X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=weighted.cpp;h=52d90c8ac4487d83acd1587b1d0f5c30b0c4cfe0;hb=4745a956b3116a719f52f341d2a2db84df4817da;hp=fa22716e397c4dddb1dd1ab29e2cd04d40b10311;hpb=fc7cf3aac8fd6106fd725b43baa8ab5ca6f836f8;p=mothur.git

diff --git a/weighted.cpp b/weighted.cpp
index fa22716..52d90c8 100644
--- a/weighted.cpp
+++ b/weighted.cpp
@@ -11,156 +11,286 @@
 
 /**************************************************************************************************/
 
-EstOutput Weighted::getValues(Tree* t) {
+EstOutput Weighted::getValues(Tree* t, int p, string o) {
     try {
 		globaldata = GlobalData::getInstance();
+		
+		data.clear(); //clear out old values
 		int numGroups;
 		vector<double> D;
+		processors = p;
+		outputDir = o;
 		
 		numGroups = globaldata->Groups.size();
+	
+		vector<double> sums = getBranchLengthSums(t);
+		
+		if (m->control_pressed) { return data; }
 		
 		//calculate number of comparisons i.e. with groups A,B,C = AB, AC, BC = 3;
-		int count = 0;
+		vector< vector<string> > namesOfGroupCombos;
 		for (int i=0; i<numGroups; i++) { 
-			for (int l = i+1; l < numGroups; l++) {	
+			for (int l = 0; l < i; l++) {	
 				//initialize weighted scores
-				WScore[globaldata->Groups[i]+globaldata->Groups[l]] = 0.0;
+				//WScore[globaldata->Groups[i]+globaldata->Groups[l]] = 0.0;
+				vector<string> groups; groups.push_back(globaldata->Groups[i]); groups.push_back(globaldata->Groups[l]);
+				namesOfGroupCombos.push_back(groups);
+			}
+		}
+		
+		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+			if(processors == 1){
+				data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), sums);
+			}else{
+				int numPairs = namesOfGroupCombos.size();
 				
-				D.push_back(0.0000); //initialize a spot in D for each combination
+				int numPairsPerProcessor = numPairs / processors;
 				
-				/********************************************************/
-				//calculate a D value for each group combo
-				for(int v=0;v<t->getNumLeaves();v++){
-					int index = v;
-					double sum = 0.0000;
-		
-					//while you aren't at root
-					while(t->tree[index].getParent() != -1){
-							
-						//if you have a BL
-						if(t->tree[index].getBranchLength() != -1){
-							sum += t->tree[index].getBranchLength();
-						}
-						index = t->tree[index].getParent();
-					}
-						
-					//get last breanch length added
-					if(t->tree[index].getBranchLength() != -1){
-						sum += t->tree[index].getBranchLength();
-					}
-						
-					//is this sum from a sequence which is in one of the users groups
-					if (inUsersGroups(t->tree[v].getGroup(), globaldata->Groups) == true) {
-						//is this sum from a sequence which is in this groupCombo
-						if ((t->tree[v].getGroup() == globaldata->Groups[i]) || (t->tree[v].getGroup() == globaldata->Groups[l])) {
-							sum /= (double)tmap->seqsPerGroup[t->tree[v].getGroup()];
-							D[count] += sum; 
-						}
+				for (int i = 0; i < processors; i++) {
+					int startPos = i * numPairsPerProcessor;
+					if(i == processors - 1){
+						numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
 					}
+					lines.push_back(linePair(startPos, numPairsPerProcessor));
 				}
-				/*********************************************************/
-				count++;
+
+				data = createProcesses(t, namesOfGroupCombos, sums);
+				
+				lines.clear();
 			}
-		}
+		#else
+			data = driver(t, namesOfGroupCombos, 0, namesOfGroupCombos.size(), sums);
+		#endif
 		
