createFilter(templateSeqs, 0.0); //just removed columns where all seqs have a gap
- //run filter on template
- for (int i = 0; i < templateSeqs.size(); i++) { if (m->control_pressed) { break; } runFilter(templateSeqs[i]); }
-
-
+ if (searchMethod == "distance") {
+ createFilter(templateSeqs, 0.0); //just removed columns where all seqs have a gap
+
+ //run filter on template copying templateSeqs into filteredTemplateSeqs
+ for (int i = 0; i < templateSeqs.size(); i++) {
+ if (m->control_pressed) { break; }
+
+ Sequence* newSeq = new Sequence(templateSeqs[i]->getName(), templateSeqs[i]->getAligned());
+ runFilter(newSeq);
+ filteredTemplateSeqs.push_back(newSeq);
+ }
+ }
}
catch(exception& e) {
m->errorOut(e, "ChimeraSlayer", "ChimeraSlayer");
if (m->control_pressed) { return 0; }
Sequence* newSeq = new Sequence(templateSeqs[i]->getName(), templateSeqs[i]->getAligned());
- filteredTemplateSeqs.push_back(newSeq);
runFilter(newSeq);
+ filteredTemplateSeqs.push_back(newSeq);
}
}
string kmerDBNameLeft;
}else if (searchMethod == "blast") {
//generate blastdb
- databaseLeft = new BlastDB(-1.0, -1.0, 1, -3);
+ databaseLeft = new BlastDB(m->getRootName(m->getSimpleName(fastafile)), -1.0, -1.0, 1, -3);
for (int i = 0; i < templateSeqs.size(); i++) { databaseLeft->addSequence(*templateSeqs[i]); }
databaseLeft->generateDB();
}else if (searchMethod == "blast") {
//generate blastdb
- databaseLeft = new BlastDB(-1.0, -1.0, 1, -3);
+ databaseLeft = new BlastDB(m->getRootName(m->getSimpleName(templateFileName)), -1.0, -1.0, 1, -3);
for (int i = 0; i < userTemplate.size(); i++) { if (m->control_pressed) { return userTemplate; } databaseLeft->addSequence(*userTemplate[i]); }
databaseLeft->generateDB();
bool rightChimeric = false;
bool leftChimeric = false;
+
+ cout << endl;
if (chimeraFlag == "yes") {
//which peice is chimeric or are both
//moved this out of maligner - 4/29/11
vector<Sequence*> refSeqs = getRefSeqs(query, thisTemplate, thisFilteredTemplate);
-
+
Maligner maligner(refSeqs, match, misMatch, divR, minSim, minCov);
Slayer slayer(window, increment, minSim, divR, iters, minSNP, minBS);
if (m->control_pressed) { return 0; }
string chimeraFlag = maligner.getResults(query, decalc);
-
+
if (m->control_pressed) { return 0; }
vector<results> Results = maligner.getOutput();
for (int i = 0; i < refSeqs.size(); i++) { delete refSeqs[i]; }
if (chimeraFlag == "yes") {
-
+
if (realign) {
- ChimeraReAligner realigner(thisTemplate, match, misMatch);
- realigner.reAlign(query, Results);
+ vector<string> parents;
+ for (int i = 0; i < Results.size(); i++) {
+ parents.push_back(Results[i].parentAligned);
+ }
+
+ ChimeraReAligner realigner;
+ realigner.reAlign(query, parents);
+
}
//get sequence that were given from maligner results
for (int j = 0; j < Results.size(); j++) {
float dist = (Results[j].regionEnd - Results[j].regionStart + 1) * Results[j].queryToParentLocal;
//only add if you are not a duplicate
- itDup = removeDups.find(Results[j].parent);
- if (itDup == removeDups.end()) { //this is not duplicate
- removeDups[Results[j].parent] = dist;
- parentNameSeq[Results[j].parent] = Results[j].parentAligned;
- }else if (dist > itDup->second) { //is this a stronger number for this parent
- removeDups[Results[j].parent] = dist;
- parentNameSeq[Results[j].parent] = Results[j].parentAligned;
