]> git.donarmstrong.com Git - mothur.git/blobdiff - distancecommand.cpp
line and label no longer persist between commands. Added phylip formated output...
[mothur.git] / distancecommand.cpp
index 90ffb8df31d5aff2a47ee0f29ce9df33397d39b7..e2ca937ed971e87194fd3623c48aec343d2f3ca5 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("");
        }
@@ -67,66 +72,81 @@ 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";
+               string phylipFile = getRootName(globaldata->getFastaFile()) + "phylip.dist";
                
+               remove(phylipFile.c_str());
                remove(distFile.c_str());
                
+               //output numSeqs to phylip formatted dist file
+               openOutputFile(phylipFile, phylipOut);
+               phylipOut << numSeqs << endl;
+               phylipOut.close();
+               
+               //#     if defined (_WIN32)
+               //figure out how to implement the fork and wait commands in windows
+               //      driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile, cutoff);
+               //#     endif
+               
+#if defined (__APPLE__) || (__MACH__)
                if(processors == 1){
-                       driver(distCalculator, seqDB, 0, numSeqs, distFile, cutoff);    
+                       driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile + "tempPhylipA", cutoff);
+                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                       remove((phylipFile + "tempPhylipA").c_str());   
                }       
                else if(processors == 2){
-               
+                       
                        int pid = fork();
                        if(pid > 0){
-                               driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", cutoff);        
+                               driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", phylipFile + "tempPhylipA", cutoff);    
                                appendFiles((distFile+"tempa"), distFile);
-                               remove((distFile + "tempa").c_str());   
+                               remove((distFile + "tempa").c_str());
+                               appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                               remove((phylipFile + "tempPhylipA").c_str());
                        }
                        else{
-                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
+                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", phylipFile + "tempPhylipB", cutoff);      
                                appendFiles((distFile+"tempb"), distFile);
-                               remove((distFile + "tempb").c_str());   
+                               remove((distFile + "tempb").c_str());
+                               appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                               remove((phylipFile + "tempPhylipB").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);
+                                       driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", phylipFile + "tempPhylipA", cutoff);
                                        appendFiles(distFile+"tempa", distFile);
-                                       remove((distFile + "tempa").c_str());   
+                                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                                       remove((distFile + "tempa").c_str());
+                                       remove((phylipFile + "tempPhylipA").c_str());   
                                }
                                else{
-                                       driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
+                                       driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", phylipFile + "tempPhylipB", cutoff);    
                                        appendFiles(distFile+"tempb", distFile);
-                                       remove((distFile + "tempb").c_str());                           
+                                       appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                                       remove((distFile + "tempb").c_str());   
+                                       remove((phylipFile + "tempPhylipB").c_str());                   
                                }
                                wait(NULL);
                        }
                        else{
-                               driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
+                               driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", phylipFile + "tempPhylipC", cutoff);  
                                appendFiles(distFile+"tempc", distFile);
-                               remove((distFile + "tempc").c_str());                   
+                               appendFiles((phylipFile + "tempPhylipC"), phylipFile);
+                               remove((distFile + "tempc").c_str());   
+                               remove((phylipFile + "tempPhylipC").c_str());           
                        }
                        wait(NULL);
                }
@@ -135,39 +155,152 @@ int DistanceCommand::execute(){
                        if(pid1 > 0){
                                int pid2 = fork();
                                if(pid2 > 0){
-                                       driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
+                                       driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", phylipFile + "tempPhylipA", cutoff);  
                                        appendFiles(distFile+"tempa", distFile);
-                                       remove((distFile + "tempa").c_str());                   
+                                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                                       remove((distFile + "tempa").c_str());
+                                       remove((phylipFile + "tempPhylipA").c_str());                   
                                }
                                else{
-                                       driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
+                                       driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", phylipFile + "tempPhylipB", cutoff);  
                                        appendFiles(distFile+"tempb", distFile);
-                                       remove((distFile + "tempb").c_str());                           
+                                       appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                                       remove((distFile + "tempb").c_str());
+                                       remove((phylipFile + "tempPhylipB").c_str());                           
                                }
                                wait(NULL);
                        }
                        else{
                                int pid3 = fork();
                                if(pid3 > 0){
-                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
+                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", phylipFile + "tempPhylipC", cutoff);      
                                        appendFiles(distFile+"tempc", distFile);
-                                       remove((distFile + "tempc").c_str());                           
+                                       appendFiles((phylipFile + "tempPhylipC"), phylipFile);
+                                       remove((distFile + "tempc").c_str());
+                                       remove((phylipFile + "tempPhylipC").c_str());   
+                                                               
                                }
                                else{
-                                       driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
+                                       driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", phylipFile + "tempPhylipD", cutoff);        
                                        appendFiles(distFile+"tempd", distFile);
-                                       remove((distFile + "tempd").c_str());                           
+                                       appendFiles((phylipFile + "tempPhylipD"), phylipFile);
+                                       remove((distFile + "tempd").c_str());   
+                                       remove((phylipFile + "tempPhylipD").c_str());                   
                                }
                                wait(NULL);
                        }
                        wait(NULL);
                }
                wait(NULL);
-       
+#elif (linux) || (__linux)
+               if(processors == 1){
+                       driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile + "tempPhylipA", cutoff);
+                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);  
+                       remove((phylipFile + "tempPhylipA").c_str());   
+               }       
+               else if(processors == 2){
+                       
+                       int pid = fork();
+                       if(pid > 0){
+                               driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", phylipFile + "tempPhylipA", cutoff);    
+                               appendFiles((distFile+"tempa"), distFile);
+                               appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                               remove((distFile + "tempa").c_str());   
+                               remove((phylipFile + "tempPhylipA").c_str());
+
+                       }
+                       else{
+                               driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", phylipFile + "tempPhylipB", cutoff);      
+                               appendFiles((distFile+"tempb"), distFile);
+                               appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                               remove((distFile + "tempb").c_str());
+                               remove((phylipFile + "tempPhylipB").c_str());   
+                       }
+                       wait();
+                       
+               }
+               else if(processors == 3){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", phylipFile + "tempPhylipA", cutoff);    
+                                       appendFiles((distFile+"tempa"), distFile);
+                                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                                       remove((distFile + "tempa").c_str());   
+                                       remove((phylipFile + "tempPhylipA").c_str());
+
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", phylipFile + "tempPhylipB", cutoff);      
+                                       appendFiles((distFile+"tempb"), distFile);
+                                       appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                                       remove((distFile + "tempb").c_str());
+                                       remove((phylipFile + "tempPhylipB").c_str());   
+                               }
+                               wait();
+                       }
+                       else{
+                               driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", phylipFile + "tempPhylipC", cutoff);  
+                               appendFiles(distFile+"tempc", distFile);
+                               appendFiles((phylipFile + "tempPhylipC"), phylipFile);
+                               remove((distFile + "tempc").c_str());   
+                               remove((phylipFile + "tempPhylipC").c_str());                           
+                       }
+                       wait();
+               }
+               else if(processors == 4){
+                       int pid1 = fork();
+                       if(pid1 > 0){
+                               int pid2 = fork();
+                               if(pid2 > 0){
+                                       driver(distCalculator, seqDB, 0, (numSeqs/sqrt(2)), distFile + "tempa", phylipFile + "tempPhylipA", cutoff);    
+                                       appendFiles((distFile+"tempa"), distFile);
+                                       appendFiles((phylipFile + "tempPhylipA"), phylipFile);
+                                       remove((distFile + "tempa").c_str());   
+                                       remove((phylipFile + "tempPhylipA").c_str());
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", phylipFile + "tempPhylipB", cutoff);      
+                                       appendFiles((distFile+"tempb"), distFile);
+                                       appendFiles((phylipFile + "tempPhylipB"), phylipFile);
+                                       remove((distFile + "tempb").c_str());
+                                       remove((phylipFile + "tempPhylipB").c_str());   
+                               }
+                               wait();
+                       }
+                       else{
+                               int pid3 = fork();
+                               if(pid3 > 0){
+                                       driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", phylipFile + "tempPhylipC", cutoff);  
+                                       appendFiles(distFile+"tempc", distFile);
+                                       appendFiles((phylipFile + "tempPhylipC"), phylipFile);
+                                       remove((distFile + "tempc").c_str());   
+                                       remove((phylipFile + "tempPhylipC").c_str());                                   
+                               }
+                               else{
+                                       driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", phylipFile + "tempPhylipD", cutoff);        
+                                       appendFiles(distFile+"tempd", distFile);
+                                       appendFiles((phylipFile + "tempPhylipD"), phylipFile);
+                                       remove((distFile + "tempd").c_str());   
+                                       remove((phylipFile + "tempPhylipD").c_str());                           
+                               }
+                               wait();
+                       }
+                       wait();
+               }
+               wait();
+               
+#else
+               driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile + "tempPhylipA", cutoff);
+               appendFiles((phylipFile + "tempPhylipA"), phylipFile);  
+               remove((phylipFile + "tempPhylipA").c_str());
+#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";
@@ -181,32 +314,44 @@ int DistanceCommand::execute(){
 
