X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=decalc.cpp;h=98545f8e24a72dca332bf7571fc0e5e9e6936988;hp=9c84429fe3cc00df92fff4e2ae358ae8553692a4;hb=cf9987b67aa49777a4c91c2d21f96e58bf17aa82;hpb=8af930b692ff87d1b7031fdd5c9a0b88cdd61245

diff --git a/decalc.cpp b/decalc.cpp
index 9c84429..98545f8 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) { 
@@ -599,7 +599,7 @@ cout << largest->second << '\t' << largest->first->score << '\t' << largest->fir
 	}
 }
 
-//***************************************************************************************************************
+***************************************************************************************************************
 //used by removeObviousOutliers which was attempt to increase sensitivity of chimera detection...not currently used...
 int DeCalculator::findLargestContrib(vector<int> seen) {
 	try{
@@ -624,7 +624,7 @@ int DeCalculator::findLargestContrib(vector<int> seen) {
 		exit(1);
 	}
 }
-//***************************************************************************************************************
+***************************************************************************************************************
 void DeCalculator::removeContrib(int bad, vector<quanMember>& quan) {
 	try{
 	
@@ -683,22 +683,23 @@ float DeCalculator::getCoef(vector<float> obs, vector<float> qav) {
 }
 //***************************************************************************************************************
 //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;  
 		
-		vector<Sequence*> seqsMatches;  
 		vector<SeqDist> distsLeft;
 		vector<SeqDist> distsRight;
 		
 		Dist* distcalculator = new eachGapDist();
 		
-		string queryUnAligned = querySeq->getUnaligned();
+		string queryUnAligned = querySeq.getUnaligned();
 		int numBases = int(queryUnAligned.length() * 0.33);
 		
 		string leftQuery = ""; //first 1/3 of the sequence
 		string rightQuery = ""; //last 1/3 of the sequence
-		string queryAligned = querySeq->getAligned();
+		string queryAligned = querySeq.getAligned();
 		
 		//left side
 		bool foundFirstBase = false;
@@ -725,7 +726,7 @@ vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*
 			//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
@@ -737,36 +738,37 @@ vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*
 				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);
 		
-		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 = db[j];
+			subjectLeft.seq = NULL;
 			subjectLeft.dist = distLeft;
 			subjectLeft.index = j;
 			
 			distsLeft.push_back(subjectLeft);
 			
 			SeqDist subjectRight;
-			subjectRight.seq = db[j];
+			subjectRight.seq = NULL;
 			subjectRight.dist = distRight;
 			subjectRight.index = j;
 			
@@ -779,6 +781,7 @@ vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*
 		//sort by smallest distance
 		sort(distsRight.begin(), distsRight.end(), compareSeqDist);
 		sort(distsLeft.begin(), distsLeft.end(), compareSeqDist);
+
 		
 		//merge results		
 		map<string, string> seen;
@@ -787,51 +790,63 @@ vector<Sequence*> DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*
 		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(distsLeft[i].seq->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[distsLeft[i].seq->getName()] = distsLeft[i].seq->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(distsRight[i].seq->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[distsRight[i].seq->getName()] = distsRight[i].seq->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(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(dists[i].seq->getName(), dists[i].seq->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;
@@ -874,17 +889,17 @@ Sequence* DeCalculator::findClosest(Sequence* querySeq, vector<Sequence*> db) {
 	}
 }
 /***************************************************************************************************************/
-map<int, int> DeCalculator::trimSeqs(Sequence* query, vector<Sequence*> topMatches) {
+map<int, int> DeCalculator::trimSeqs(Sequence& query, vector<Sequence>& topMatches) {
 	try {
 		
 		int frontPos = 0;  //should contain first position in all seqs that is not a gap character
-		int rearPos = query->getAligned().length();
+		int rearPos = query.getAligned().length();
 		
 		//********find first position in topMatches that is a non gap character***********//
 		//find first position all query seqs that is a non gap character
 		for (int i = 0; i < topMatches.size(); i++) {
 			
-			string aligned = topMatches[i]->getAligned();
+			string aligned = topMatches[i].getAligned();
 			int pos = 0;
 			
 			//find first spot in this seq
@@ -900,7 +915,7 @@ map<int, int> DeCalculator::trimSeqs(Sequence* query, vector<Sequence*> topMatch
 		}
 		
 		
-		string aligned = query->getAligned();
+		string aligned = query.getAligned();
 		int pos = 0;
 			
 		//find first position in query that is a non gap character
@@ -918,7 +933,7 @@ map<int, int> DeCalculator::trimSeqs(Sequence* query, vector<Sequence*> topMatch
 		//********find last position in topMatches that is a non gap character***********//
 		for (int i = 0; i < topMatches.size(); i++) {
 			
-			string aligned = topMatches[i]->getAligned();
+			string aligned = topMatches[i].getAligned();
 			int pos = aligned.length();
 			
 			//find first spot in this seq
@@ -934,7 +949,7 @@ map<int, int> DeCalculator::trimSeqs(Sequence* query, vector<Sequence*> topMatch
 		}
 		
 		
-		aligned = query->getAligned();
+		aligned = query.getAligned();
 		pos = aligned.length();
 		
 		//find last position in query that is a non gap character
@@ -951,25 +966,24 @@ map<int, int> DeCalculator::trimSeqs(Sequence* query, vector<Sequence*> topMatch
 		map<int, int> trimmedPos;
 		//check to make sure that is not whole seq
 		if ((rearPos - frontPos - 1) <= 0) {  
-			m->mothurOut("[ERROR]: when I trim your sequences, the entire sequence is trimmed."); m->mothurOutEndLine();  
-			query->setAligned("");
+			query.setAligned("");
 			//trim topMatches
 			for (int i = 0; i < topMatches.size(); i++) {
-				topMatches[i]->setAligned("");
+				topMatches[i].setAligned("");
 			}
 			
 		}else {
 
 			//trim query
-			string newAligned = query->getAligned();
+			string newAligned = query.getAligned();
 			newAligned = newAligned.substr(frontPos, (rearPos-frontPos+1));
-			query->setAligned(newAligned);
+			query.setAligned(newAligned);
 			
 			//trim topMatches
 			for (int i = 0; i < topMatches.size(); i++) {
-				newAligned = topMatches[i]->getAligned();
+				newAligned = topMatches[i].getAligned();
 				newAligned = newAligned.substr(frontPos, (rearPos-frontPos+1));
-				topMatches[i]->setAligned(newAligned);
+				topMatches[i].setAligned(newAligned);
 			}
 			
 			for (int i = 0; i < newAligned.length(); i++) {