X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=subsample.cpp;h=261297df67cfc1c5a8933a72f8bba307defbef88;hb=0cefb55a2616975bd4a144fc345693695ffc9bb6;hp=e6dd845adf9b64bec6c932f4cf72de6d8b9fe909;hpb=e41b86a600fe30f5df9507d7e55027e7b8bd7dd6;p=mothur.git

diff --git a/subsample.cpp b/subsample.cpp
index e6dd845..261297d 100644
--- a/subsample.cpp
+++ b/subsample.cpp
@@ -7,7 +7,204 @@
 //
 
 #include "subsample.h"
+//**********************************************************************************************************************
+Tree* SubSample::getSample(Tree* T, TreeMap* tmap, TreeMap* newTmap, int size, map<string, string> originalNameMap) {
+    try {
+        Tree* newTree = NULL;
+        
+        map<string, vector<string> > newGroups;
+        vector<string> subsampledSeqs = getSample(tmap, size, newGroups);
+        
+        //remove seqs not in sample from treemap
+        for (map<string, vector<string> >::iterator it = newGroups.begin(); it != newGroups.end(); it++) {
+            for (int i = 0; i < (it->second).size(); i++) {
+                newTmap->addSeq((it->second)[i], it->first);
+            }
+        }
+        
+        newTree = new Tree(newTmap);
+        newTree->getCopy(T, originalNameMap);
+        
+        return newTree;
+    }
+    catch(exception& e) {
+        m->errorOut(e, "SubSample", "getSample-Tree");
+        exit(1);
+    }
+}
+/**********************************************************************************************************************
+Tree* SubSample::getSample(Tree* T, TreeMap* tmap, map<string, string> whole, int size) {
+    try {
+        Tree* newTree = NULL;
+        
+        vector<string> subsampledSeqs = getSample(tmap, size);
+        map<string, string> sampledNameMap = deconvolute(whole, subsampledSeqs); 
+        
+        //remove seqs not in sample from treemap
+        for (int i = 0; i < tmap->namesOfSeqs.size(); i++) {
+            //is that name in the subsample?
+            int count = 0;
+            for (int j = 0; j < subsampledSeqs.size(); j++) {
+                if (tmap->namesOfSeqs[i] == subsampledSeqs[j]) { break; } //found it
+                count++;
+            }
 
