changed dups parameter to dereplicate in chimera.uchime.

[mothur.git] / classify.cpp

diff --git a/classify.cpp b/classify.cpp
index f44e66c359873f6830dac72d99fb1727a73c27ad..8aa3cdb381ed7a389667ce61962d47cefac15ddd 100644 (file)
--- a/classify.cpp
+++ b/classify.cpp
@@ -61,7 +61,8 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
                                        names.push_back(temp.getName());
                                        database->addSequence(temp);    
                                }
-                               database->generateDB();
+                               if ((method == "kmer") && (!shortcuts)) {;} //don't print
+                else {database->generateDB(); }
                        }else if ((method == "kmer") && (!needToGenerate)) {    
                                ifstream kmerFileTest(kmerDBName.c_str());
                                database->readKmerDB(kmerFileTest);     
@@ -201,7 +202,7 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
                                fastaFile.close();
 
                 if ((method == "kmer") && (!shortcuts)) {;} //don't print
-                else {database->generateDB(); }
+                else {database->generateDB(); } 
                                
                        }else if ((method == "kmer") && (!needToGenerate)) {    
                                ifstream kmerFileTest(kmerDBName.c_str());
@@ -353,3 +354,37 @@ vector<string> Classify::parseTax(string tax) {
 }
 /**************************************************************************************************/
 
+double Classify::getLogExpSum(vector<double> probabilities, int& maxIndex){
+       try {
+        //     http://jblevins.org/notes/log-sum-exp
+        
+        double maxProb = probabilities[0];
+        maxIndex = 0;
+        
+        int numProbs = (int)probabilities.size();
+        
+        for(int i=1;i<numProbs;i++){
+            if(probabilities[i] >= maxProb){
+                maxProb = probabilities[i];
+                maxIndex = i;
+            }
+        }
+        
+        double probSum = 0.0000;
+        
+        for(int i=0;i<numProbs;i++){
+            probSum += exp(probabilities[i] - maxProb);                
+        }
+        
+        probSum = log(probSum) + maxProb;
+        
+        return probSum;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Classify", "getLogExpSum");
+               exit(1);
+       }
+}
+
+/**************************************************************************************************/
+