#include "dist.h"
#include "eachgapdist.h"
#include "ignoregaps.h"
-
+#include "eachgapdist.h"
//***************************************************************************************************************
void DeCalculator::setMask(string ms) {
}
//***************************************************************************************************************
//gets closest matches to each end, since chimeras will most likely have different parents on each end
-vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*> db, int& numWanted, vector<int>& indexes) {
+vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*>& thisTemplate, vector<Sequence*>& thisFilteredTemplate, int numWanted, int minSim) {
try {
- indexes.clear();
+ //indexes.clear();
vector<Sequence*> seqsMatches;
//if you are a base
if (isalpha(queryAligned[i])) { baseCount++; }
//if you have 1/3
- if (baseCount >= numBases) { rightSpot = i; break; } //last 1/3
+ if (baseCount > numBases + 1) { rightSpot = i; break; } //last 1/3
}
//trim end
break;
}
}
- rightQuery = queryAligned.substr(rightSpot, (lastBaseSpot-rightSpot)); //sequence from pos spot to end
+ rightQuery = queryAligned.substr(rightSpot, (lastBaseSpot-rightSpot+1)); //sequence from pos spot to end
Sequence queryLeft(querySeq->getName(), 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;
//sort by smallest distance
sort(distsRight.begin(), distsRight.end(), compareSeqDist);
sort(distsLeft.begin(), distsLeft.end(), compareSeqDist);
-
+
+
//merge results
map<string, string> seen;
map<string, string>::iterator it;
vector<SeqDist> 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++;
- }
-
- if (numWanted > dists.size()) {
- //m->mothurOut("numwanted is larger than the number of template sequences, adjusting numwanted."); m->mothurOutEndLine();
- numWanted = dists.size();
- }
+ //cout << numWanted << endl;
+ for (int i = 0; i < dists.size(); i++) {
+// cout << db[dists[i].index]->getName() << '\t' << dists[i].dist << endl;
-//cout << numWanted << endl;
- for (int i = 0; i < numWanted; i++) {
-//cout << dists[i].seq->getName() << '\t' << dists[i].dist << endl;
+ 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);
+ }
- 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;