-		data.clear(); //clear out old values
+		return data;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Weighted", "getValues");
+		exit(1);
+	}
+}
+/**************************************************************************************************/
+
+EstOutput Weighted::createProcesses(Tree* t, vector< vector<string> > namesOfGroupCombos, vector<double>& sums) {
+	try {
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+		int process = 1;
+		int num = 0;
+		vector<int> processIDS;
+		
+		EstOutput results;
+		
+		//loop through and create all the processes you want
+		while (process != processors) {
+			int pid = fork();
+			
+			if (pid > 0) {
+				processIDS.push_back(pid);  //create map from line number to pid so you can append files in correct order later
+				process++;
+			}else if (pid == 0){
+	
+				EstOutput Myresults;
+				Myresults = driver(t, namesOfGroupCombos, lines[process].start, lines[process].num, sums);
+			
+				m->mothurOut("Merging results."); m->mothurOutEndLine();
+				
+				//pass numSeqs to parent
+				ofstream out;
+
+				string tempFile = outputDir + toString(getpid()) + ".weighted.results.temp";
+	
+				m->openOutputFile(tempFile, out);
+	
+				out << Myresults.size() << endl;
+				for (int i = 0; i < Myresults.size(); i++) {  out << Myresults[i] << '\t';  } out << endl;
+				out.close();
+				
+				exit(0);
+			}else { m->mothurOut("unable to spawn the necessary processes."); m->mothurOutEndLine(); exit(0); }
+		}
+	
+		results = driver(t, namesOfGroupCombos, lines[0].start, lines[0].num, sums);
+	
+		//force parent to wait until all the processes are done
+		for (int i=0;i<(processors-1);i++) { 
+			int temp = processIDS[i];
+			wait(&temp);
+		}
 	
