some more mods to shh.seqs

[mothur.git] / shhhercommand.cpp

diff --git a/shhhercommand.cpp b/shhhercommand.cpp
index 39a27d593abd1a3da6d9f93ab89e79b893c72089..b21d4140b84280251cb208705bf0fca70919d0e0 100644 (file)
--- a/shhhercommand.cpp
+++ b/shhhercommand.cpp
@@ -32,7 +32,7 @@
 vector<string> ShhherCommand::getValidParameters(){    
        try {
                string Array[] =  {     
-                       "file", "flow", "lookup", "cutoff", "sigma", "outputdir","inputdir", "processors"       
+                       "file", "flow", "lookup", "cutoff", "sigma", "outputdir","inputdir", "processors", "maxiter", "mindelta", "order"       
                };
                
                vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
@@ -111,7 +111,7 @@ ShhherCommand::ShhherCommand(string option) {
                        
                        //valid paramters for this command
                        string AlignArray[] =  {
-                               "file", "flow", "lookup", "cutoff", "sigma", "outputdir","inputdir", "processors"       
+                               "file", "flow", "lookup", "cutoff", "sigma", "outputdir","inputdir", "processors", "maxiter", "mindelta", "order"       
                        };
                        
                        vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
@@ -211,6 +211,12 @@ ShhherCommand::ShhherCommand(string option) {
                        temp = validParameter.validFile(parameters, "sigma", false);if (temp == "not found")    {       temp = "60";            }
                        convert(temp, sigma); 
                        
+                       flowOrder = validParameter.validFile(parameters, "order", false);
+                       if (flowOrder == "not found"){ flowOrder = "TACG";              }
+                       else if(flowOrder.length() != 4){
+                               m->mothurOut("The value of the order option must be four bases long\n");
+                       }
+                       
                        globaldata = GlobalData::getInstance();
                }
                        
@@ -250,9 +256,6 @@ int ShhherCommand::execute(){
                int tag = 1976;
                MPI_Status status; 
 
-               double begClock = clock();
-               unsigned long int begTime = time(NULL);
-               
                if(pid == 0){
 
                        for(int i=1;i<ncpus;i++){
@@ -288,14 +291,15 @@ int ShhherCommand::execute(){
                        }
                        
                        for(int i=0;i<numFiles;i++){
-                               flowFileName = flowFileVector[i];
+                               double begClock = clock();
+                               unsigned long int begTime = time(NULL);
 
-                               
+                               flowFileName = flowFileVector[i];
                                
                                m->mothurOut("\n>>>>>\tProcessing " + flowFileName + " (file " + toString(i+1) + " of " + toString(numFiles) + ")\t<<<<<\n");
                                m->mothurOut("Reading flowgrams...\n");
-                               
                                getFlowData();
+
                                m->mothurOut("Identifying unique flowgrams...\n");
                                getUniques();
 
@@ -370,7 +374,7 @@ int ShhherCommand::execute(){
                                                MPI_Send(&nSeqsPerOTU[0], numOTUs, MPI_INT, i, tag, MPI_COMM_WORLD);
                                                MPI_Send(&cumNumSeqs[0], numOTUs, MPI_INT, i, tag, MPI_COMM_WORLD);
                                        }
-                               
+                                       
                                        calcCentroidsDriver(0, numOTUsOnCPU);
                                        
                                        for(int i=1;i<ncpus;i++){
@@ -489,7 +493,6 @@ int ShhherCommand::execute(){
                }
                else{
                        int abort = 1;
-                       bool live = 1;
 
                        MPI_Recv(&abort, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                        if(abort){      return 0;       }
@@ -499,6 +502,8 @@ int ShhherCommand::execute(){
 
                        for(int i=0;i<numFiles;i++){
                                //Now into the pyrodist part
+                               bool live = 1;
+
                                char fileName[1024];
                                MPI_Recv(&fileName, 1024, MPI_CHAR, 0, tag, MPI_COMM_WORLD, &status);
                                MPI_Recv(&numSeqs, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
@@ -554,12 +559,12 @@ int ShhherCommand::execute(){
                                int total;
 
