def run(self, args):
if len(args.input) < 2:
logging.error("You have to specify at least 2 files")
files = args.input
output = args.output
if output is None:
output = sys.stdout
gem.merge(files[0], files[1:], output,
threads=int(args.threads), same_content=args.same,
paired=args.paired, compress=args.compress)
def test_file_merge_pairwise_same():
reads_1 = files.open(testfiles["test.map"])
reads_2 = files.open(testfiles["test.map"])
merged = gem.merge(reads_1, [reads_2],
output=results_dir + "/merge_result.map",
threads=8, paired=True)
num_reads = sum(1 for r in merged)
assert num_reads == 10
def test_file_merge_pairwise_same_content_big_sub():
reads_1 = files.open(testfiles["20t.map.gz"])
reads_2 = files.open(testfiles["20t_sub.map.gz"])
merged = gem.merge(reads_1, [reads_2],
output=results_dir + "/merge_result.map",
threads=8, paired=True)
num_reads = sum(1 for r in merged)
assert num_reads == 20000
def test_file_merge_pairwise_same_content_big_uncompressed():
subprocess.call("cp %s %s; gunzip %s;" % (testfiles["20t.map.gz"], results_dir, results_dir + "/20t.map.gz"), shell=True)
reads_1 = files.open(results_dir + "/20t.map")
reads_2 = files.open(results_dir + "/20t.map")
merged = gem.merge(reads_1, [reads_2],
output=results_dir + "/merge_result.map",
threads=8, same_content=True)
num_reads = sum(1 for r in merged)
assert num_reads == 20000
def test_merging_maps_chr21():
s = [
"chr21_mapping_initial.map", "chr21_mapping_denovo.map",
"chr21_mapping_initial_split.map", "chr21_mapping_trim_20.map",
"chr21_mapping_trim_20_split.map"
]
fs = []
for f in s:
fs.append(files.open(testfiles[f]))
m = gem.merge(fs[0], fs[1:])
count = 0
ms = 0
for read in m:
count += 1
ms += read.num_maps
assert count == 2000, count
assert ms == 3590, ms
def test_merging_maps_chr21():
s = ["chr21_mapping_initial.map",
"chr21_mapping_denovo.map",
"chr21_mapping_initial_split.map",
"chr21_mapping_trim_20.map",
"chr21_mapping_trim_20_split.map"
]
fs = []
for f in s:
fs.append(files.open(testfiles[f]))
m = gem.merge(fs[0], fs[1:])
count = 0
ms = 0
for read in m:
count += 1
ms += read.num_maps
assert count == 2000, count
assert ms == 3590, ms
def test_merging_maps():
input = files.open(testfiles["test_merge_target.map"])
source_1 = files.open(testfiles["test_merge_source_1.map"])
source_2 = files.open(testfiles["test_merge_source_2.map"])
#result = gem.mapper(input, index, results_dir + "/merged.mapping")
merger = gem.merge(input, [source_1, source_2])
count = 0
for read in merger:
count += 1
if read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13866:124450":
assert read.to_map().split("\t")[3] == "0+1", read.to_map()
assert read.to_map().split("\t")[4] == "chr2:-:162359617:74G"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13753:124452":
assert read.to_map().split("\t")[3] == "0+1", read.to_map()
assert read.to_map().split("\t")[4] == "chr9:+:38397301:54G20"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:14211:124259":
assert read.to_map().split("\t")[3] == "1", read.to_map()
assert read.to_map().split("\t")[4] == "chr15:+:72492866:75"
assert count == 10
def test_merging_maps():
input = files.open(testfiles["test_merge_target.map"])
source_1 = files.open(testfiles["test_merge_source_1.map"])
source_2 = files.open(testfiles["test_merge_source_2.map"])
#result = gem.mapper(input, index, results_dir + "/merged.mapping")
merger = gem.merge(input, [source_1, source_2])
count = 0
for read in merger:
count += 1
if read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13866:124450":
assert read.to_map().split("\t")[3] == "0+1", read.to_map()
assert read.to_map().split("\t")[4] == "chr2:-:162359617:74G"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13753:124452":
assert read.to_map().split("\t")[3] == "0+1", read.to_map()
assert read.to_map().split("\t")[4] == "chr9:+:38397301:54G20"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:14211:124259":
assert read.to_map().split("\t")[3] == "1", read.to_map()
assert read.to_map().split("\t")[4] == "chr15:+:72492866:75"
assert count == 10
def test_merging_maps_to_file():
input = files.open(testfiles["test_merge_target.map"])
source_1 = files.open(testfiles["test_merge_source_1.map"])
source_2 = files.open(testfiles["test_merge_source_2.map"])
result = results_dir + "/merged.mapping"
merger = gem.merge(input, [source_1, source_2], output=result)
count = 0
for read in merger:
count += 1
if read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13866:124450 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "0+1"
assert read.to_map().split("\t")[4] == "chr2:-:162359617:74G"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13753:124452 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "0+1"
assert read.to_map().split("\t")[4] == "chr9:+:38397301:54G20"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:14211:124259 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "1"
assert read.to_map().split("\t")[4] == "chr15:+:72492866:75"
assert count == 10
assert os.path.exists(result)
def test_merging_maps_to_file():
input = files.open(testfiles["test_merge_target.map"])
source_1 = files.open(testfiles["test_merge_source_1.map"])
source_2 = files.open(testfiles["test_merge_source_2.map"])
result = results_dir + "/merged.mapping"
merger = gem.merge(input, [source_1, source_2], output=result)
count = 0
for read in merger:
count += 1
if read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13866:124450 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "0+1"
assert read.to_map().split("\t")[4] == "chr2:-:162359617:74G"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:13753:124452 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "0+1"
assert read.to_map().split("\t")[4] == "chr9:+:38397301:54G20"
elif read.tag == "HWI-ST661:153:D0FTJACXX:2:1102:14211:124259 1:N:0:GCCAAT":
assert read.to_map().split("\t")[3] == "1"
assert read.to_map().split("\t")[4] == "chr15:+:72492866:75"
assert count == 10
assert os.path.exists(result)
def test_file_merge_async():
reads_1 = files.open(testfiles["test.map"])
reads_2 = files.open(testfiles["test.map"])
merged = gem.merge(reads_1, [reads_2])
num_reads = sum(1 for r in merged)
assert num_reads == 10