]> git.donarmstrong.com Git - mothur.git/blobdiff - screenseqscommand.cpp
projects / mothur.git / blobdiff
? search:
summary | shortlog | log | commit | commitdiff | tree
raw | inline | side by side
added load.logfile command. changed summary.single output for subsample=t.
[mothur.git] / screenseqscommand.cpp
diff --git a/screenseqscommand.cpp b/screenseqscommand.cpp
index 9494865c4f66e302182b8fd221e0b44951cb2518..7f385d39c729e1acbc2d33325e9640c706f095e2 100644 (file)
--- a/screenseqscommand.cpp
+++ b/screenseqscommand.cpp
@@ -70,6 +70,33 @@ string ScreenSeqsCommand::getHelpString(){
                exit(1);
        }
 }
+//**********************************************************************************************************************
+string ScreenSeqsCommand::getOutputFileNameTag(string type, string inputName=""){      
+       try {
+        string outputFileName = "";
+               map<string, vector<string> >::iterator it;
+        
+        //is this a type this command creates
+        it = outputTypes.find(type);
+        if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
+        else {
+            if (type == "fasta")            {   outputFileName =  "good" + m->getExtension(inputName);   }
+            else if (type == "taxonomy")    {   outputFileName =  "good" + m->getExtension(inputName);   }
+            else if (type == "name")        {   outputFileName =  "good" + m->getExtension(inputName);   }
+            else if (type == "group")       {   outputFileName =  "good" + m->getExtension(inputName);   }
+            else if (type == "accnos")      {   outputFileName =  "bad.accnos";   }
+            else if (type == "qfile")       {   outputFileName =  "good" + m->getExtension(inputName);   }
+            else if (type == "alignreport") {   outputFileName =  "good.align.report";                   }
+            else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
+        }
+        return outputFileName;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "ScreenSeqsCommand", "getOutputFileNameTag");
+               exit(1);
+       }
+}
+
 //**********************************************************************************************************************
 ScreenSeqsCommand::ScreenSeqsCommand(){        
        try {
@@ -288,7 +315,7 @@ int ScreenSeqsCommand::execute(){
                        getSummary(positions); 
                } 
                else { 
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                 positions = m->divideFile(fastafile, processors);
                 for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }
                        #else 
@@ -296,6 +323,7 @@ int ScreenSeqsCommand::execute(){
                 else {
                     int numFastaSeqs = 0;
                     positions = m->setFilePosFasta(fastafile, numFastaSeqs); 
+                    if (positions.size() < processors) { processors = positions.size(); }
                 
                     //figure out how many sequences you have to process
                     int numSeqsPerProcessor = numFastaSeqs / processors;
@@ -308,8 +336,8 @@ int ScreenSeqsCommand::execute(){
                        #endif
                }
                                        
-               string goodSeqFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + "good" + m->getExtension(fastafile);
-               string badAccnosFile =  outputDir + m->getRootName(m->getSimpleName(fastafile)) + "bad.accnos";
+               string goodSeqFile = outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("fasta", fastafile);
+               string badAccnosFile =  outputDir + m->getRootName(m->getSimpleName(fastafile)) + getOutputFileNameTag("accnos");
                
                int numFastaSeqs = 0;
                set<string> badSeqNames;
@@ -398,17 +426,10 @@ int ScreenSeqsCommand::execute(){
                        MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
                                        
 #else
-                                               
-       //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-                       if(processors == 1){
-                               numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);            
-                       }else{
-                               numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames); 
-            }
-       //#else
-       //              numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);            
-       //#endif
-            if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
+        if(processors == 1){ numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);      }       
+        else{ numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames); }
+        
+        if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 #endif         
 
                #ifdef USE_MPI
@@ -520,7 +541,7 @@ int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
                string seqName, seqList, group;
                set<string>::iterator it;
 
-               string goodNameFile = outputDir + m->getRootName(m->getSimpleName(namefile)) + "good" + m->getExtension(namefile);
+               string goodNameFile = outputDir + m->getRootName(m->getSimpleName(namefile)) + getOutputFileNameTag("name", namefile);
                outputNames.push_back(goodNameFile);  outputTypes["name"].push_back(goodNameFile);
                
                ofstream goodNameOut;   m->openOutputFile(goodNameFile, goodNameOut);
@@ -566,7 +587,7 @@ int ScreenSeqsCommand::screenNameGroupFile(set<string> badSeqNames){
                        ifstream inputGroups;
                        m->openInputFile(groupfile, inputGroups);
 
-                       string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
+                       string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
                        outputNames.push_back(goodGroupFile);   outputTypes["group"].push_back(goodGroupFile);
                        
                        ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
@@ -618,7 +639,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
                vector<int> longHomoPolymer;
                
         vector<unsigned long long> positions;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                positions = m->divideFile(fastafile, processors);
                for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }   
 #else
@@ -626,6 +647,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
         else {
             int numFastaSeqs = 0;
             positions = m->setFilePosFasta(fastafile, numFastaSeqs); 
+            if (positions.size() < processors) { processors = positions.size(); }
             
