working on sra and get.mimarkspackage commands. added file parameter to cluster...

[mothur.git] / summarysharedcommand.cpp

diff --git a/summarysharedcommand.cpp b/summarysharedcommand.cpp
index 9893ea747a1a58799280b259c0950f47837b1af7..66a0113899cf32595b1af20c024f8b8210c23aa1 100644 (file)
--- a/summarysharedcommand.cpp
+++ b/summarysharedcommand.cpp
@@ -17,7 +17,7 @@ vector<string> SummarySharedCommand::setParameters(){
                CommandParameter plabel("label", "String", "", "", "", "", "","",false,false); parameters.push_back(plabel);
         CommandParameter psubsample("subsample", "String", "", "", "", "", "","phylip",false,false); parameters.push_back(psubsample);
                CommandParameter pdistance("distance", "Boolean", "", "F", "", "", "","phylip",false,false); parameters.push_back(pdistance);
-               CommandParameter pcalc("calc", "Multiple", "sharedchao-sharedsobs-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-whittaker-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-structchi2-hamming-gower-memchi2-memchord-memeuclidean-mempearson", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan", "", "", "","",true,false,true); parameters.push_back(pcalc);
+               CommandParameter pcalc("calc", "Multiple", "sharedchao-sharedsobs-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan-kstest-whittaker-sharednseqs-ochiai-anderberg-kulczynski-kulczynskicody-lennon-morisitahorn-braycurtis-odum-canberra-structeuclidean-structchord-hellinger-manhattan-structpearson-soergel-spearman-structkulczynski-speciesprofile-structchi2-hamming-gower-memchi2-memchord-memeuclidean-mempearson-jsd-rjsd", "sharedsobs-sharedchao-sharedace-jabund-sorabund-jclass-sorclass-jest-sorest-thetayc-thetan", "", "", "","",true,false,true); parameters.push_back(pcalc);
         CommandParameter poutput("output", "Multiple", "lt-square", "lt", "", "", "","",false,false); parameters.push_back(poutput);
                CommandParameter pall("all", "Boolean", "", "F", "", "", "","",false,false); parameters.push_back(pall);
         CommandParameter piters("iters", "Number", "", "1000", "", "", "","",false,false); parameters.push_back(piters);
@@ -184,9 +184,6 @@ SummarySharedCommand::SummarySharedCommand(string option)  {
                        string temp = validParameter.validFile(parameters, "all", false);                               if (temp == "not found") { temp = "false"; }
                        all = m->isTrue(temp);
                        
-                       temp = validParameter.validFile(parameters, "distance", false);                                 if (temp == "not found") { temp = "false"; }
-                       createPhylip = m->isTrue(temp);
-                       
             temp = validParameter.validFile(parameters, "iters", false);                       if (temp == "not found") { temp = "1000"; }
                        m->mothurConvert(temp, iters); 
             
@@ -204,6 +201,10 @@ SummarySharedCommand::SummarySharedCommand(string option)  {
             
             if (subsample == false) { iters = 0; }
             
+            temp = validParameter.validFile(parameters, "distance", false);                                    if (temp == "not found") { temp = "false"; }
+                       createPhylip = m->isTrue(temp);
+            if (subsample) { createPhylip = true; }
+            
                        temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
                        m->setProcessors(temp);
                        m->mothurConvert(temp, processors); 
@@ -293,6 +294,10 @@ SummarySharedCommand::SummarySharedCommand(string option)  {
                                                        sumCalculators.push_back(new MemEuclidean());
                                                }else if (Estimators[i] == "mempearson") { 
                                                        sumCalculators.push_back(new MemPearson());
+                                               }else if (Estimators[i] == "jsd") {
+                                                       sumCalculators.push_back(new JSD());
+                                               }else if (Estimators[i] == "rjsd") {
+                                                       sumCalculators.push_back(new RJSD());
                                                }
                                        }
                                }
@@ -606,7 +611,7 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
         
             
             if(processors == 1){
-                driver(thisLookup, 0, numGroups, sumFileName+".temp", sumAllFileName+".temp", calcDists);
+                driver(thisItersLookup, 0, numGroups, sumFileName+".temp", sumAllFileName+".temp", calcDists);
                 m->appendFiles((sumFileName + ".temp"), sumFileName);
                 m->mothurRemove((sumFileName + ".temp"));
                 if (mult) {
@@ -621,17 +626,17 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                 //loop through and create all the processes you want
                 while (process != processors) {
-                    int pid = fork();
+                    pid_t pid = fork();
                     
                     if (pid > 0) {
                         processIDS.push_back(pid); 
                         process++;
                     }else if (pid == 0){
-                        driver(thisLookup, lines[process].start, lines[process].end, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);   
+                        driver(thisItersLookup, lines[process].start, lines[process].end, sumFileName + m->mothurGetpid(process) + ".temp", sumAllFileName + m->mothurGetpid(process) + ".temp", calcDists);
                         
