X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=chimeraslayercommand.cpp;h=e7dc92e5919d1f5f7078fd5a11171cbb39030861;hp=d8e0952a8b9f89cd237946a27ea624f11fa42898;hb=b206f634aae1b4ce13978d203247fb64757d5482;hpb=859e3a473a3e63e0060c49be70b80f9289253da2

diff --git a/chimeraslayercommand.cpp b/chimeraslayercommand.cpp
index d8e0952..e7dc92e 100644
--- a/chimeraslayercommand.cpp
+++ b/chimeraslayercommand.cpp
@@ -603,10 +603,7 @@ ChimeraSlayerCommand::ChimeraSlayerCommand(string option)  {
 			m->mothurConvert(temp, numwanted);
             
 			temp = validParameter.validFile(parameters, "dereplicate", false);	
-			if (temp == "not found") { 
-				if (groupfile != "")    {  temp = "false";					}
-				else                    {  temp = "true"; 	}
-			}
+			if (temp == "not found") { temp = "false";			}
 			dups = m->isTrue(temp);
 			
 			blastlocation = validParameter.validFile(parameters, "blastlocation", false);	
@@ -762,7 +759,7 @@ int ChimeraSlayerCommand::execute(){
 				if (processors == 1) {
                     numSeqs = driverGroups(outputFileName, accnosFileName, trimFastaFileName, fileToPriority, fileGroup, newCountFile);
                     if (hasCount && dups) {
-                        CountTable c; c.readTable(nameFileNames[s]);
+                        CountTable c; c.readTable(nameFileNames[s], true, false);
                         if (!m->isBlank(newCountFile)) {
                             ifstream in2;
                             m->openInputFile(newCountFile, in2);
@@ -794,7 +791,7 @@ int ChimeraSlayerCommand::execute(){
                     }else {
                         if (hasCount) {
                             set<string> doNotRemove;
-                            CountTable c; c.readTable(newCountFile);
+                            CountTable c; c.readTable(newCountFile, true, true);
                             vector<string> namesInTable = c.getNamesOfSeqs();
                             for (int i = 0; i < namesInTable.size(); i++) {
                                 int temp = c.getNumSeqs(namesInTable[i]);
@@ -837,6 +834,11 @@ int ChimeraSlayerCommand::execute(){
 			}
 		}
 		
+        itTypes = outputTypes.find("count");
+		if (itTypes != outputTypes.end()) {
+			if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setCountTableFile(current); }
+		}
+        
 		m->mothurOutEndLine();
 		m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 		for (int i = 0; i < outputNames.size(); i++) {	m->mothurOut(outputNames[i]); m->mothurOutEndLine();	}	
@@ -866,8 +868,8 @@ int ChimeraSlayerCommand::MPIExecuteGroups(string outputFileName, string accnosF
 		map<string, map<string, int> >::iterator itFile;
 		vector<string> filenames;
 		for(itFile = fileToPriority.begin(); itFile != fileToPriority.end(); itFile++) { filenames.push_back(itFile->first); }
-		
-		int numGroupsPerProcessor = filenames.size() / processors;
+        
+        int numGroupsPerProcessor = ceil(filenames.size() / (double) processors);
 		int startIndex =  pid * numGroupsPerProcessor;
 		int endIndex = (pid+1) * numGroupsPerProcessor;
 		if(pid == (processors - 1)){	endIndex = filenames.size(); 	}
@@ -1549,7 +1551,7 @@ int ChimeraSlayerCommand::createProcessesGroups(string outputFName, string accno
 		if (fileToPriority.size() < processors) { processors = fileToPriority.size(); }
         
         CountTable newCount;
-        if (hasCount && dups) { newCount.readTable(countFile); }
+        if (hasCount && dups) { newCount.readTable(countFile, true, false); }
 		
 		int groupsPerProcessor = fileToPriority.size() / processors;
 		int remainder = fileToPriority.size() % processors;
@@ -1626,7 +1628,7 @@ int ChimeraSlayerCommand::createProcessesGroups(string outputFName, string accno
 		//Create processor worker threads.
 		for(int i=1; i<processors; i++ ){
 			string extension = toString(i) + ".temp";
-			slayerData* tempslayer = new slayerData(group2NameMap, hasCount, dups, (accnos + extension+".byCount"), (outputFName + extension), (fasta + extension), (accnos + extension), templatefile, search, blastlocation, trimera, trim, realign, m, breakUp[i], fileGroup, ksize, match, mismatch, window, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, divR, priority, i);
+			slayerData* tempslayer = new slayerData(group2NameMap, hasCount, dups, (accnos + toString(i) +".byCount"), (outputFName + extension), (fasta + extension), (accnos + extension), templatefile, search, blastlocation, trimera, trim, realign, m, breakUp[i], fileGroup, ksize, match, mismatch, window, minSimilarity, minCoverage, minBS, minSNP, parents, iters, increment, numwanted, divR, priority, i);
 			pDataArray.push_back(tempslayer);
 			processIDS.push_back(i);
 			
@@ -1821,10 +1823,10 @@ int ChimeraSlayerCommand::driver(linePair filePos, string outputFName, string fi
 			
 			delete candidateSeq;
 			//report progress
-			if((count) % 100 == 0){	m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();		}
+			if((count) % 100 == 0){	m->mothurOutJustToScreen("Processing sequence: " + toString(count) + "\n");		}
 		}
 		//report progress
-		if((count) % 100 != 0){	m->mothurOut("Processing sequence: " + toString(count)); m->mothurOutEndLine();		}
+		if((count) % 100 != 0){	m->mothurOutJustToScreen("Processing sequence: " + toString(count)+ "\n"); 		}
 		
 		int numNoParents = chimera->getNumNoParents();
 		if (numNoParents == count) { m->mothurOut("[WARNING]: megablast returned 0 potential parents for all your sequences. This could be due to formatdb.exe not being setup properly, please check formatdb.log for errors."); m->mothurOutEndLine(); } 
@@ -1971,10 +1973,10 @@ int ChimeraSlayerCommand::driverMPI(int start, int num, MPI_File& inMPI, MPI_Fil
 			delete candidateSeq;
 			
 			//report progress
-			if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;	m->mothurOutJustToLog("Processing sequence: " + toString(i+1) + "\n");		}
+			if((i+1) % 100 == 0){  cout << "Processing sequence: " << (i+1) << endl;		}
 		}
 		//report progress
-		if(num % 100 != 0){		cout << "Processing sequence: " << num << endl;	m->mothurOutJustToLog("Processing sequence: " + toString(num) + "\n"); 	}
+		if(num % 100 != 0){		cout << "Processing sequence: " << num << endl;		}
 		
 		int numNoParents = chimera->getNumNoParents();
 		if (numNoParents == num) { cout << "[WARNING]: megablast returned 0 potential parents for all your sequences. This could be due to formatdb.exe not being setup properly, please check formatdb.log for errors." << endl; }
@@ -2159,7 +2161,7 @@ map<string, int> ChimeraSlayerCommand::sortFastaFile(string fastaFile, string na
         int error;
         if (hasCount) { 
             CountTable ct;
-            ct.readTable(nameFile);
+            ct.readTable(nameFile, true, false);
             
             for(map<string, string>::iterator it = seqs.begin(); it != seqs.end(); it++) {
                 int num = ct.getNumSeqs(it->first);