Merge pull request #250 from diekhans/markd-species-errmsg

Fixed checking for all species being remove

src/Makefile

@@ -14,14 +14,19 @@ CXX?=g++
 # Notes: - "-Wno-sign-compare" eliminates a high number of warnings (see footnote below). Please adopt
 #          a strict signed-only usage strategy to avoid mistakes since we are not warned about this.
 #        - The order of object files in $(OBJS) IS IMPORTANT (see lldouble.hh)
-CXXFLAGS := -Wall -Wno-sign-compare -pedantic -O3 -std=c++11 ${CXXFLAGS}#-DDEBUG -g -ggdb -pg -DDEBUG_STATES
+# for debugging set
+#    OPT = -g
+#    CXXFLAGS = -DDEBUG -g -ggdb -pg -DDEBUG_STATES
+#  in common.mk or the environment
+OPT ?=  -O3
+CXXFLAGS += -Wall -Wno-sign-compare -pedantic ${OPT} -std=c++11
 
 ifeq (,$(findstring $(ZIPINPUT),0 false False FALSE))  # if ZIPINPUT is not defined or is something else than 0, false, False or FALSE
 	CPPFLAGS += -DZIPINPUT
-	LIBS    = -lboost_iostreams
+	LIBS    += -lboost_iostreams
 endif
 
-INCLS	= -I../include
+INCLS	+= -I../include
 
 OBJS	= genbank.o properties.o pp_profile.o pp_hitseq.o pp_scoring.o statemodel.o namgene.o \
  types.o gene.o evaluation.o motif.o geneticcode.o hints.o extrinsicinfo.o projectio.o \
@@ -77,43 +82,43 @@ all: $(OBJS) $(TOBJS) $(DUMOBJS) $(PROGR) info
 
 augustus: augustus.cc $(OBJS) $(DUMOBJS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp augustus ../bin/
+	cp -f augustus ../bin/
 
 etraining: etraining.cc $(TOBJS) $(OBJS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp etraining ../bin/
+	cp -f etraining ../bin/
 
 evaluate: evaluate.cc $(OBJS) $(DUMOBJS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
 
 fastBlockSearch: fastBlockSearch.cc pp_fastBlockSearcher.o \
 	types.o properties.o geneticcode.o pp_profile.o lldouble.o
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp fastBlockSearch ../bin/
+	cp -f fastBlockSearch ../bin/
 
 pp_simScore: pp_simscore.cc fasta.o types.o properties.o geneticcode.o pp_profile.o lldouble.o
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp pp_simScore ../bin/
+	cp -f pp_simScore ../bin/
 
 prepareAlign: pp_prepare_align.cc
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^
-	cp prepareAlign ../bin/
+	cp -f prepareAlign ../bin/
 
 load2db: load2db.cc fasta.o hints.o types.o properties.o lldouble.o
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp load2db ../bin/
+	cp -f load2db ../bin/
 
 load2sqlitedb: load2sqlitedb.cc fasta.o hints.o types.o properties.o lldouble.o sqliteDB.o
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp load2sqlitedb ../bin/
+	cp -f load2sqlitedb ../bin/
 
 getSeq: getSeq.cc $(OBJS) $(DUMOBJS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp getSeq ../bin/
+	cp -f getSeq ../bin/
 
 espoca: espoca.cc $(OBJS) $(DUMOBJS)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) $(LDFLAGS) -o $@ $^ $(INCLS) $(LIBS)
-	cp espoca ../bin/
+	cp -f espoca ../bin/
 
 info:
 	echo "$(CXXFLAGS) $(CPPFLAGS)" > $(INFO)

src/compgenepred.cc

@@ -79,7 +79,9 @@ CompGenePred::CompGenePred() : tree(Constant::treefile) {
     for(int j=0; j<speciesNames.size(); j++){
       cout<<"species "<<j<<"\t"<<speciesNames[j]<<endl;
     }
-
+    if (speciesNames.size() < 2) {
+        throw ProjectError("must have at least two species for CompGenePred");
+    }
 
     if (Properties::hasProperty("trainFeatureFile")) {
       trainFeature_from_file();

src/phylotree.cc

@@ -107,8 +107,8 @@ PhyloTree::PhyloTree(string filename){
 	    ifstream ifstrm(only_species.c_str());
 	    if (ifstrm.is_open()){
 		vector<string> keep; // the subset of species to be kept
-		char buf[256];
-		while(ifstrm.getline(buf,255)){
+		char buf[8192];
+		while(ifstrm.getline(buf, sizeof(buf)-1)){
 		    stringstream stm(buf);
 		    string s;
 		    if(stm >> s)
@@ -123,13 +123,16 @@ PhyloTree::PhyloTree(string filename){
 			    break;
 			}
 		    }
-		    if(!found){ //if species name ist not in list, remove leaf
+		    if(!found){ //if species name is not in list, remove leaf
 			drop(species[i]);
 		    }
 		}
-		if(species.size() < 2 || species.size() < keep.size())
-		    throw ProjectError(only_species + " has the wrong format. correct format:\n\n" + 
-				       "hg19\n" + "mm9\n" + "galGal3\n" + "...\n" + 
+		if(numSp < 2 || numSp < keep.size())
+		    throw ProjectError(only_species + " has the wrong format or doesn't match tree. Correct format:\n\n" + 
+				       "hg19	/dir/to/genome/human.fa\n" +
+                                       "mm9	/dir/to/genome/mouse.fa\n" +
+                                       "galGal3	/dir/to/genome/chicken.fa\n" +
+                                       "...\n" + 
 				       "The species identifiers must be the same as in the phylogenetic tree. At least two species identifiers must be given.");
 	    }
 	    else

src/randseqaccess.cc

@@ -98,7 +98,8 @@ void SpeciesCollection::readExtrinsicCFGFile(vector<string> &speciesNames){
 			    }
 			    else{
 				if(find(speciesNames.begin(), speciesNames.end(), skey) == speciesNames.end()) {
-				    throw ProjectError(skey + " is not a valid species identifier.");
+				    throw ProjectError("species identifier '" + skey + "' referenced in extrinsicCfgFile '"
+                                                       +  filename + "' not found in defined set of species");
 				}
 				addSpeciesToGroup(skey,groupCount);
 				cout <<" "<<skey;