]> git.donarmstrong.com Git - mothur.git/blobdiff - distancecommand.cpp
removed readseqs, readfasta, readnexus, readclustal, readseqsphylip and updated seque...
[mothur.git] / distancecommand.cpp
index 140aa6085b4007a3998ffc6a769b266464b16e6a..6eb7d576833e0ba944d8f445ff662f2e1d86e661 100644 (file)
@@ -20,12 +20,17 @@ DistanceCommand::DistanceCommand(){
        try {
                globaldata = GlobalData::getInstance();
                validCalculator = new ValidCalculators();
-               ends = globaldata->getEnds();
+               countends = globaldata->getCountEnds();
                convert(globaldata->getProcessors(), processors);
                convert(globaldata->getCutOff(), cutoff);
                
+               //open file
+               string filename = globaldata->getFastaFile();
+               openInputFile(filename, in);
+
+               
                int i;
-               if (ends != "T") {
+               if (isTrue(countends) == true) {
                        for (i=0; i<globaldata->Estimators.size(); i++) {
                                if (validCalculator->isValidCalculator("distance", globaldata->Estimators[i]) == true) { 
                                        if (globaldata->Estimators[i] == "nogaps") { 
@@ -33,7 +38,7 @@ DistanceCommand::DistanceCommand(){
                                        }else if (globaldata->Estimators[i] == "eachgap") { 
                                                distCalculator = new eachGapDist();     
                                        }else if (globaldata->Estimators[i] == "onegap") {
-                                               distCalculator = new oneGapDist();                                      }
+                                       distCalculator = new oneGapDist();                                      }
                                }
                        }
                }else {
@@ -49,7 +54,7 @@ DistanceCommand::DistanceCommand(){
                                }
                        }
                }
-                               
+               
                //reset calc for next command
                globaldata->setCalc("");
        }
@@ -67,113 +72,183 @@ DistanceCommand::DistanceCommand(){
 int DistanceCommand::execute(){
        try {
                
-               //read file
-               string filename = globaldata->inputFileName;
-               
-               if(globaldata->getFastaFile() != "") {
-                       readSeqs =  new ReadFasta(filename); }
-               else if(globaldata->getNexusFile() != "") {
-                       readSeqs = new ReadNexus(filename); }
-               else if(globaldata->getClustalFile() != "") {
-                       readSeqs = new ReadClustal(filename); }
-               else if(globaldata->getPhylipFile() != "") {
-                       readSeqs = new ReadPhylip(filename); }
-                       
-               readSeqs->read();
-               seqDB = readSeqs->getDB();
-       
+               //reads fasta file and fills sequenceDB
+               if(globaldata->getFastaFile() != "") {  seqDB = new SequenceDB(in);  }
+               else { cout << "Error no fasta file." << endl; return 0; }
+                               
                int numSeqs = seqDB->getNumSeqs();
+               cutoff += 0.005;
                
                string distFile = getRootName(globaldata->getFastaFile()) + "dist";
                
                remove(distFile.c_str());
                
-       //      #       if defined (WIN_VERSION)
+               //#     if defined (_WIN32)
+               //figure out how to implement the fork and wait commands in windows
                //      driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);
-       //      #       endif
-               
-       //      #       if defined (LINUX_VERSION)
-                       if(processors == 1){
-                               driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);    
-                       }       
-                       else if(processors == 2){
+               //#     endif
                
