Merge branch 'master' of https://github.com/fbreitwieser/kraken-hll

* 'master' of https://github.com/fbreitwieser/kraken-hll:
  Fix gzstream building and licensing
  Fix gzstream compilation, update on HLL

Conflicts:
	src/hyperloglogplus.h

scripts/krakenhll

@@ -1,6 +1,7 @@
 #!/usr/bin/env perl
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

scripts/krakenhll-build

@@ -1,6 +1,7 @@
 #!/usr/bin/env perl
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

scripts/krakenhll-build_db.sh

@@ -1,7 +1,8 @@
 #!/bin/bash
 #vim: noai:ts=2:sw=2
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

scripts/krakenhll-check_for_jellyfish.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
-# modified by Florian Breitwieser, 2017
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

scripts/krakenhll-download

@@ -3,6 +3,7 @@
 
 # krakenhll-download.pl - based on centrifuge-download
 # (c) Florian Breitwieser, 2017
+# licensed under GPL-3
 
 use strict;
 use warnings;

scripts/krakenhll-report

@@ -1,6 +1,7 @@
 #!/usr/bin/env perl
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

scripts/krakenhll-standard_installation.sh

@@ -1,6 +1,7 @@
 #!/bin/bash
 
-# Copyright 2013-2015, Derrick Wood <[email protected]>
+# Original file Copyright 2013-2015, Derrick Wood <[email protected]>
+# Portions (c) 2017, Florian Breitwieser <[email protected]> as part of KrakenHLL
 #
 # This file is part of the Kraken taxonomic sequence classification system.
 #

src/Makefile

@@ -1,9 +1,10 @@
 CXX = g++
 FOPENMP?=-fopenmp
-CXXFLAGS = -Wall -std=c++0x $(FOPENMP) -O2 -Wfatal-errors ${CPPFLAGS}
+CXXFLAGS = -Wall -std=c++0x $(FOPENMP) -I./gzstream -O2 -Wfatal-errors ${CPPFLAGS}
 #CXXFLAGS = -Wall -std=c++11 $(FOPENMP) -O3 -Wfatal-errors
 PROGS = classify db_sort set_lcas make_seqid_to_taxid_map db_shrink build_taxdb grade_classification dump_taxdb read_uid_mapping
-LIBFLAGS = -L. -I./gzstream -L./gzstream -lz -lgzstream ${LDFLAGS}
+#LIBFLAGS = -L. -lz -lgzstream ${LDFLAGS}
+LIBFLAGS = -L. -lz ${LDFLAGS}
 
 .PHONY: all install clean
 
@@ -13,7 +14,7 @@ install: $(PROGS)
 	cp $(PROGS) $(KRAKEN_DIR)/
 
 clean:
-	rm -f $(PROGS) *.o
+	rm -rf $(PROGS) *.o *.dSYM
 
 db_shrink: krakendb.o quickfile.o
 
@@ -25,7 +26,7 @@ grade_classification: taxdb.h report-cols.h
 
 read_uid_mapping: quickfile.o
 
-classify: classify.cpp krakendb.o quickfile.o krakenutil.o seqreader.o uid_mapping.o hyperloglogplus.h taxdb.h report-cols.h
+classify: classify.cpp krakendb.o quickfile.o krakenutil.o seqreader.o uid_mapping.o gzstream.o hyperloglogplus.h taxdb.h report-cols.h
 	$(CXX) $(CXXFLAGS) -o classify $^ $(LIBFLAGS)
 
 build_taxdb: taxdb.h report-cols.h quickfile.o
@@ -43,6 +44,9 @@ krakendb.o: krakendb.cpp krakendb.hpp quickfile.hpp
 seqreader.o: seqreader.cpp seqreader.hpp quickfile.hpp
 	$(CXX) $(CXXFLAGS) -c seqreader.cpp
 
