Merge branch '1.40.0'

makefile-internal

@@ -1,20 +1,21 @@
 64BIT_VERSION ?= yes
-USEREADLINE ?= yes
-USEBOOST ?= yes
-RELEASE_DATE = "\"2/2/2017\""
+USEREADLINE ?= no
+USEBOOST ?= no
+RELEASE_DATE = "\"2/7/2017\""
 VERSION = "\"1.39.1\""
 
 # Optimize to level 3:
     CXXFLAGS += -O3
 
+
 ifeq  ($(strip $(64BIT_VERSION)),yes)
     #if you are a mac user use the following line
-    TARGET_ARCH += -arch x86_64
+    #TARGET_ARCH += -arch x86_64
 
     #if you using cygwin to build Windows the following line
-    #CXX = x86_64-w64-mingw32-g++
-    #CC = x86_64-w64-mingw32-g++
-    #TARGET_ARCH += -m64 -static
+    CXX = x86_64-w64-mingw32-g++
+    CC = x86_64-w64-mingw32-g++
+    TARGET_ARCH += -m64 -static
 
     #if you are a linux user use the following line - specific processor builds in future??
     #CXXFLAGS += -mtune=generic

source/classifier/phylotree.cpp

@@ -18,6 +18,7 @@ PhyloTree::PhyloTree(){
 		numSeqs = 0;
 		tree.push_back(TaxNode("Root"));
 		tree[0].heirarchyID = "0";
+        tree[0].level = 0;
 		maxLevel = 0;
 		calcTotals = true;
 		addSeqToTree("unknown", "unknown;");
@@ -78,6 +79,7 @@ PhyloTree::PhyloTree(string tfile){
 		numSeqs = 0;
 		tree.push_back(TaxNode("Root"));
 		tree[0].heirarchyID = "0";
+        tree[0].level = 0;
 		maxLevel = 0;
 		calcTotals = true;
 		string name, tax;
@@ -219,6 +221,7 @@ int PhyloTree::addSeqToTree(string seqName, string seqTaxonomy){
 				tree.push_back(TaxNode(taxon));
 				numNodes++;
 				tree[currentNode].children[taxon] = numNodes-1;
+                tree[currentNode].level = level;
 				tree[numNodes-1].parent = currentNode;
 				
 				currentNode = tree[currentNode].children[taxon];
@@ -296,7 +299,7 @@ void PhyloTree::assignHeirarchyIDs(int index){
                 
 			tree[it->second].heirarchyID = tree[index].heirarchyID + '.' + toString(counter);
 			counter++;
-			tree[it->second].level = tree[index].level + 1;
+			//tree[it->second].level = tree[index].level + 1;
 						
 			//save maxLevel for binning the unclassified seqs
 			if (tree[it->second].level > maxLevel) { maxLevel = tree[it->second].level; } 

source/commands/chimeravsearchcommand.cpp

@@ -1390,8 +1390,10 @@ int ChimeraVsearchCommand::driver(string outputFName, string filename, string ac
             
             string name = seq.getName();
             
-            name = name.substr(0, name.length()-1); //rip off last ;
-            name = name.substr(0, name.find_last_of(';'));
+            if (templatefile == "self") {
+                name = name.substr(0, name.length()-1); //rip off last ;
+                name = name.substr(0, name.find_last_of(';'));
+            }
             
             out << name << endl; numChimeras++;
         }

source/commands/makecontigscommand.cpp

@@ -825,6 +825,10 @@ unsigned long long MakeContigsCommand::processSingleFileOption(map<string, int>&
 //**********************************************************************************************************************
 unsigned long long MakeContigsCommand::processMultipleFileOption(map<string, int>& totalGroupCounts, vector<string>& theseOutputFileNames, map<string, string>& theseAllFileNames) {
     try {
+        
+        //read file
+        vector< vector<string> > fileInputs = readFileNames(file);  if (m->control_pressed) { return 0; }
+        
         unsigned long long numReads = 0;
         
         map<string, string> cvars;
@@ -861,10 +865,6 @@ unsigned long long MakeContigsCommand::processMultipleFileOption(map<string, int
         theseOutputFileNames.push_back(compositeQualFile);
         theseOutputFileNames.push_back(compositeMisMatchFile);
         
