightQuery = queryUnAligned.substr(int(queryUnAligned.length() * 0.66)); //last 1/3 of the sequence
+
+ Sequence* queryLeft = new Sequence(q->getName()+"left", leftQuery);
+ Sequence* queryRight = new Sequence(q->getName()+"right", rightQuery);
+
+ vector<int> tempIndexesRight = databaseLeft->findClosestMegaBlast(queryRight, num+1);
+ vector<int> tempIndexesLeft = databaseLeft->findClosestMegaBlast(queryLeft, num+1);
+
+ //if ((tempIndexesRight.size() == 0) && (tempIndexesLeft.size() == 0)) { m->mothurOut("megablast returned " + toString(tempIndexesRight.size()) + " results for the right end, and " + toString(tempIndexesLeft.size()) + " for the left end. Needed " + toString(num+1) + ". Unable to process sequence " + q->getName()); m->mothurOutEndLine(); return refResults; }
+
+ vector<int> smaller;
+ vector<int> larger;
+
+ if (tempIndexesRight.size() < tempIndexesLeft.size()) { smaller = tempIndexesRight; larger = tempIndexesLeft; }
+ else { smaller = tempIndexesLeft; larger = tempIndexesRight; }
+
+ //merge results
+ map<int, int> seen;
+ map<int, int>::iterator it;
+
+ vector<int> mergedResults;
+ for (int i = 0; i < smaller.size(); i++) {
+ //add left if you havent already
+ it = seen.find(smaller[i]);
+ if (it == seen.end()) {
+ mergedResults.push_back(smaller[i]);
+ seen[smaller[i]] = smaller[i];
+ }
+
+ //add right if you havent already
+ it = seen.find(larger[i]);
+ if (it == seen.end()) {
+ mergedResults.push_back(larger[i]);
+ seen[larger[i]] = larger[i];
+ }
+ }
+
+ for (int i = smaller.size(); i < larger.size(); i++) {
+ it = seen.find(larger[i]);
+ if (it == seen.end()) {
+ mergedResults.push_back(larger[i]);
+ seen[larger[i]] = larger[i];
+ }
+ }
+
+ if (mergedResults.size() < numWanted) { numWanted = mergedResults.size(); }
+//cout << q->getName() << " merged results size = " << mergedResults.size() << '\t' << "numwanted = " << numWanted << endl;
+ for (int i = 0; i < numWanted; i++) {
+//cout << db[mergedResults[i]]->getName() << '\t' << mergedResults[i] << endl;
+ if (db[mergedResults[i]]->getName() != q->getName()) {
+ Sequence* temp = new Sequence(db[mergedResults[i]]->getName(), db[mergedResults[i]]->getAligned());
+ refResults.push_back(temp);
+ indexes.push_back(mergedResults[i]);
+ }
+//cout << mergedResults[i] << endl;
+ }
+//cout << "done " << q->getName() << endl;
+ delete queryRight;
+ delete queryLeft;
+
+ return refResults;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Maligner", "getBlastSeqs");
+ exit(1);
+ }
+}
+//***************************************************************************************************************
+vector<Sequence*> Maligner::getKmerSeqs(Sequence* q, int num) {
+ try {
+ indexes.clear();
+
+ //get parts of query
+ string queryUnAligned = q->getUnaligned();
+ string leftQuery = queryUnAligned.substr(0, int(queryUnAligned.length() * 0.33)); //first 1/3 of the sequence
+ string rightQuery = queryUnAligned.substr(int(queryUnAligned.length() * 0.66)); //last 1/3 of the sequence
+
+ Sequence* queryLeft = new Sequence(q->getName(), leftQuery);
+ Sequence* queryRight = new Sequence(q->getName(), rightQuery);
+
+ vector<int> tempIndexesLeft = databaseLeft->findClosestSequences(queryLeft, numWanted);
+ vector<int> tempIndexesRight = databaseRight->findClosestSequences(queryRight, numWanted);
+
+ //merge results
+ map<int, int> seen;
+ map<int, int>::iterator it;
+
+ vector<int> mergedResults;
+ for (int i = 0; i < tempIndexesLeft.size(); i++) {
+ //add left if you havent already
+ it = seen.find(tempIndexesLeft[i]);
+ if (it == seen.end()) {
+ mergedResults.push_back(tempIndexesLeft[i]);
+ seen[tempIndexesLeft[i]] = tempIndexesLeft[i];
+ }
+
+ //add right if you havent already
+ it = seen.find(tempIndexesRight[i]);
+ if (it == seen.end()) {
+ mergedResults.push_back(tempIndexesRight[i]);
+ seen[tempIndexesRight[i]] = tempIndexesRight[i];
+ }
+ }
+
+//cout << q->getName() << endl;
+ vector<Sequence*> refResults;
+ for (int i = 0; i < numWanted; i++) {
+//cout << db[mergedResults[i]]->getName() << endl;
+ Sequence* temp = new Sequence(db[mergedResults[i]]->getName(), db[mergedResults[i]]->getAligned());
+ refResults.push_back(temp);
+ indexes.push_back(mergedResults[i]);
+ }
+//cout << endl;
+ delete queryRight;
+ delete queryLeft;
+
+ return refResults;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "Maligner", "getKmerSeqs");
exit(1);
}
}
projects / mothur.git / blobdiff