added a few evenness calculators and fixed a couple of bugs in filter.seqs and pre...

[mothur.git] / clustercommand.cpp

diff --git a/clustercommand.cpp b/clustercommand.cpp
index 7484a621645c2449bfc9f158a391549f33da71dd..3f27560ae9bfbc69c5ac3d49875847a829f49bf3 100644 (file)
--- a/clustercommand.cpp
+++ b/clustercommand.cpp
@@ -22,7 +22,7 @@ ClusterCommand::ClusterCommand(string option)  {
                
                else {
                        //valid paramters for this command
-                       string Array[] =  {"cutoff","precision","method","showabund","timing","outputdir","inputdir"};
+                       string Array[] =  {"cutoff","precision","method","showabund","timing","hard","outputdir","inputdir"};
                        vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
                        
                        OptionParser parser(option);
@@ -56,16 +56,19 @@ ClusterCommand::ClusterCommand(string option)  {
                        length = temp.length();
                        convert(temp, precision); 
                        
+                       temp = validParameter.validFile(parameters, "hard", false);                     if (temp == "not found") { temp = "F"; }
+                       hard = isTrue(temp);
+                       
                        temp = validParameter.validFile(parameters, "cutoff", false);
                        if (temp == "not found") { temp = "10"; }
                        convert(temp, cutoff); 
-                       cutoff += (5 / (precision * 10.0));
+                       cutoff += (5 / (precision * 10.0));  
                        
                        method = validParameter.validFile(parameters, "method", false);
                        if (method == "not found") { method = "furthest"; }
                        
-                       if ((method == "furthest") || (method == "nearest") || (method == "average")) { }
-                       else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest or average."); m->mothurOutEndLine(); abort = true; }
+                       if ((method == "furthest") || (method == "nearest") || (method == "average") || (method == "weighted")) { }
+                       else { m->mothurOut("Not a valid clustering method.  Valid clustering algorithms are furthest, nearest, average, and weighted."); m->mothurOutEndLine(); abort = true; }
 
                        showabund = validParameter.validFile(parameters, "showabund", false);
                        if (showabund == "not found") { showabund = "T"; }
@@ -88,6 +91,7 @@ ClusterCommand::ClusterCommand(string option)  {
                                if (method == "furthest")       {       cluster = new CompleteLinkage(rabund, list, matrix, cutoff, method); }
                                else if(method == "nearest"){   cluster = new SingleLinkage(rabund, list, matrix, cutoff, method); }
                                else if(method == "average"){   cluster = new AverageLinkage(rabund, list, matrix, cutoff, method);     }
+                               else if(method == "weighted"){  cluster = new WeightedLinkage(rabund, list, matrix, cutoff, method);    }
                                tag = cluster->getTag();
                                
                                if (outputDir == "") { outputDir += hasPath(globaldata->inputFileName); }
@@ -114,7 +118,7 @@ ClusterCommand::ClusterCommand(string option)  {
 void ClusterCommand::help(){
        try {
                m->mothurOut("The cluster command can only be executed after a successful read.dist command.\n");
-               m->mothurOut("The cluster command parameter options are method, cuttoff, precision, showabund and timing. No parameters are required.\n");
+               m->mothurOut("The cluster command parameter options are method, cuttoff, hard, precision, showabund and timing. No parameters are required.\n");
                m->mothurOut("The cluster command should be in the following format: \n");
                m->mothurOut("cluster(method=yourMethod, cutoff=yourCutoff, precision=yourPrecision) \n");
                m->mothurOut("The acceptable cluster methods are furthest, nearest and average.  If no method is provided then furthest is assumed.\n\n");      
@@ -176,8 +180,14 @@ int ClusterCommand::execute(){
                        loops++;
 
                        cluster->update(cutoff);
+       
                        float dist = matrix->getSmallDist();
-                       float rndDist = roundDist(dist, precision);
+                       float rndDist;
+                       if (hard) {
+                               rndDist = ceilDist(dist, precision); 
+                       }else{
+                               rndDist = roundDist(dist, precision); 
+                       }
 
                        if(previousDist <= 0.0000 && dist != previousDist){
                                printData("unique");
@@ -209,7 +219,7 @@ int ClusterCommand::execute(){
                //delete globaldata's copy of the sparsematrix and listvector to free up memory
                delete globaldata->gSparseMatrix;  globaldata->gSparseMatrix = NULL;
                delete globaldata->gListVector;  globaldata->gListVector = NULL;
-               
+       
                //saves .list file so you can do the collect, rarefaction and summary commands without doing a read.list
                if (globaldata->getFormat() == "phylip") { globaldata->setPhylipFile(""); }
                else if (globaldata->getFormat() == "column") { globaldata->setColumnFile(""); }
@@ -221,8 +231,13 @@ int ClusterCommand::execute(){
                sabundFile.close();
                rabundFile.close();
                listFile.close();
-               
-               if (saveCutoff != cutoff) { m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine();  }
+       
+               if (saveCutoff != cutoff) { 
+                       if (hard)       {  saveCutoff = ceilDist(saveCutoff, precision);        }
+                       else            {       saveCutoff = roundDist(saveCutoff, precision);  }
+
+                       m->mothurOut("changed cutoff to " + toString(cutoff)); m->mothurOutEndLine(); 
+               }
                
                m->mothurOutEndLine();
                m->mothurOut("Output File Names: "); m->mothurOutEndLine();