enable multi-threaded bam read for lumpy_filter

also run lumpy_tests in travis

closes arq5x#41

.gitignore

@@ -5,3 +5,10 @@ ssshtest
 data/
 data.tar.gz
 src/utils/BamTools
+lumpy_tests/*/*.bam
+lumpy_tests/*/*.bai
+lumpy_tests/*/*.histo
+lumpy_tests/*/*.stats
+lumpy_tests/*/*.vcf
+lumpy_tests/*/*.bedpe
+src/filter/lumpy_filter

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "scripts/bamkit"]
 	path = scripts/bamkit
 	url = https://github.com/cc2qe/bamkit.git
+[submodule "lib/htslib"]
+	path = lib/htslib
+	url = https://github.com/samtools/htslib.git

.travis.yml

@@ -6,9 +6,11 @@ script:
     - ./scripts/lumpyexpress -h
     - python scripts/cnvanator_to_bedpes.py --cnvkit  -b 100 --del_o delo2 --dup_o dupo2 -c data/example.cns
     - bash test/functional-test.sh
+    - cd lumpy_tests && bash test.sh
 
 install:
     - sudo apt-get update
+    - sudo apt-get install libssl-dev gettext gettext-base libtool autoconf
     - unset PYTHONPATH
     - unset PYTHONHOME
     - unset PYTHONUSER
@@ -25,4 +27,5 @@ install:
     - conda config --add channels bioconda
     - conda create -q -n travis python=$TRAVIS_PYTHON_VERSION pysam samtools htslib bedtools
     - source activate travis
+    - autoconf --version
     - make

Makefile

@@ -41,7 +41,7 @@ UTIL_SUBDIRS =	$(SRC_DIR)/utils/bedFile \
 				$(SRC_DIR)/utils/sqlite3
 
 
-all:	lumpyexpress lumpy lumpy_filter
+all:	lumpy_filter lumpyexpress lumpy
 
 lumpy:	
 	[ -d $(OBJ_DIR) ] || mkdir -p $(OBJ_DIR)
@@ -65,11 +65,16 @@ lumpy:
 	done
 
 lumpy_filter: htslib
+	[ -d $(BIN_DIR) ] || mkdir -p $(BIN_DIR)
 	$(MAKE) --no-print-directory -C src/filter/
 	cp src/filter/lumpy_filter $(BIN_DIR)
 
+
 htslib:
-	$(MAKE) --no-print-directory -C lib/htslib
+	$(shell cd lib/htslib && autoreconf)
+	cd lib/htslib && \
+	./configure --disable-bz2 --disable-lzma --enable-libcurl
+	CFLAGS="$(CFLAGS) -DBGZF" $(MAKE) -C lib/htslib --no-print-directory CFLAGS="-DBGZF_MT"
 
 lumpyexpress:
 	[ -d $(BIN_DIR) ] || mkdir -p $(BIN_DIR)

src/filter/Makefile

@@ -1,4 +1,4 @@
 
 all:
-	$(CC) -I../../lib/htslib/ -o lumpy_filter filter.c ../../lib/htslib/libhts.a -lpthread -lz
+	$(CC) -I../../lib/htslib/ -o lumpy_filter filter.c ../../lib/htslib/libhts.a -lcrypto -lcurl -lpthread -lz
 

src/filter/filter.c

@@ -11,6 +11,7 @@
 #include <stdbool.h>
 
 #include "htslib/sam.h"
+#include "htslib/bgzf.h"
 
 #define MIN(X, Y) (((X) < (Y)) ? (X) : (Y))
 #define MAX(X, Y) (((X) > (Y)) ? (X) : (Y))
@@ -186,14 +187,18 @@ int count_tags(char *sa_tag)
 
 int main(int argc, char **argv)
 {
-    if (argc != 4)
+    if (argc < 4)
         errx(1,
-             "usage\t:%s <bam> <split out> <discord out>",
+             "usage\t:%s <bam> <split out> <discord out> (optional #threads)",
              argv[0]);
 
     char *bam_file_name = argv[1];
     char *split_file_name = argv[2];
     char *disc_file_name = argv[3];
+    int threads = 2;
+    if (argc == 5) {
+        threads = atoi(argv[4]);
+    }
 
     samFile *disc = sam_open(disc_file_name, "wb");
 
@@ -203,6 +208,11 @@ int main(int argc, char **argv)
     if(in == NULL)
         errx(1, "Unable to open BAM/SAM file.");
 
+    // TODO: handle cram.
+    if (threads > 1) {
+        bgzf_mt(in->fp.bgzf, threads, 256);
+    }
+
     hts_idx_t *idx = sam_index_load(in, bam_file_name);
     if(idx == NULL)
         errx(1,"Unable to open BAM/SAM index.");