                         //only do this if you want a distance file
                         if (createPhylip) {
-                            string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + toString(getpid()) + ".dist";
+                            string tempdistFileName = m->getRootName(m->getSimpleName(sumFileName)) + m->mothurGetpid(process) + ".dist";
                             ofstream outtemp;
                             m->openOutputFile(tempdistFileName, outtemp);
                             
@@ -654,10 +659,10 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
                 }
                 
                 //parent do your part
-                driver(thisLookup, lines[0].start, lines[0].end, sumFileName + toString(getpid()) + ".temp", sumAllFileName + toString(getpid()) + ".temp", calcDists);   
-                m->appendFiles((sumFileName + toString(getpid()) + ".temp"), sumFileName);
-                m->mothurRemove((sumFileName + toString(getpid()) + ".temp"));
-                if (mult) { m->appendFiles((sumAllFileName + toString(getpid()) + ".temp"), sumAllFileName); }
+                driver(thisItersLookup, lines[0].start, lines[0].end, sumFileName + m->mothurGetpid(process) + ".temp", sumAllFileName + m->mothurGetpid(process) + ".temp", calcDists);
+                m->appendFiles((sumFileName + m->mothurGetpid(process) + ".temp"), sumFileName);
+                m->mothurRemove((sumFileName + m->mothurGetpid(process) + ".temp"));
+                if (mult) { m->appendFiles((sumAllFileName + m->mothurGetpid(process) + ".temp"), sumAllFileName); }
                 
                 //force parent to wait until all the processes are done
                 for (int i = 0; i < processIDS.size(); i++) {
@@ -703,7 +708,7 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
                 
                 vector<summarySharedData*> pDataArray; 
                 DWORD   dwThreadIdArray[processors-1];
-                HANDLE  hThreadArray[processors-1]; 
+                HANDLE  hThreadArray[processors-1];
                 
                 //Create processor worker threads.
                 for( int i=1; i<processors; i++ ){
@@ -716,11 +721,12 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
                         temp->setGroup(thisLookup[k]->getGroup());
                         newLookup.push_back(temp);
                     }
+                
                     
                     //for each bin
-                    for (int k = 0; k < thisLookup[0]->getNumBins(); k++) {
+                    for (int k = 0; k < thisItersLookup[0]->getNumBins(); k++) {
                         if (m->control_pressed) { for (int j = 0; j < newLookup.size(); j++) {  delete newLookup[j];  } return 0; }
-                        for (int j = 0; j < thisLookup.size(); j++) { newLookup[j]->push_back(thisLookup[j]->getAbundance(k), thisLookup[j]->getGroup()); }
+                        for (int j = 0; j < thisItersLookup.size(); j++) { newLookup[j]->push_back(thisItersLookup[j]->getAbundance(k), thisItersLookup[j]->getGroup()); }
                     }
                     
                     // Allocate memory for thread data.
@@ -732,7 +738,7 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
                 }
                 
                 //parent do your part
-                driver(thisLookup, lines[0].start, lines[0].end, sumFileName +"0.temp", sumAllFileName + "0.temp", calcDists);   
+                driver(thisItersLookup, lines[0].start, lines[0].end, sumFileName +"0.temp", sumAllFileName + "0.temp", calcDists);
                 m->appendFiles((sumFileName + "0.temp"), sumFileName);
                 m->mothurRemove((sumFileName + "0.temp"));
                 if (mult) { m->appendFiles((sumAllFileName + "0.temp"), sumAllFileName); }
@@ -742,10 +748,13 @@ int SummarySharedCommand::process(vector<SharedRAbundVector*> thisLookup, string
                 
                 //Close all thread handles and free memory allocations.
                 for(int i=0; i < pDataArray.size(); i++){
+                    if (pDataArray[i]->count != (pDataArray[i]->end-pDataArray[i]->start)) {
+                        m->mothurOut("[ERROR]: process " + toString(i) + " only processed " + toString(pDataArray[i]->count) + " of " + toString(pDataArray[i]->end-pDataArray[i]->start) + " groups assigned to it, quitting. \n"); m->control_pressed = true; 
+                    }
                     m->appendFiles((sumFileName + toString(processIDS[i]) + ".temp"), sumFileName);
                     m->mothurRemove((sumFileName + toString(processIDS[i]) + ".temp"));
                     
-                    for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  } 
+                    for (int j = 0; j < pDataArray[i]->thisLookup.size(); j++) {  delete pDataArray[i]->thisLookup[j];  }
                     
                     if (createPhylip) {
                         for (int k = 0; k < calcDists.size(); k++) {