 /**************************************************************************************************/
 /////// need to fix to work with calcs and sequencedb
-int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLine, int endLine, string dFileName, float cutoff){
+int DistanceCommand::driver(Dist* distCalculator, SequenceDB* align, int startLine, int endLine, string dFileName, string pFilename, float cutoff){
        try {
                int startTime = time(NULL);
-       
+               
+               //column file
                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 j=0;j<i;j++){   
-                               distCalculator->calcDist(align->get(i), align->get(j));
+               //column file
+               ofstream philFile(pFilename.c_str(), ios::trunc);
+               philFile.setf(ios::fixed, ios::showpoint);
+               philFile << setprecision(4);
+               
+               for(int i=startLine;i<endLine;i++){
+                       
+                       for(int j=0;j<i;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;
                                }
-                       
+                               philFile << dist << '\t';
                        }
+                       
+                       philFile << endl;
+                       
                        if(i % 100 == 0){
                                cout << i << '\t' << time(NULL) - startTime << endl;
                        }
-               
+                       
                }
                cout << endLine-1 << '\t' << time(NULL) - startTime << endl;
-       
+               
+               //philFile.close();
+               //distFile.close();
+               
                return 1;
        }
        catch(exception& e) {
@@ -217,7 +362,7 @@ 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);
        }       
-
+       
 }
 
 /**************************************************************************************************/
@@ -240,7 +385,7 @@ void DistanceCommand::appendFiles(string temp, string filename) {
                                output << line << endl;   // Appending back newline char 
                        }
                }       
-                       
+               
                input.close();
                output.close();
        }