+		if (m->control_pressed) { return results; }
+		
+		//get data created by processes
+		for (int i=0;i<(processors-1);i++) { 
+			ifstream in;
+			string s = outputDir + toString(processIDS[i]) + ".weighted.results.temp";
+			m->openInputFile(s, in);
+			
+			//get quantiles
+			while (!in.eof()) {
+				int num;
+				in >> num; m->gobble(in);
+				
+				if (m->control_pressed) { break; }
+
+				double w; 
+				for (int j = 0; j < num; j++) {
+					in >> w;
+					results.push_back(w);
+				}
+				m->gobble(in);
+			}
+			in.close();
+			remove(s.c_str());
+		}
+		
+		m->mothurOut("DONE."); m->mothurOutEndLine(); m->mothurOutEndLine();
+		
+		return results;
+#endif		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Weighted", "createProcesses");
+		exit(1);
+	}
+}
+/**************************************************************************************************/
+EstOutput Weighted::driver(Tree* t, vector< vector<string> > namesOfGroupCombos, int start, int num, vector<double>& sums) { 
+ try {
+		EstOutput results;
+		vector<double> D;
+		
+		int count = 0;
+		for (int h = start; h < (start+num); h++) {
+		
+			if (m->control_pressed) { return results; }
+		
+			//initialize weighted score
+			string groupA = namesOfGroupCombos[h][0]; 
+			string groupB = namesOfGroupCombos[h][1];
+			
+			WScore[groupA+groupB] = 0.0;
+			D.push_back(0.0000); //initialize a spot in D for each combination
+			
+			//adding the wieghted sums from group i
+			for (int j = 0; j < t->groupNodeInfo[groupA].size(); j++) { //the leaf nodes that have seqs from group i
+				map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupA][j]].pcount.find(groupA);
+				int numSeqsInGroupI = it->second;
+				
+				double weightedSum = ((numSeqsInGroupI * sums[t->groupNodeInfo[groupA][j]]) / (double)tmap->seqsPerGroup[groupA]);
+			
+				D[count] += weightedSum;
+			}
+			
+			//adding the wieghted sums from group l
+			for (int j = 0; j < t->groupNodeInfo[groupB].size(); j++) { //the leaf nodes that have seqs from group l
+				map<string, int>::iterator it = t->tree[t->groupNodeInfo[groupB][j]].pcount.find(groupB);
+				int numSeqsInGroupL = it->second;
+				
+				double weightedSum = ((numSeqsInGroupL * sums[t->groupNodeInfo[groupB][j]]) / (double)tmap->seqsPerGroup[groupB]);
+			
+				D[count] += weightedSum;
+			}
+			count++;
+		}
+		
+		//calculate u for the group comb 
+		int total = t->getNumNodes();
+		int twentyPercent = (total * 0.20);
+		
 		for(int i=0;i<t->getNumNodes();i++){
-			//calculate weighted score for each of the group comb i.e. with groups A,B,C = AB, AC, BC.
-			for (int b=0; b<numGroups; b++) { 
-				for (int l = b+1; l < numGroups; l++) {
-					//calculate a u value for each combo
-					double u;
-					//does this node have descendants from group b-1
-					it = t->tree[i].pcount.find(globaldata->Groups[b]);
-					//if it does u = # of its descendants with a certain group / total number in tree with a certain group
-					if (it != t->tree[i].pcount.end()) {
-						u = (double) t->tree[i].pcount[globaldata->Groups[b]] / (double) tmap->seqsPerGroup[globaldata->Groups[b]];
-					}else { u = 0.00; }
-		
-					//does this node have descendants from group l
-					it = t->tree[i].pcount.find(globaldata->Groups[l]);
-					//if it does subtract their percentage from u
-					if (it != t->tree[i].pcount.end()) {
-						u -= (double) t->tree[i].pcount[globaldata->Groups[l]] / (double) tmap->seqsPerGroup[globaldata->Groups[l]];
-					}
-						
-					u = abs(u) * t->tree[i].getBranchLength();
-					
-					//save groupcombs u value
-					WScore[globaldata->Groups[b]+globaldata->Groups[l]] += u;
-				/*********************************************************/
+			for (int h = start; h < (start+num); h++) {
+				
+				string groupA = namesOfGroupCombos[h][0]; 
+				string groupB = namesOfGroupCombos[h][1];
+
+				if (m->control_pressed) { return results; }
+				
+				double u;
+				//does this node have descendants from groupA
+				it = t->tree[i].pcount.find(groupA);
+				//if it does u = # of its descendants with a certain group / total number in tree with a certain group
+				if (it != t->tree[i].pcount.end()) {
+					u = (double) t->tree[i].pcount[groupA] / (double) tmap->seqsPerGroup[groupA];
+				}else { u = 0.00; }
+			
+							
+				//does this node have descendants from group l
+				it = t->tree[i].pcount.find(groupB);
+				//if it does subtract their percentage from u
+				if (it != t->tree[i].pcount.end()) {
+					u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 				}
+							
+				u = abs(u * t->tree[i].getBranchLength());
+						
+				//save groupcombs u value
+				WScore[(groupA+groupB)] += u;
 			}
+			//report progress
+			if((i) % twentyPercent == 0){	m->mothurOut("Percentage complete: " + toString(int((i / (float)total) * 100.0))); m->mothurOutEndLine();		}
 		}
-  
-		//calculate weighted score for each group combination
+		
+		m->mothurOut("Percentage complete: 100"); m->mothurOutEndLine();
+		
+		/********************************************************/
+		//calculate weighted score for the group combination
 		double UN;	
 		count = 0;
-		for (int i=0; i<numGroups; i++) { 
-			for (int l = i+1; l < numGroups; l++) {
-				UN = (WScore[globaldata->Groups[i]+globaldata->Groups[l]] / D[count]);
+		for (int h = start; h < (start+num); h++) {
+			UN = (WScore[namesOfGroupCombos[h][0]+namesOfGroupCombos[h][1]] / D[count]);
 		
-				if (isnan(UN) || isinf(UN)) { UN = 0; } 
-				data.push_back(UN);
-				count++;
-			}
+			if (isnan(UN) || isinf(UN)) { UN = 0; } 
+			results.push_back(UN);
+			count++;
 		}
-		return data;
+				
+		return results; 
 	}
 	catch(exception& e) {
-		cout << "Standard Error: " << e.what() << " has occurred in the Weighted class Function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
-		exit(1);
-	}
-	catch(...) {
-		cout << "An unknown error has occurred in the Weighted class function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		m->errorOut(e, "Weighted", "driver");
 		exit(1);
 	}
-
 }
-
 /**************************************************************************************************/
-EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) { 
+EstOutput Weighted::getValues(Tree* t, string groupA, string groupB, vector<double>& sums) { 
  try {
 		globaldata = GlobalData::getInstance();
 		
 		data.clear(); //clear out old values
 		
+		if (m->control_pressed) { return data; }
+		
 		//initialize weighted score
 		WScore[(groupA+groupB)] = 0.0;
-		float D = 0.0;
+		double D = 0.0;
 		
