]> git.donarmstrong.com Git - mothur.git/blobdiff - distancecommand.cpp
added screen.seqs command - pds
[mothur.git] / distancecommand.cpp
index c20ddb91aecf7a4c0f7460cce4b11601c64ac597..1bc16f34bc231147a3610a894ad3ad04c2d49bed 100644 (file)
@@ -20,14 +20,12 @@ DistanceCommand::DistanceCommand(){
        try {
                globaldata = GlobalData::getInstance();
                validCalculator = new ValidCalculators();
-               ends = globaldata->getEnds();
-               seqDB = globaldata->gSequenceDB;
+               countends = globaldata->getCountEnds();
                convert(globaldata->getProcessors(), processors);
                convert(globaldata->getCutOff(), cutoff);
-               distFile = getRootName(globaldata->getFastaFile()) + "dist";
                
                int i;
-               if (ends != "T") {
+               if (countends == "T") {
                        for (i=0; i<globaldata->Estimators.size(); i++) {
                                if (validCalculator->isValidCalculator("distance", globaldata->Estimators[i]) == true) { 
                                        if (globaldata->Estimators[i] == "nogaps") { 
@@ -35,7 +33,7 @@ DistanceCommand::DistanceCommand(){
                                        }else if (globaldata->Estimators[i] == "eachgap") { 
                                                distCalculator = new eachGapDist();     
                                        }else if (globaldata->Estimators[i] == "onegap") {
-                                               distCalculator = new oneGapDist();                                      }
+                                       distCalculator = new oneGapDist();                                      }
                                }
                        }
                }else {
@@ -51,12 +49,9 @@ DistanceCommand::DistanceCommand(){
                                }
                        }
                }
-                               
-                               
+               
                //reset calc for next command
                globaldata->setCalc("");
-
-               
        }
        catch(exception& e) {
                cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function DistanceCommand. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
@@ -71,27 +66,53 @@ 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();
+               
                int numSeqs = seqDB->getNumSeqs();
+               cutoff += 0.005;
+               
+               string distFile = getRootName(globaldata->getFastaFile()) + "dist";
                