             //figure out how many sequences you have to process
             int numSeqsPerProcessor = numFastaSeqs / processors;
@@ -645,7 +667,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
                        driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 #else
                int numSeqs = 0;
-               //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+               //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                        if(processors == 1){
                                numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
                        }else{
@@ -750,7 +772,7 @@ int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<in
                                count++;
                        }
                        //if((count) % 100 == 0){       m->mothurOut("Optimizing sequence: " + toString(count)); m->mothurOutEndLine();         }
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                unsigned long long pos = in.tellg();
                                if ((pos == -1) || (pos >= filePos.end)) { break; }
                        #else
@@ -776,7 +798,7 @@ int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition,
                int num = 0;
                vector<int> processIDS;
 
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -898,8 +920,8 @@ int ScreenSeqsCommand::screenGroupFile(set<string> badSeqNames){
                string seqName, group;
                set<string>::iterator it;
                
-               string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + "good" + m->getExtension(groupfile);
-               outputNames.push_back(goodGroupFile);  outputTypes["group"].push_back(goodGroupFile);
+               string goodGroupFile = outputDir + m->getRootName(m->getSimpleName(groupfile)) + getOutputFileNameTag("group", groupfile);
+        outputNames.push_back(goodGroupFile);  outputTypes["group"].push_back(goodGroupFile);
                ofstream goodGroupOut;  m->openOutputFile(goodGroupFile, goodGroupOut);
                
                while(!inputGroups.eof()){
@@ -950,7 +972,7 @@ int ScreenSeqsCommand::screenAlignReport(set<string> badSeqNames){
                string seqName, group;
                set<string>::iterator it;
                
-               string goodAlignReportFile = outputDir + m->getRootName(m->getSimpleName(alignreport)) + "good" + m->getExtension(alignreport);
+               string goodAlignReportFile = outputDir + m->getRootName(m->getSimpleName(alignreport)) + getOutputFileNameTag("alignreport");
                outputNames.push_back(goodAlignReportFile);  outputTypes["alignreport"].push_back(goodAlignReportFile);
                ofstream goodAlignReportOut;    m->openOutputFile(goodAlignReportFile, goodAlignReportOut);
 
@@ -1014,7 +1036,7 @@ int ScreenSeqsCommand::screenTaxonomy(set<string> badSeqNames){
                string seqName, tax;
                set<string>::iterator it;
                
-               string goodTaxFile = outputDir + m->getRootName(m->getSimpleName(taxonomy)) + "good" + m->getExtension(taxonomy);
+               string goodTaxFile = outputDir + m->getRootName(m->getSimpleName(taxonomy)) + getOutputFileNameTag("taxonomy", taxonomy);
                outputNames.push_back(goodTaxFile);  outputTypes["taxonomy"].push_back(goodTaxFile);
                ofstream goodTaxOut;    m->openOutputFile(goodTaxFile, goodTaxOut);
                                
@@ -1063,7 +1085,7 @@ int ScreenSeqsCommand::screenQual(set<string> badSeqNames){
                m->openInputFile(qualfile, in);
                set<string>::iterator it;
                
-               string goodQualFile = outputDir + m->getRootName(m->getSimpleName(qualfile)) + "good" + m->getExtension(qualfile);
+               string goodQualFile = outputDir + m->getRootName(m->getSimpleName(qualfile)) + getOutputFileNameTag("qfile", qualfile);
                outputNames.push_back(goodQualFile);  outputTypes["qfile"].push_back(goodQualFile);
                ofstream goodQual;      m->openOutputFile(goodQualFile, goodQual);
                
@@ -1170,7 +1192,7 @@ int ScreenSeqsCommand::driver(linePair filePos, string goodFName, string badAccn
                        count++;
                        }
                        
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                unsigned long long pos = inFASTA.tellg();
                                if ((pos == -1) || (pos >= filePos.end)) { break; }
                        #else
@@ -1282,7 +1304,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
         int process = 1;
                int num = 0;
 
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -1363,7 +1385,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
             if (i!=0) {extension += toString(i) + ".temp"; processIDS.push_back(i); }
             
                        // Allocate memory for thread data.
-                       sumScreenData* tempSum = new sumScreenData(startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, filename, m, lines[i].start, lines[i].end,goodFileName+extension, badAccnos+extension, &badSeqNames);
+                       sumScreenData* tempSum = new sumScreenData(startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, filename, m, lines[i].start, lines[i].end,goodFileName+extension, badAccnos+extension);
                        pDataArray.push_back(tempSum);
                        
                        //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
@@ -1380,6 +1402,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
                //Close all thread handles and free memory allocations.
                for(int i=0; i < pDataArray.size(); i++){
                        num += pDataArray[i]->count;
+            for (set<string>::iterator it = pDataArray[i]->badSeqNames.begin(); it != pDataArray[i]->badSeqNames.end(); it++) {        badSeqNames.insert(*it);       }
                        CloseHandle(hThreadArray[i]);
                        delete pDataArray[i];
                }
Mothur--16S microbiota sequencing software; mirror of main repository with my branches