+
+ if(Results[j].queryToParentLocal >= 90){ //local match has to be over 90% similarity
+
+ itDup = removeDups.find(Results[j].parent);
+ if (itDup == removeDups.end()) { //this is not duplicate
+ removeDups[Results[j].parent] = dist;
+ parentNameSeq[Results[j].parent] = Results[j].parentAligned;
+ }else if (dist > itDup->second) { //is this a stronger number for this parent
+ removeDups[Results[j].parent] = dist;
+ parentNameSeq[Results[j].parent] = Results[j].parentAligned;
+ }
+
}
+
}
for (itDup = removeDups.begin(); itDup != removeDups.end(); itDup++) {
SeqDist member;
member.seq = seq;
member.dist = itDup->second;
-
seqs.push_back(member);
}
seqs.pop_back();
}
}
-
+
//put seqs into vector to send to slayer
+
+// cout << query->getAligned() << endl;
vector<Sequence*> seqsForSlayer;
- for (int k = 0; k < seqs.size(); k++) { seqsForSlayer.push_back(seqs[k].seq); }
+ for (int k = 0; k < seqs.size(); k++) {
+// cout << seqs[k].seq->getAligned() << endl;
+ seqsForSlayer.push_back(seqs[k].seq);
+
+ }
if (m->control_pressed) { for (int k = 0; k < seqs.size(); k++) { delete seqs[k].seq; } return 0; }
//free memory
for (int k = 0; k < seqs.size(); k++) { delete seqs[k].seq; }
}
-
+ //cout << endl << endl;
return 0;
}
catch(exception& e) {
//find closest seqs to query in template - returns copies of seqs so trim does not destroy - remember to deallocate
Sequence* newSeq = new Sequence(q->getName(), q->getAligned());
runFilter(newSeq);
- refSeqs = decalc->findClosest(newSeq, thisTemplate, thisFilteredTemplate, numWanted);
+ refSeqs = decalc->findClosest(newSeq, thisTemplate, thisFilteredTemplate, numWanted, minSim);
+ delete newSeq;
}else if (searchMethod == "blast") {
refSeqs = getBlastSeqs(q, thisTemplate, numWanted); //fills indexes
}else if (searchMethod == "kmer") {
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
+//cout << "whole length = " << queryUnAligned.length() << '\t' << "left length = " << leftQuery.length() << '\t' << "right length = "<< rightQuery.length() << endl;
+ Sequence* queryLeft = new Sequence(q->getName(), leftQuery);
+ Sequence* queryRight = new Sequence(q->getName(), rightQuery);
- Sequence* queryLeft = new Sequence(q->getName()+"left", leftQuery);
- Sequence* queryRight = new Sequence(q->getName()+"right", rightQuery);
-
- vector<int> tempIndexesLeft = databaseLeft->findClosestMegaBlast(queryLeft, num+1);
- vector<int> tempIndexesRight = databaseLeft->findClosestMegaBlast(queryRight, num+1);
-
- vector<int> smaller;
- vector<int> larger;
+ vector<int> tempIndexesLeft = databaseLeft->findClosestMegaBlast(queryLeft, num+1, minSim);
+ vector<int> tempIndexesRight = databaseLeft->findClosestMegaBlast(queryRight, num+1, minSim);
+
- if (tempIndexesRight.size() < tempIndexesLeft.size()) { smaller = tempIndexesRight; larger = tempIndexesLeft; }
- else { smaller = tempIndexesLeft; larger = tempIndexesRight; }
+ //cout << q->getName() << '\t' << leftQuery << '\t' << "leftMatches = " << tempIndexesLeft.size() << '\t' << rightQuery << " rightMatches = " << tempIndexesRight.size() << endl;
+// 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++) {
+
+ int index = 0;
+// for (int i = 0; i < smaller.size(); i++) {
+ while(index < tempIndexesLeft.size() && index < tempIndexesRight.size()){
+
+ if (m->control_pressed) { delete queryRight; delete queryLeft; return refResults; }