-        
-        //read file
-        vector< vector<string> > fileInputs = readFileNames(file);
-        
         if (gz) {
             numReads = createProcessesGroups(fileInputs, compositeGroupFile, compositeFastaFile, compositeScrapFastaFile, compositeQualFile, compositeScrapQualFile, compositeMisMatchFile, totalGroupCounts, theseAllFileNames); theseOutputFileNames.push_back(compositeGroupFile);
         }else {
@@ -2976,6 +2976,9 @@ vector< vector<string> > MakeContigsCommand::readFileNames(string filename){
             
             bool skip = false;
             string line = m->getline(in);  m->gobble(in);
+            
+            if (m->debug) {  m->mothurOut("[DEBUG]: " + line +"\n");  }
+            
             vector<string> pieces = m->splitWhiteSpace(line);
             
             string group = "";
@@ -3267,6 +3270,8 @@ vector< vector<string> > MakeContigsCommand::readFileNames(string filename){
             gz = true;
         }else { gz = false; }
         
+        if (files.size() == 0) { m->control_pressed = true; }
+        
         return files;
     }
     catch(exception& e) {

source/commands/pcrseqscommand.cpp

@@ -1161,16 +1161,14 @@ int PcrSeqsCommand::readGroup(set<string> names){
 		while(!in.eof()){
 			if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 			
-			in >> name;				//read from first column
-			in >> group;			//read from second column
+			in >> name;		m->gobble(in);		//read from first column
+			in >> group;	m->gobble(in);		//read from second column
 			
 			//if this name is in the accnos file
 			if (names.count(name) == 0) {
 				wroteSomething = true;
 				out << name << '\t' << group << endl;
 			}else {  removedCount++;  }
-            
-			m->gobble(in);
 		}
 		in.close();
 		out.close();

source/commands/preclustercommand.cpp

@@ -790,7 +790,8 @@ int PreClusterCommand::readFASTA(){
 					if (it == nameMap.end()) { m->mothurOut(seq.getName() + " is not in your names file, please correct."); m->mothurOutEndLine(); exit(1); }
 					else{
                         string second = it->second;
-						seqPNode tempNode(m->getNumNames(second), seq, second);
+                        int numReps = m->getNumNames(second);
+						seqPNode tempNode(numReps, seq, second);
 						alignSeqs.push_back(tempNode);
 						lengths.insert(seq.getAligned().length());
 					}	
@@ -810,6 +811,7 @@ int PreClusterCommand::readFASTA(){
         if (lengths.size() > 1) { method = "unaligned"; }
         else if (lengths.size() == 1) {  method = "aligned"; filterSeqs(); }
         
+        
 		return alignSeqs.size();
 	}
 	

source/commands/renamefilecommand.cpp

@@ -59,8 +59,8 @@ string RenameFileCommand::getHelpString(){
     try {
         string helpString = "";
         helpString += "The rename.file command allows you to rename files and updates the current files saved by mothur.\n";
-        helpString += "The rename.file command parameters are: phylip, column, list, rabund, sabund, name, group, design, tree, shared, relabund, fasta, qfile, sff, oligos, accnos, biom, count, summary, file, taxonomy, constaxonomy, input, output, prefix, deletedold and shorten.\n";
-        helpString += "The output parameter allows you to provide an output file name for the input file you provide.\n";
+        helpString += "The rename.file command parameters are: phylip, column, list, rabund, sabund, name, group, design, tree, shared, relabund, fasta, qfile, sff, oligos, accnos, biom, count, summary, file, taxonomy, constaxonomy, input, new, prefix, deletedold and shorten.\n";
+        helpString += "The new parameter allows you to provide an output file name for the input file you provide.\n";
         helpString += "The shorten parameter is used to inicate you want mothur to generate output file names for you. For example: stability.trim.contigs.good.unique.good.filter.unique.precluster.pick.pick.pick.an.unique_list.shared would become stability.an.shared. Default=true.";
         helpString += "The prefix parameter allows you to enter your own prefix for shortened names.";
         helpString += "The deleteold parameter indicates whether you want to delete the old file.  Default=true.";
@@ -479,6 +479,8 @@ RenameFileCommand::RenameFileCommand(string option)  {
             if ((mothurGenerated) && (outputfile != "") && (numFiles != 1)) {
                 m->mothurOut("[ERROR]: You must allow mothur to generate the filenames or input one file at a time with a new name, not both.\n"); abort= true;
             }
+            
+            if (outputDir != "") { outputfile = outputDir + m->getSimpleName(outputfile);  }
         }
         
