Fixes some problems related to FASTA header management

Makefile

@@ -11,12 +11,12 @@ INCDIR = .
 ################################# for bloom filter code #################################
 CXX=g++
 # directory of sdsl library.so
-SDSL_PATH=$(PREFIX)/lib
+SDSL_PATH=./sdsl-lite/COMPILED/lib
 # also requires head file of roaring, sdsl, jellyfish, tclap in ~/include
-CXXFLAGS= -std=c++11 -Wall -O3 -I$(PREFIX)/include
+CXXFLAGS= -std=c++11 -Wall -O3 -I./sdsl-lite/COMPILED/include
 
 LD_LIB=-L $(SDSL_PATH)
-LD_FLAG=-lsdsl -ldivsufsort -ldivsufsort64 
+LD_FLAG=-lsdsl -ldivsufsort -ldivsufsort64
 
 ################################ end bloom filter def ##################################
 

install.sh

@@ -1,4 +1,5 @@
 #!/bin/bash
 chmod +x ./sdsl-lite/install.sh ./sdsl-lite/build/*.sh
-./sdsl-lite/install.sh $PREFIX
+mkdir -p ./sdsl-lite/COMPILED
+./sdsl-lite/install.sh ./sdsl-lite/COMPILED
 make all

src/generate_bf.cc

@@ -44,7 +44,13 @@ int BFGenerator::readFasta(string & fasta_filename) {
 				genome_vector.emplace_back(genome);
 			}
 
-			id = line.substr(1);
+			// This is necessary to correctly manage headers containing additional information
+			id = split(line.substr(1), ' ')[0];
+
+			// This is necessary because the chromosome id from VCF is forced to start with chr
+			if(id.compare(0, 3, "chr") != 0)
+			  id = "chr" + id;
+
 			DNA_sequence.clear();
 		}
 		else {//  if (line[0] != '>'){ // not needed because implicit

src/qv.cc

@@ -2342,7 +2342,11 @@ int main(const int argc, const char *argv[])
 	#undef CHRLENS_EXT
 
 			for (size_t i = 0; i < ref.size; i++) {
-				fprintf(chrlens, "%s %lu\n", ref.seqs[i].name, ref.seqs[i].size);
+				// This is necessary because the chromosome id from VCF is forced to start with chr
+				std::string seq_name (ref.seqs[i].name);
+				if(seq_name.compare(0, 3, "chr") != 0)
+					seq_name = "chr" + seq_name;
+				fprintf(chrlens, "%s %lu\n", seq_name.c_str(), ref.seqs[i].size);
 			}
 
 			fclose(chrlens);