+
//add left if you havent already
- it = seen.find(smaller[i]);
+ it = seen.find(tempIndexesLeft[index]);
if (it == seen.end()) {
- mergedResults.push_back(smaller[i]);
- seen[smaller[i]] = smaller[i];
+ mergedResults.push_back(tempIndexesLeft[index]);
+ seen[tempIndexesLeft[index]] = tempIndexesLeft[index];
}
//add right if you havent already
- it = seen.find(larger[i]);
+ it = seen.find(tempIndexesRight[index]);
if (it == seen.end()) {
- mergedResults.push_back(larger[i]);
- seen[larger[i]] = larger[i];
+ mergedResults.push_back(tempIndexesRight[index]);
+ seen[tempIndexesRight[index]] = tempIndexesRight[index];
}
+ index++;
}
+
- for (int i = smaller.size(); i < larger.size(); i++) {
+ for (int i = index; i < tempIndexesLeft.size(); i++) {
+ if (m->control_pressed) { delete queryRight; delete queryLeft; return refResults; }
+
//add right if you havent already
- it = seen.find(larger[i]);
+ it = seen.find(tempIndexesLeft[i]);
if (it == seen.end()) {
- mergedResults.push_back(larger[i]);
- seen[larger[i]] = larger[i];
+ mergedResults.push_back(tempIndexesLeft[i]);
+ seen[tempIndexesLeft[i]] = tempIndexesLeft[i];
}
}
- //numWanted = mergedResults.size();
- //cout << q->getName() << " merged results size = " << mergedResults.size() << '\t' << "numwanted = " << numWanted << endl;
- for (int i = 0; i < mergedResults.size(); i++) {
- //cout << db[mergedResults[i]]->getName() << '\t' << mergedResults[i] << endl;
+ for (int i = index; i < tempIndexesRight.size(); i++) {
+ if (m->control_pressed) { delete queryRight; delete queryLeft; return refResults; }
+ //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];
+ }
+ }
+
+
+ for (int i = 0; i < mergedResults.size(); i++) {
+ //cout << mergedResults[i] << '\t' << db[mergedResults[i]]->getName() << endl;
if (db[mergedResults[i]]->getName() != q->getName()) {
Sequence* temp = new Sequence(db[mergedResults[i]]->getName(), db[mergedResults[i]]->getAligned());
refResults.push_back(temp);
- //cout << db[mergedResults[i]]->getName() << endl;
+
}
-
- //cout << mergedResults[i] << endl;
}
- //cout << "done " << q->getName() << endl;
+
+
delete queryRight;
delete queryLeft;
//***************************************************************************************************************
vector<Sequence*> ChimeraSlayer::getKmerSeqs(Sequence* q, vector<Sequence*>& db, int num) {
try {
+ vector<Sequence*> refResults;
//get parts of query
string queryUnAligned = q->getUnaligned();
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);
+ vector<int> tempIndexesLeft = databaseLeft->findClosestSequences(queryLeft, num);
+ vector<int> tempIndexesRight = databaseRight->findClosestSequences(queryRight, num);
//merge results
map<int, int> seen;
map<int, int>::iterator it;
vector<int> mergedResults;
for (int i = 0; i < tempIndexesLeft.size(); i++) {
+
+ if (m->control_pressed) { delete queryRight; delete queryLeft; return refResults; }
+
//add left if you havent already
it = seen.find(tempIndexesLeft[i]);
if (it == seen.end()) {
//numWanted = mergedResults.size();
//cout << q->getName() << endl;
- vector<Sequence*> refResults;
+
for (int i = 0; i < mergedResults.size(); i++) {
//cout << db[mergedResults[i]]->getName() << endl;
if (db[mergedResults[i]]->getName() != q->getName()) {
projects / mothur.git / blobdiff