     }

source/commands/seqsummarycommand.cpp

@@ -197,18 +197,20 @@ int SeqSummaryCommand::execute(){
 				
 		long long numSeqs = 0;
         long long size = 0;
+        long long numUniques = 0;
 		map<int, long long> startPosition;
 		map<int, long long> endPosition;
 		map<int, long long> seqLength;
 		map<int, long long> ambigBases;
 		map<int, long long> longHomoPolymer;
 		
-		if (namefile != "") { nameMap = m->readNames(namefile); }
+        if (namefile != "") { nameMap = m->readNames(namefile); numUniques = nameMap.size(); }
         else if (countfile != "") {
             CountTable ct;
             ct.readTable(countfile, false, false);
             nameMap = ct.getNameMap();
             size = ct.getNumSeqs();
+            numUniques = ct.getNumUniqueSeqs();
         }
 		
 		if (m->control_pressed) { return 0; }
@@ -244,9 +246,19 @@ int SeqSummaryCommand::execute(){
         
         //set size
         if (countfile != "") {}//already set
-        else if (namefile == "") { size = numSeqs;  }
+        else if (namefile == "") { size = numSeqs; }
         else { for (map<int, long long>::iterator it = startPosition.begin(); it != startPosition.end(); it++) { size += it->second; } }
         
+        if ((namefile != "") || (countfile != "")) {
+            string type = "count";
+            if (namefile != "") { type = "name"; }
+            if (numSeqs != numUniques) { // do fasta and name/count files match
+                m->mothurOut("[ERROR]: Your " + type + " file contains " + toString(numUniques) + " unique sequences, but your fasta file contains " + toString(numSeqs) + ". File mismatch detected, quitting command.\n"); m->control_pressed = true;
+            }
+        }
+        
+        if (m->control_pressed) {  m->mothurRemove(summaryFile); return 0; }
+        
         long long ptile0_25	= 1+(long long)(size * 0.025); //number of sequences at 2.5%
         long long ptile25		= 1+(long long)(size * 0.250); //number of sequences at 25%
         long long ptile50		= 1+(long long)(size * 0.500);

source/mothurout.cpp

@@ -309,14 +309,16 @@ void MothurOut::setFileName(string filename)  {
 void MothurOut::setDefaultPath(string pathname)  {
 	try {
 	
-		//add / to name if needed
-		string lastChar = pathname.substr(pathname.length()-1);
-		#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
-			if (lastChar != "/") { pathname += "/"; }
-		#else
-			if (lastChar != "\\") { pathname += "\\"; }	
-		#endif
-		
+        if (pathname != "") {
+            //add / to name if needed
+            string lastChar = pathname.substr(pathname.length()-1);
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+            if (lastChar != "/") { pathname += "/"; }
+#else
+            if (lastChar != "\\") { pathname += "\\"; }
+#endif
+        }
+        
 		defaultPath = getFullPathName(pathname);
 		
 	}
@@ -329,14 +331,15 @@ void MothurOut::setDefaultPath(string pathname)  {
 void MothurOut::setBlastPath(string pathname)  {
     try {
         
-        //add / to name if needed
-        string lastChar = pathname.substr(pathname.length()-1);
+        if (pathname != "") {
+            //add / to name if needed
+            string lastChar = pathname.substr(pathname.length()-1);
 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
-        if (lastChar != "/") { pathname += "/"; }
+            if (lastChar != "/") { pathname += "/"; }
 #else
-        if (lastChar != "\\") { pathname += "\\"; }
+            if (lastChar != "\\") { pathname += "\\"; }
 #endif
-        
+        }
         blastPath = getFullPathName(pathname);
         
     }