-                               int pid = fork();
-                               if(pid > 0){
-                                       driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", cutoff);        
-                                       appendFiles((distFile+"tempa"), distFile);
+#if defined (__APPLE__) || (__MACH__)
+               if(processors == 1){
+                       driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);    
+               }       
+               else if(processors == 2){
+                       
+                       int pid = fork();
+                       if(pid > 0){
+                               driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", cutoff);        
+                               appendFiles((distFile+"tempa"), distFile);
+                               remove((distFile + "tempa").c_str());   
+                       }
+                       else{
+                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
+                               appendFiles((distFile+"tempb"), distFile);
+                               remove((distFile + "tempb").c_str());   
+                       }
+                       wait(NULL);
+                       
+               }
+               else if(processors == 3){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", cutoff);
+                                       appendFiles(distFile+"tempa", distFile);
                                        remove((distFile + "tempa").c_str());   
                                }
                                else{
-                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
-                                       appendFiles((distFile+"tempb"), distFile);
-                                       remove((distFile + "tempb").c_str());   
+                                       driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
-
                        }
-                       else if(processors == 3){
-                               int pid1 = fork();
-                               if(pid1 > 0){
-                                       int pid2 = fork();
-                                       if(pid2 > 0){
-                                               driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", cutoff);
-                                               appendFiles(distFile+"tempa", distFile);
-                                               remove((distFile + "tempa").c_str());   
-                                       }
-                                       else{
-                                               driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
-                                               appendFiles(distFile+"tempb", distFile);
-                                               remove((distFile + "tempb").c_str());                           
-                                       }
-                                       wait(NULL);
+                       else{
+                               driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
+                               appendFiles(distFile+"tempc", distFile);
+                               remove((distFile + "tempc").c_str());                   
+                       }
+                       wait(NULL);
+               }
+               else if(processors == 4){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());                   
                                }
                                else{
-                                       driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
-                                       appendFiles(distFile+"tempc", distFile);
-                                       remove((distFile + "tempc").c_str());                   
+                                       driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
                        }
-                       else if(processors == 4){
-                               int pid1 = fork();
-                               if(pid1 > 0){
-                                       int pid2 = fork();
-                                       if(pid2 > 0){
-                                               driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
-                                               appendFiles(distFile+"tempa", distFile);
-                                               remove((distFile + "tempa").c_str());                   
-                                       }
-                                       else{
-                                               driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
-                                               appendFiles(distFile+"tempb", distFile);
-                                               remove((distFile + "tempb").c_str());                           
-                                       }
-                                       wait(NULL);
+                       else{
+                               int pid3 = fork();
+                               if(pid3 > 0){
+                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
+                                       appendFiles(distFile+"tempc", distFile);
+                                       remove((distFile + "tempc").c_str());                           
                                }
                                else{
-                                       int pid3 = fork();
-                                       if(pid3 > 0){
-                                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
-                                               appendFiles(distFile+"tempc", distFile);
-                                               remove((distFile + "tempc").c_str());                           
-                                       }
-                                       else{
-                                               driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
-                                               appendFiles(distFile+"tempd", distFile);
-                                               remove((distFile + "tempd").c_str());                           
-                                       }
-                                       wait(NULL);
+                                       driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
+                                       appendFiles(distFile+"tempd", distFile);
+                                       remove((distFile + "tempd").c_str());                           
                                }
                                wait(NULL);
                        }
                        wait(NULL);
-               //#     endif
-       
+               }
+               wait(NULL);
+#elif (linux) || (__linux)
+               if(processors == 1){
+                       driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);    
+               }       
+               else if(processors == 2){
+                       
+                       int pid = fork();
+                       if(pid > 0){
+                               driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", cutoff);        
+                               appendFiles((distFile+"tempa"), distFile);
+                               remove((distFile + "tempa").c_str());   
+                       }
+                       else{
+                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
+                               appendFiles((distFile+"tempb"), distFile);
+                               remove((distFile + "tempb").c_str());   
+                       }
+                       wait();
+                       
+               }
+               else if(processors == 3){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", cutoff);
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());   
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
+                               }
+                               wait();
+                       }
+                       else{
+                               driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
+                               appendFiles(distFile+"tempc", distFile);
+                               remove((distFile + "tempc").c_str());                   
+                       }
+                       wait();
+               }
+               else if(processors == 4){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());                   
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
+                               }
+                               wait();
+                       }
+                       else{
+                               int pid3 = fork();
+                               if(pid3 > 0){
+                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
+                                       appendFiles(distFile+"tempc", distFile);
+                                       remove((distFile + "tempc").c_str());                           
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
+                                       appendFiles(distFile+"tempd", distFile);
+                                       remove((distFile + "tempd").c_str());                           
+                               }
+                               wait();
+                       }
+                       wait();
+               }
+               wait();
+               
+#else
+               driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);
+#endif
+               
                delete distCalculator;
-       
+               
                return 0;
-
+               
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function execute. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -190,33 +265,29 @@ int DistanceCommand::execute(){
 int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLine, int endLine, string dFileName, float cutoff){
        try {
                int startTime = time(NULL);
-       
+               
                ofstream distFile(dFileName.c_str(), ios::trunc);
                distFile.setf(ios::fixed, ios::showpoint);
                distFile << setprecision(4);
-       
-               for(int i=startLine;i<endLine;i++){
                
+               for(int i=startLine;i<endLine;i++){
+                       
                        for(int j=0;j<i;j++){
-//cout << "unaligned" << endl;
-//cout << align->get(i).getUnaligned() << "  " << align->get(j).getUnaligned() << endl;
-//cout << "aligned" << endl;
-//cout << align->get(i).getAligned() << "  " << align->get(j).getAligned() << endl;
-                               distCalculator->calcDist(align->get(i), align->get(j));
+                               distCalculator->calcDist(*(align->get(i)), *(align->get(j)));
                                double dist = distCalculator->getDist();
-
+                               
                                if(dist <= cutoff){
-                                       distFile << align->get(i).getName() << ' ' << align->get(j).getName() << ' ' << dist << endl;
+                                       distFile << align->get(i)->getName() << ' ' << align->get(j)->getName() << ' ' << dist << endl;
                                }
-                       
+                               
                        }
                        if(i % 100 == 0){
                                cout << i << '\t' << time(NULL) - startTime << endl;
                        }
-               
+                       
                }
                cout << endLine-1 << '\t' << time(NULL) - startTime << endl;
-       
+               
                return 1;
        }
        catch(exception& e) {
@@ -227,7 +298,7 @@ int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLi
                cout << "An unknown error has occurred in the DistanceCommand class function driver. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
                exit(1);
        }       
-
+       
 }
 
 /**************************************************************************************************/
@@ -250,7 +321,7 @@ void DistanceCommand::appendFiles(string temp, string filename) {
                                output << line << endl;   // Appending back newline char 
                        }
                }       
-                       
+               
                input.close();
                output.close();
        }