                                while(live){
-
+                                       
                                        MPI_Recv(&total, 1, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                                        singleTau.assign(total, 0.0000);
                                        seqNumber.assign(total, 0);
                                        seqIndex.assign(total, 0);
-                                       
+
                                        MPI_Recv(&change[0], numOTUs, MPI_SHORT, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&centroids[0], numOTUs, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&singleTau[0], total, MPI_DOUBLE, 0, tag, MPI_COMM_WORLD, &status);
@@ -567,13 +572,12 @@ int ShhherCommand::execute(){
                                        MPI_Recv(&seqIndex[0], total, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&nSeqsPerOTU[0], total, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&cumNumSeqs[0], numOTUs, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
-                                       
+
                                        calcCentroidsDriver(startOTU, endOTU);
 
                                        MPI_Send(&centroids[0], numOTUs, MPI_INT, 0, tag, MPI_COMM_WORLD);
                                        MPI_Send(&change[0], numOTUs, MPI_SHORT, 0, tag, MPI_COMM_WORLD);
 
-                                       
                                        MPI_Recv(&centroids[0], numOTUs, MPI_INT, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&weight[0], numOTUs, MPI_DOUBLE, 0, tag, MPI_COMM_WORLD, &status);
                                        MPI_Recv(&change[0], numOTUs, MPI_SHORT, 0, tag, MPI_COMM_WORLD, &status);
@@ -592,8 +596,8 @@ int ShhherCommand::execute(){
                                }
                        }
                }               
-
                MPI_Barrier(MPI_COMM_WORLD);
+
                return 0;
 
        }
@@ -1210,7 +1214,7 @@ void ShhherCommand::getOTUData(string listFileName){
                otuData.assign(numSeqs, 0);
                cumNumSeqs.assign(numOTUs, 0);
                nSeqsPerOTU.assign(numOTUs, 0);
-               aaP.resize(numOTUs);
+               aaP.clear();aaP.resize(numOTUs);
                
                seqNumber.clear();
                aaI.clear();
@@ -1266,6 +1270,8 @@ void ShhherCommand::getOTUData(string listFileName){
                        for(int j=nSeqsPerOTU[i];j<numSeqs;j++){
                                aaP[i].push_back(0);
                        }
+                       
+                       
                }
                
                for(int i=1;i<numOTUs;i++){
@@ -1275,6 +1281,7 @@ void ShhherCommand::getOTUData(string listFileName){
                seqIndex = seqNumber;
                
                listFile.close();       
+               
        }
        catch(exception& e) {
                m->errorOut(e, "ShhherCommand", "getOTUData");
@@ -1393,6 +1400,7 @@ void ShhherCommand::calcCentroidsDriver(int start, int finish){
        
        try{
                
+       
                for(int i=start;i<finish;i++){
                        
                        double count = 0;
@@ -1404,7 +1412,7 @@ void ShhherCommand::calcCentroidsDriver(int start, int finish){
                        for(int j=0;j<nSeqsPerOTU[i];j++){
                                count += singleTau[seqNumber[cumNumSeqs[i] + j]];
                        }
-                       
+
                        if(nSeqsPerOTU[i] > 0 && count > MIN_COUNT){
                                vector<double> adF(nSeqsPerOTU[i]);
                                vector<int> anL(nSeqsPerOTU[i]);
@@ -2008,7 +2016,7 @@ void ShhherCommand::writeQualities(vector<int> otuCounts){
 
 void ShhherCommand::writeSequences(vector<int> otuCounts){
        try {
-               string bases = "TACG";
+               flowOrder = "TACG";
                
                string fastaFileName = flowFileName.substr(0,flowFileName.find_last_of('.')) + ".pn.fasta";
                ofstream fastaFile;
@@ -2024,7 +2032,7 @@ void ShhherCommand::writeSequences(vector<int> otuCounts){
                                
                                for(int j=8;j<numFlowCells;j++){
                                        
-                                       char base = bases[j % 4];
+                                       char base = flowOrder[j % 4];
                                        for(int k=0;k<uniqueFlowgrams[index * numFlowCells + j];k++){
                                                fastaFile << base;
                                        }