-						
-		/********************************************************/
-		//calculate a D value for the group combo
-		for(int v=0;v<t->getNumLeaves();v++){
-			int index = v;
-			double sum = 0.0000;
+		vector<string> groups; groups.push_back(groupA); groups.push_back(groupB);
 		
-			//while you aren't at root
-			while(t->tree[index].getParent() != -1){
-							
-				//if you have a BL
-				if(t->tree[index].getBranchLength() != -1){
-					sum += t->tree[index].getBranchLength();
-				}
-				index = t->tree[index].getParent();
-			}
-						
-			//get last breanch length added
-			if(t->tree[index].getBranchLength() != -1){
-				sum += t->tree[index].getBranchLength();
-			}
-						
-			if ((t->tree[v].getGroup() == groupA) || (t->tree[v].getGroup() == groupB)) {
-				sum /= (double)tmap->seqsPerGroup[t->tree[v].getGroup()];
-				D += sum; 
-			}
+		//adding the wieghted sums from group i
+		for (int j = 0; j < t->groupNodeInfo[groups[0]].size(); j++) { //the leaf nodes that have seqs from group i
+			map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[0]][j]].pcount.find(groups[0]);
+			int numSeqsInGroupI = it->second;
+			
+			double weightedSum = ((numSeqsInGroupI * sums[t->groupNodeInfo[groups[0]][j]]) / (double)tmap->seqsPerGroup[groups[0]]);
+		
+			D += weightedSum;
+		}
+		
+		//adding the wieghted sums from group l
+		for (int j = 0; j < t->groupNodeInfo[groups[1]].size(); j++) { //the leaf nodes that have seqs from group l
+			map<string, int>::iterator it = t->tree[t->groupNodeInfo[groups[1]][j]].pcount.find(groups[1]);
+			int numSeqsInGroupL = it->second;
+			
+			double weightedSum = ((numSeqsInGroupL * sums[t->groupNodeInfo[groups[1]][j]]) / (double)tmap->seqsPerGroup[groups[1]]);
+		
+			D += weightedSum;
 		}
-		/********************************************************/				
 		
 		//calculate u for the group comb 
 		for(int i=0;i<t->getNumNodes();i++){
+		
+			if (m->control_pressed) { return data; }
+			
 			double u;
 			//does this node have descendants from groupA
 			it = t->tree[i].pcount.find(groupA);
@@ -177,7 +307,7 @@ EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 				u -= (double) t->tree[i].pcount[groupB] / (double) tmap->seqsPerGroup[groupB];
 			}
 						
-			u = abs(u) * t->tree[i].getBranchLength();
+			u = abs(u * t->tree[i].getBranchLength());
 					
 			//save groupcombs u value
 			WScore[(groupA+groupB)] += u;
@@ -195,21 +325,49 @@ EstOutput Weighted::getValues(Tree* t, string groupA, string groupB) {
 		return data; 
 	}
 	catch(exception& e) {
-		cout << "Standard Error: " << e.what() << " has occurred in the Weighted class Function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+		m->errorOut(e, "Weighted", "getValues");
 		exit(1);
 	}
-	catch(...) {
-		cout << "An unknown error has occurred in the Weighted class function getValues. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+}
+/**************************************************************************************************/
+vector<double> Weighted::getBranchLengthSums(Tree* t) { 
+	try {
+		
+		vector<double> sums;
+		
+		for(int v=0;v<t->getNumLeaves();v++){
+			
+			if (m->control_pressed) { return sums; }
+				
+			int index = v;
+			double sum = 0.0000;
+	
+			//while you aren't at root
+			while(t->tree[index].getParent() != -1){
+						
+				//if you have a BL
+				if(t->tree[index].getBranchLength() != -1){
+					sum += abs(t->tree[index].getBranchLength());
+				}
+				index = t->tree[index].getParent();
+			}
+					
+			//get last breanch length added
+			if(t->tree[index].getBranchLength() != -1){
+				sum += abs(t->tree[index].getBranchLength());
+			}
+			
+			sums.push_back(sum);
+		}
+		
+		return sums;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Weighted", "getBranchLengthSums");
 		exit(1);
 	}
-
 }
-
-
-
-
-
-
+/**************************************************************************************************/