+gzstream.o: gzstream/gzstream.C gzstream/gzstream.h
+	$(CXX) $(CXXFLAGS) -c -O gzstream/gzstream.C
+
 quickfile.o: quickfile.cpp quickfile.hpp
 	$(CXX) $(CXXFLAGS) -c quickfile.cpp
 

src/classify.cpp

@@ -115,19 +115,19 @@ ostream* cout_or_file(string file) {
 }
 
 void loadKrakenDB(KrakenDB& database, string DB_filename, string Index_filename) {
-	QuickFile db_file;
-	db_file.open_file(DB_filename);
-	if (Populate_memory) {
-		db_file.load_file();
-	}
-	database = KrakenDB(db_file.ptr());
-	QuickFile idx_file;
-	idx_file.open_file(Index_filename);
-	if (Populate_memory)
-		idx_file.load_file();
-
-	KrakenDBIndex db_index(idx_file.ptr());
-	database.set_index(&db_index);
+  QuickFile db_file;
+  db_file.open_file(DB_filename);
+  if (Populate_memory) {
+    db_file.load_file();
+  }
+  database = KrakenDB(db_file.ptr());
+  QuickFile idx_file;
+  idx_file.open_file(Index_filename);
+  if (Populate_memory)
+    idx_file.load_file();
+
+  KrakenDBIndex db_index(idx_file.ptr());
+  database.set_index(&db_index);
 }
 
 int main(int argc, char **argv) {
@@ -243,18 +243,18 @@ int main(int argc, char **argv) {
       taxdb.readGenomeSizes(fname);
     }
 
-	TaxReport<uint32_t,ReadCounts> rep = TaxReport<uint32_t, ReadCounts>(*Report_output, taxdb, taxon_counts, false);
-	rep.setReportCols(vector<string> { 
-		"%",
-		"reads", 
+  TaxReport<uint32_t,ReadCounts> rep = TaxReport<uint32_t, ReadCounts>(*Report_output, taxdb, taxon_counts, false);
+  rep.setReportCols(vector<string> { 
+    "%",
+    "reads", 
     "taxReads",
     "kmers",
     "dup",
     "cov", 
-		"taxID", 
-		"rank", 
-		"taxName"});
-	rep.printReport("kraken","blu");
+    "taxID", 
+    "rank", 
+    "taxName"});
+  rep.printReport("kraken","blu");
   }
 
   for (size_t i = 0; i < Open_fstreams.size(); ++i) {
@@ -368,11 +368,11 @@ void process_file(char *filename) {
 
 inline
 uint32_t get_taxon_for_kmer(KrakenDB& database, uint64_t* kmer_ptr, uint64_t& current_bin_key,
-		int64_t& current_min_pos, int64_t& current_max_pos) {
-	uint32_t* val_ptr = database.kmer_query(
-			database.canonical_representation(*kmer_ptr), &current_bin_key,
-			&current_min_pos, &current_max_pos);
-	return val_ptr ? *val_ptr : 0;
+    int64_t& current_min_pos, int64_t& current_max_pos) {
+  uint32_t* val_ptr = database.kmer_query(
+      database.canonical_representation(*kmer_ptr), &current_bin_key,
+      &current_min_pos, &current_max_pos);
+  return val_ptr ? *val_ptr : 0;
 }
 
 
@@ -513,10 +513,10 @@ bool classify_sequence(DNASequence &dna, ostringstream &koss,
         taxon = get_taxon_for_kmer(*KrakenDatabases[i], kmer_ptr,
         db_statuses[i].current_bin_key, db_statuses[i].current_min_pos, db_statuses[i].current_max_pos);
 
-	  //uint32_t* val_ptr = KrakenDatabases[i]->kmer_query(
-		//	KrakenDatabases[i]->canonical_representation(*kmer_ptr), &db_statuses[i].current_bin_key,
-		//	&db_statuses[i].current_min_pos, &db_statuses[i].current_max_pos);
-	  //taxon = val_ptr ? *val_ptr : 0;
+        //uint32_t* val_ptr = KrakenDatabases[i]->kmer_query(
+        //  KrakenDatabases[i]->canonical_representation(*kmer_ptr), &db_statuses[i].current_bin_key,
+        //  &db_statuses[i].current_min_pos, &db_statuses[i].current_max_pos);
+        //taxon = val_ptr ? *val_ptr : 0;
           if (taxon) break;
         }
 

src/gzstream/Makefile

@@ -33,7 +33,7 @@
 # ----------------------------------------------------------------------------
 
 # CXX      = CC -n32 -LANG:std   # for SGI Irix 6.5, MIPSpro CC version 7.30
-CXX      = g++-7   # for Linux RedHat 6.1, g++ version 2.95.2
+CXX      = g++   # for Linux RedHat 6.1, g++ version 2.95.2
 
 CPPFLAGS = -I. -O -fPIC
 LDFLAGS  = -L. -lgzstream -lz