+            if (m->control_pressed) { return newTree; }
+            
+            //if you didnt find it, remove it 
+            if (count == subsampledSeqs.size()) { 
+                tmap->removeSeq(tmap->namesOfSeqs[i]);
+                i--; //need this because removeSeq removes name from namesOfSeqs
+            }
+        }
+        
+        //create new tree
+        int numUniques = sampledNameMap.size();
+        if (sampledNameMap.size() == 0) { numUniques = subsampledSeqs.size(); }
+        
+        newTree = new Tree(numUniques, tmap); //numNodes, treemap
+        newTree->getSubTree(T, subsampledSeqs, sampledNameMap);
+        
+        return newTree;
+    }
+    catch(exception& e) {
+        m->errorOut(e, "SubSample", "getSample-Tree");
+        exit(1);
+    }
+}*/
+//**********************************************************************************************************************
+//assumes whole maps dupName -> uniqueName
+map<string, string> SubSample::deconvolute(map<string, string> whole, vector<string>& wanted) {
+    try {
+        map<string, string> nameMap;
+        
+        //whole will be empty if user gave no name file, so we don't need to make a new one
+        if (whole.size() == 0) { return nameMap; }
+        
+        vector<string> newWanted;
+        for (int i = 0; i < wanted.size(); i++) {
+            
+            if (m->control_pressed) { break; }
+            
+            string dupName = wanted[i];
+            
+            map<string, string>::iterator itWhole = whole.find(dupName);
+            if (itWhole != whole.end()) {
+                string repName = itWhole->second;
+                
+                //do we already have this rep?
+                map<string, string>::iterator itName = nameMap.find(repName);
+                if (itName != nameMap.end()) { //add this seqs to dups list
+                    (itName->second) += "," + dupName;
+                }else { //first sighting of this seq
+                    nameMap[repName] = dupName;
+                    newWanted.push_back(repName);
+                }
+            }else { m->mothurOut("[ERROR]: "+dupName+" is not in your name file, please correct.\n"); m->control_pressed = true; }
+        }
+        
+        wanted = newWanted;
+        return nameMap;
+    }
+	catch(exception& e) {
+		m->errorOut(e, "SubSample", "deconvolute");
+		exit(1);
+	}
+}
+//**********************************************************************************************************************
+vector<string> SubSample::getSample(TreeMap* tMap, int size, map<string, vector<string> >& sample) {
+    try {
+        vector<string> temp2;
+        sample["doNotIncludeMe"] = temp2;
+        
+        vector<string> namesInSample;
+        
+        vector<string> Groups = tMap->getNamesOfGroups();    
+        for (int i = 0; i < Groups.size(); i++) {
+            
+            if (m->inUsersGroups(Groups[i], m->getGroups())) {
+                if (m->control_pressed) { break; }
+                
+                vector<string> thisGroup; thisGroup.push_back(Groups[i]);
+                vector<string> thisGroupsSeqs = tMap->getNamesSeqs(thisGroup);
+                int thisSize = thisGroupsSeqs.size();
+                vector<string> temp;
+                sample[Groups[i]] = temp;
+                
+                if (thisSize >= size) {	
+                    
+                    random_shuffle(thisGroupsSeqs.begin(), thisGroupsSeqs.end());
+                    
+                    for (int j = 0; j < size; j++) { sample[Groups[i]].push_back(thisGroupsSeqs[j]); namesInSample.push_back(thisGroupsSeqs[j]); }
+                    for (int j = size; j < thisSize; j++) { sample["doNotIncludeMe"].push_back(thisGroupsSeqs[j]); }
+                    
+                }else {  m->mothurOut("[ERROR]: You have selected a size that is larger than "+Groups[i]+" number of sequences.\n"); m->control_pressed = true; }
+            }
+        } 
+        
+        return namesInSample;
+    }
+	catch(exception& e) {
+		m->errorOut(e, "SubSample", "getSample-TreeMap");
+		exit(1);
+	}
+}	
+
+//**********************************************************************************************************************
+vector<string> SubSample::getSample(TreeMap* tMap, int size) {
+    try {
+        vector<string> sample;
+        
+        vector<string> Groups = tMap->getNamesOfGroups();    
+        for (int i = 0; i < Groups.size(); i++) {
+            
+            if (m->inUsersGroups(Groups[i], m->getGroups())) {
+                if (m->control_pressed) { break; }
+                
+                vector<string> thisGroup; thisGroup.push_back(Groups[i]);
+                vector<string> thisGroupsSeqs = tMap->getNamesSeqs(thisGroup);
+                int thisSize = thisGroupsSeqs.size();
+                
+                if (thisSize >= size) {	
+                    
+                    random_shuffle(thisGroupsSeqs.begin(), thisGroupsSeqs.end());
+                    
+                    for (int j = 0; j < size; j++) { sample.push_back(thisGroupsSeqs[j]); }
+                }else {  m->mothurOut("[ERROR]: You have selected a size that is larger than "+Groups[i]+" number of sequences.\n"); m->control_pressed = true; }
+            }
+        } 
+        
+        return sample;
+    }
+	catch(exception& e) {
+		m->errorOut(e, "SubSample", "getSample-TreeMap");
+		exit(1);
+	}
+}	
+//**********************************************************************************************************************
+vector<string> SubSample::getSample(TreeMap* tMap, vector<string> Groups) {
+    try {
+        vector<string> sample;
+        
+        //vector<string> Groups = tMap->getNamesOfGroups();    
+        for (int i = 0; i < Groups.size(); i++) {
+            
+            if (m->control_pressed) { break; }
+            
+            vector<string> thisGroup; thisGroup.push_back(Groups[i]);
+            vector<string> thisGroupsSeqs = tMap->getNamesSeqs(thisGroup);
+            int thisSize = thisGroupsSeqs.size();
+                
+            for (int j = 0; j < thisSize; j++) { sample.push_back(thisGroupsSeqs[j]); }
+        } 
+        
+        return sample;
+    }
+	catch(exception& e) {
+		m->errorOut(e, "SubSample", "getSample-TreeMap");
+		exit(1);
+	}
+}	
 //**********************************************************************************************************************
 vector<string> SubSample::getSample(vector<SharedRAbundVector*>& thislookup, int size) {
 	try {
@@ -53,7 +250,7 @@ vector<string> SubSample::getSample(vector<SharedRAbundVector*>& thislookup, int
 		
 		//subsampling may have created some otus with no sequences in them
 		eliminateZeroOTUS(thislookup);
-		
+        
 		if (m->control_pressed) { return m->currentBinLabels; }
 		
 		//save mothurOut's binLabels to restore for next label
@@ -64,7 +261,7 @@ vector<string> SubSample::getSample(vector<SharedRAbundVector*>& thislookup, int
 		
 	}
 	catch(exception& e) {
-		m->errorOut(e, "SubSample", "getSample");
+		m->errorOut(e, "SubSample", "getSample-shared");
 		exit(1);
 	}
 }	
@@ -125,8 +322,51 @@ int SubSample::eliminateZeroOTUS(vector<SharedRAbundVector*>& thislookup) {
 		exit(1);
 	}
 }
+//**********************************************************************************************************************
+int SubSample::getSample(SAbundVector*& sabund, int size) {
+	try {
+		
+        OrderVector* order = new OrderVector();
+        *order = sabund->getOrderVector(NULL);
+        
+		int numBins = order->getNumBins();
+		int thisSize = order->getNumSeqs();
+        
+		if (thisSize > size) {
+			random_shuffle(order->begin(), order->end());
+			
+            RAbundVector* rabund = new RAbundVector(numBins);
+			rabund->setLabel(order->getLabel());
 
-
+			for (int j = 0; j < size; j++) {
+                
+				if (m->control_pressed) { delete order; delete rabund; return 0; }
+				
+				int bin = order->get(j);
+				
+				int abund = rabund->get(bin);
+				rabund->set(bin, (abund+1));
+			}
+			
+            delete sabund;
+            sabund = new SAbundVector();
+            *sabund = rabund->getSAbundVector();
+            delete rabund;
+            
+		}else if (thisSize < size) { m->mothurOut("[ERROR]: The size you requested is larger than the number of sequences in the sabund vector. You requested " + toString(size) + " and you only have " + toString(thisSize) + " seqs in your sabund vector.\n"); m->control_pressed = true; }
+		
+		if (m->control_pressed) { return 0; }
+        
+		delete order;
+		
+		return 0;
+		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "SubSampleCommand", "getSample");
+		exit(1);
+	}
+}			
 //**********************************************************************************************************************