-               system(("rm "+distFile).c_str() );
+               remove(distFile.c_str());
+               
+               //#     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 (__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);        
-//                             system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-//                             system(("rm " + distFile + "tempa").c_str());                           
+                               appendFiles((distFile+"tempa"), distFile);
+                               remove((distFile + "tempa").c_str());   
                        }
                        else{
                                driver(distCalculator, seqDB, (numSeqs/sqrt(2)), numSeqs, distFile + "tempb", cutoff);  
-//                             system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-//                             system(("rm " + distFile + "tempb").c_str());                           
+                               appendFiles((distFile+"tempb"), distFile);
+                               remove((distFile + "tempb").c_str());   
                        }
                        wait(NULL);
-
+                       
                }
                else if(processors == 3){
                        int pid1 = fork();
@@ -99,37 +120,20 @@ int DistanceCommand::execute(){
                                int pid2 = fork();
                                if(pid2 > 0){
                                        driver(distCalculator, seqDB, 0, sqrt(3) * numSeqs / 3, distFile + "tempa", cutoff);
-                                       #ifdef HAVE_CAT
-                                               system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-                                       #else
-                                               #ifdef HAVE_COPY
-//get system call from pat system(("copy " + distFile + "tempa").c_str());
-                                               #else
-                                                       cout << "Sorry but I can't continue because this operating system doesn't appear to support the cat() or copy() system calls." << endl;
-                                               #endif
-                                       #endif
-                                       
-                                       #ifdef HAVE_RM
-                                               system(("rm " + distFile + "tempa").c_str());   
-                                       #else
-                                               #ifdef HAVE_ERASE
-                                                       system(("erase " + distFile + "tempa").c_str());
-                                               #else
-                                                       cout << "Sorry but I can't remove the required files because this operating system doesn't appear to support the rm() or erase() system calls." << endl;
-                                               #endif
-                                       #endif
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());   
                                }
                                else{
                                        driver(distCalculator, seqDB, sqrt(3) * numSeqs / 3, sqrt(6) * numSeqs / 3, distFile + "tempb", cutoff);        
-                                       system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempb").c_str());                           
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
                        }
                        else{
                                driver(distCalculator, seqDB, sqrt(6) * numSeqs / 3, numSeqs, distFile + "tempc", cutoff);      
-                               system(("cat " + distFile + "tempc" + " >> " + distFile).c_str());
-                               system(("rm " + distFile + "tempc").c_str());                           
+                               appendFiles(distFile+"tempc", distFile);
+                               remove((distFile + "tempc").c_str());                   
                        }
                        wait(NULL);
                }
@@ -139,13 +143,13 @@ int DistanceCommand::execute(){
                                int pid2 = fork();
                                if(pid2 > 0){
                                        driver(distCalculator, seqDB, 0, numSeqs / 2, distFile + "tempa", cutoff);      
-                                       system(("cat " + distFile + "tempa" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempa").c_str());                           
+                                       appendFiles(distFile+"tempa", distFile);
+                                       remove((distFile + "tempa").c_str());                   
                                }
                                else{
                                        driver(distCalculator, seqDB, numSeqs / 2, (numSeqs/sqrt(2)), distFile + "tempb", cutoff);      
-                                       system(("cat " + distFile + "tempb" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempb").c_str());                           
+                                       appendFiles(distFile+"tempb", distFile);
+                                       remove((distFile + "tempb").c_str());                           
                                }
                                wait(NULL);
                        }
@@ -153,24 +157,104 @@ int DistanceCommand::execute(){
                                int pid3 = fork();
                                if(pid3 > 0){
                                        driver(distCalculator, seqDB, (numSeqs/sqrt(2)), (sqrt(3) * numSeqs / 2), distFile + "tempc", cutoff);  
-                                       system(("cat " + distFile + "tempc" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempc").c_str());                           
+                                       appendFiles(distFile+"tempc", distFile);
+                                       remove((distFile + "tempc").c_str());                           
                                }
                                else{
                                        driver(distCalculator, seqDB, (sqrt(3) * numSeqs / 2), numSeqs, distFile + "tempd", cutoff);    
-                                       system(("cat " + distFile + "tempd" + " >> " + distFile).c_str());
-                                       system(("rm " + distFile + "tempd").c_str());                           
+                                       appendFiles(distFile+"tempd", distFile);
+                                       remove((distFile + "tempd").c_str());                           
                                }
                                wait(NULL);
                        }
                        wait(NULL);
                }
                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";
@@ -187,29 +271,29 @@ 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 j=0;j<i;j++){
+               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;
                                }
-                       
+                               
                        }
                        if(i % 100 == 0){
                                cout << i << '\t' << time(NULL) - startTime << endl;
                        }
-               
+                       
                }
                cout << endLine-1 << '\t' << time(NULL) - startTime << endl;
-       
+               
                return 1;
        }
        catch(exception& e) {
@@ -220,8 +304,40 @@ 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);
        }       
-
+       
 }
 
 /**************************************************************************************************/
-
+void DistanceCommand::appendFiles(string temp, string filename) {
+       try{
+               ofstream output;
+               ifstream input;
+               
+               //open output file in append mode
+               openOutputFileAppend(filename, output);
+               
+               //open temp file for reading
+               openInputFile(temp, input);
+               
+               string line;
+               //read input file and write to output file
+               while(input.eof() != true) {
+                       getline(input, line); //getline removes the newline char
+                       if (line != "") {
+                               output << line << endl;   // Appending back newline char 
+                       }
+               }       
+               
+               input.close();
+               output.close();
+       }
+       catch(exception& e) {
+               cout << "Standard Error: " << e.what() << " has occurred in the DistanceCommand class Function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }
+       catch(...) {
+               cout << "An unknown error has occurred in the DistanceCommand class function appendFiles. Please contact Pat Schloss at pschloss@microbio.umass.edu." << "\n";
+               exit(1);
+       }       
+}
+/**************************************************************************************************/
\ No newline at end of file