X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=decalc.cpp;h=3a8c9d5711fa56631a4cac6cac70658fc2ad27f9;hb=62c36830aae6dd6151898ec6e07df59c8aed79fe;hp=8d90a78a492e63ee97ede0b577e37fde39336fb4;hpb=f1e2f581f91d88e9d6c7a69e495b36031b29e4ab;p=mothur.git diff --git a/decalc.cpp b/decalc.cpp index 8d90a78..3a8c9d5 100644 --- a/decalc.cpp +++ b/decalc.cpp @@ -12,7 +12,7 @@ #include "dist.h" #include "eachgapdist.h" #include "ignoregaps.h" - +#include "eachgapdist.h" //*************************************************************************************************************** void DeCalculator::setMask(string ms) { @@ -683,9 +683,9 @@ float DeCalculator::getCoef(vector obs, vector qav) { } //*************************************************************************************************************** //gets closest matches to each end, since chimeras will most likely have different parents on each end -vector DeCalculator::findClosest(Sequence* querySeq, vector db, int& numWanted, vector& indexes) { +vector DeCalculator::findClosest(Sequence* querySeq, vector& thisTemplate, vector& thisFilteredTemplate, int numWanted, int minSim) { try { - indexes.clear(); + //indexes.clear(); vector seqsMatches; @@ -726,7 +726,7 @@ vector DeCalculator::findClosest(Sequence* querySeq, vector= numBases) { rightSpot = i; break; } //last 1/3 + if (baseCount > numBases + 1) { rightSpot = i; break; } //last 1/3 } //trim end @@ -738,27 +738,28 @@ vector DeCalculator::findClosest(Sequence* querySeq, vectorgetName(), leftQuery); Sequence queryRight(querySeq->getName(), rightQuery); + //cout << querySeq->getName() << '\t' << leftSpot << '\t' << rightSpot << '\t' << firstBaseSpot << '\t' << lastBaseSpot << endl; //cout << queryUnAligned.length() << '\t' << queryLeft.getUnaligned().length() << '\t' << queryRight.getUnaligned().length() << endl; - for(int j = 0; j < db.size(); j++){ + for(int j = 0; j < thisFilteredTemplate.size(); j++){ - string dbAligned = db[j]->getAligned(); + string dbAligned = thisFilteredTemplate[j]->getAligned(); string leftDB = dbAligned.substr(firstBaseSpot, (leftSpot-firstBaseSpot+1)); //first 1/3 of the sequence - string rightDB = dbAligned.substr(rightSpot, (lastBaseSpot-rightSpot)); //last 1/3 of the sequence - - Sequence dbLeft(db[j]->getName(), leftDB); - Sequence dbRight(db[j]->getName(), rightDB); + string rightDB = dbAligned.substr(rightSpot, (lastBaseSpot-rightSpot+1)); //last 1/3 of the sequence + + Sequence dbLeft(thisFilteredTemplate[j]->getName(), leftDB); + Sequence dbRight(thisFilteredTemplate[j]->getName(), rightDB); distcalculator->calcDist(queryLeft, dbLeft); float distLeft = distcalculator->getDist(); distcalculator->calcDist(queryRight, dbRight); float distRight = distcalculator->getDist(); - + SeqDist subjectLeft; subjectLeft.seq = NULL; subjectLeft.dist = distLeft; @@ -780,6 +781,7 @@ vector DeCalculator::findClosest(Sequence* querySeq, vector seen; @@ -788,51 +790,63 @@ vector DeCalculator::findClosest(Sequence* querySeq, vector dists; float lastRight = distsRight[0].dist; float lastLeft = distsLeft[0].dist; - int lasti = 0; - for (int i = 0; i < distsLeft.size(); i++) { + + float maxDist = 1.0 - (minSim / 100.0); + + for (int i = 0; i < numWanted+1; i++) { + if (m->control_pressed) { return seqsMatches; } + //add left if you havent already - it = seen.find(db[distsLeft[i].index]->getName()); - if (it == seen.end()) { + it = seen.find(thisTemplate[distsLeft[i].index]->getName()); + if (it == seen.end() && distsLeft[i].dist <= maxDist) { dists.push_back(distsLeft[i]); - seen[db[distsLeft[i].index]->getName()] = db[distsLeft[i].index]->getName(); + seen[thisTemplate[distsLeft[i].index]->getName()] = thisTemplate[distsLeft[i].index]->getName(); lastLeft = distsLeft[i].dist; +// cout << "loop-left\t" << db[distsLeft[i].index]->getName() << '\t' << distsLeft[i].dist << endl; } //add right if you havent already - it = seen.find(db[distsRight[i].index]->getName()); - if (it == seen.end()) { + it = seen.find(thisTemplate[distsRight[i].index]->getName()); + if (it == seen.end() && distsRight[i].dist <= maxDist) { dists.push_back(distsRight[i]); - seen[db[distsRight[i].index]->getName()] = db[distsRight[i].index]->getName(); + seen[thisTemplate[distsRight[i].index]->getName()] = thisTemplate[distsRight[i].index]->getName(); lastRight = distsRight[i].dist; +// cout << "loop-right\t" << db[distsRight[i].index]->getName() << '\t' << distsRight[i].dist << endl; } - if (dists.size() > numWanted) { lasti = i; break; } //you have enough results + if (i == numWanted) { break; } + } - //add in dups - lasti++; - int i = lasti; - while (i < distsLeft.size()) { - if (distsLeft[i].dist == lastLeft) { dists.push_back(distsLeft[i]); numWanted++; } - else { break; } - i++; + //are we still above the minimum similarity cutoff + if ((lastLeft >= minSim) || (lastRight >= minSim)) { + //add in ties from left + int i = numWanted; + while (i < distsLeft.size()) { + if (distsLeft[i].dist == lastLeft) { dists.push_back(distsLeft[i]); } + else { break; } + i++; + } + + //add in ties from right + i = numWanted; + while (i < distsRight.size()) { + if (distsRight[i].dist == lastRight) { dists.push_back(distsRight[i]); } + else { break; } + i++; + } } - i = lasti; - while (i < distsRight.size()) { - if (distsRight[i].dist == lastRight) { dists.push_back(distsRight[i]); numWanted++; } - else { break; } - i++; - } + //cout << numWanted << endl; + for (int i = 0; i < dists.size(); i++) { +// cout << db[dists[i].index]->getName() << '\t' << dists[i].dist << endl; - if (numWanted > dists.size()) { m->mothurOut("numwanted is larger than the number of template sequences, adjusting numwanted."); m->mothurOutEndLine(); numWanted = dists.size(); } + if ((thisTemplate[dists[i].index]->getName() != querySeq->getName()) && (((1.0-dists[i].dist)*100) >= minSim)) { + Sequence* temp = new Sequence(thisTemplate[dists[i].index]->getName(), thisTemplate[dists[i].index]->getAligned()); //have to make a copy so you can trim and filter without stepping on eachother. + //cout << querySeq->getName() << '\t' << thisTemplate[dists[i].index]->getName() << '\t' << dists[i].dist << endl; + seqsMatches.push_back(temp); + } -//cout << numWanted << endl; - for (int i = 0; i < numWanted; i++) { -//cout << dists[i].seq->getName() << '\t' << dists[i].dist << endl; - Sequence* temp = new Sequence(db[dists[i].index]->getName(), db[dists[i].index]->getAligned()); //have to make a copy so you can trim and filter without stepping on eachother. - seqsMatches.push_back(temp); - indexes.push_back(dists[i].index); } return seqsMatches; @@ -952,7 +966,6 @@ map DeCalculator::trimSeqs(Sequence* query, vector topMatch map trimmedPos; //check to make sure that is not whole seq if ((rearPos - frontPos - 1) <= 0) { - m->mothurOut("[ERROR]: when I trim " + query->getName() + ", the entire sequence is trimmed. Skipping."); m->mothurOutEndLine(); query->setAligned(""); //trim topMatches for (int i = 0; i < topMatches.size(); i++) {