def create_pipeline(command, config):
if command == "genomeindex":
return GenomeIndexPipeline(config)
elif command == "mappable_regions":
return MappableRegionsPipeline(config)
elif command == "bins":
return BinsPipeline(config)
elif command == "mapping":
return MappingPipeline(config)
elif command == "varbin":
return VarbinPipeline(config)
elif command == "scclust":
return Rpipeline(config)
elif command == "extract_10x":
return Extract10xPipeline(config)
elif command == "varbin_10x":
return Varbin10xPipeline(config)
elif command == "prepare":
pipelines = [
GenomeIndexPipeline(config),
MappableRegionsPipeline(config),
BinsPipeline(config),
]
return CompositePipeline(config, pipelines)
elif command == "process":
pipelines = [
MappingPipeline(config),
VarbinPipeline(config),
Rpipeline(config),
]
return CompositePipeline(config, pipelines)
raise ValueError(f"Unexpected command: {command}")
async def test_bowtie_mappings(tests_config, event_loop, chrom):
event_loop.set_debug(True)
pipeline = MappableRegionsPipeline(tests_config)
bowtie = await pipeline.async_start_bowtie()
reads_generator = pipeline.generate_reads([chrom], 100)
writer = asyncio.Task(
pipeline.async_write_reads_generator(bowtie.stdin, reads_generator))
infile = bowtie.stdout
while True:
line = await infile.readline()
if not line:
break
line = line.decode()
if line[0] == '@':
# comment
continue
mapping = Mapping.parse_sam(line)
if mapping.flag == 0:
# print(mapping)
chromosome, pos = mapping.name.split('.')
assert int(pos) == mapping.start
assert chrom == chromosome
await writer
async def test_async_generate_mappable_regions(tests_config, event_loop):
# Enable debugging
event_loop.set_debug(True)
pipeline = MappableRegionsPipeline(tests_config)
await pipeline.async_generate_mappable_regions(['chrM'], 100)
async def test_async_mappable_regions_50(tests_config, event_loop, chrom):
event_loop.set_debug(True)
pipeline = MappableRegionsPipeline(tests_config)
filename = os.path.join('tests/data',
"{}.50mer.mappable.regions.txt.gz".format(chrom))
gold_df = pd.read_csv(filename,
sep='\t',
compression='gzip',
header=None,
names=['chrom', 'start', 'end'])
with io.StringIO() as outfile:
await pipeline.async_generate_mappable_regions([chrom],
50,
outfile=outfile)
infile = io.StringIO(outfile.getvalue())
df = pd.read_csv(infile,
sep='\t',
header=None,
names=['chrom', 'start', 'end'])
print(df.head())
assert np.all(df.columns == gold_df.columns)
assert np.all(df.start == gold_df.start)
assert np.all(df.end == gold_df.end)
def test_async_mappable_regions_50_hisat(tests_config, chrom, hisat2):
filename = os.path.join('sgains/tests/data',
"{}.50mer.mappable.regions.txt.gz".format(chrom))
gold_df = pd.read_csv(filename,
sep='\t',
compression='gzip',
header=None,
names=['chrom', 'start', 'end'])
with io.StringIO() as outfile:
pipeline = MappableRegionsPipeline(tests_config, hisat2)
pipeline.generate_mappable_regions([chrom], 50, outfile=outfile)
outfile.flush()
infile = io.StringIO(outfile.getvalue())
df = pd.read_csv(infile,
sep='\t',
header=None,
names=['chrom', 'start', 'end'])
print(df.head())
assert np.all(df.columns == gold_df.columns)
assert len(df) == len(gold_df)
assert np.all(df.start == gold_df.start)
assert np.all(df.end == gold_df.end)
async def test_generate_reads(tests_config):
pipeline = MappableRegionsPipeline(tests_config)
generator = pipeline.generate_reads(['chr1'], 100)
for num, rec in enumerate(generator):
print(rec.id, len(rec))
if num >= 10:
break
generator.close()
def test_reads_generator(tests_config):
pipeline = MappableRegionsPipeline(tests_config)
reads_generator = pipeline.generate_reads(['chrM'], 100)
# ouptut_writer = pipeline.mappable_regions_writer()
count = 0
for _read in reads_generator:
# print(read)
count += 1
print(count)
assert 16472 == count
async def test_async_mappings_generator(tests_config, event_loop):
# Enable debugging
event_loop.set_debug(True)
pipeline = MappableRegionsPipeline(tests_config)
bowtie = await pipeline.async_start_bowtie()
reads_generator = pipeline.generate_reads(['chrM'], 100)
writer = asyncio.Task(
pipeline.async_write_reads_generator(bowtie.stdin, reads_generator))
async for mapping in pipeline.async_mappable_regions_generator(
bowtie.stdout):
print(mapping)
await bowtie.wait()
await writer
def run(self, args):
print("mappable-regions subcommand called with args: {}".format(args))
self.process_args(args)
pipeline = MappableRegionsPipeline(self.config)
pipeline.run()
def test_generate_mappable_regions_bwa(tests_config, bwa):
assert bwa is not None
pipeline = MappableRegionsPipeline(tests_config, bwa)
pipeline.generate_mappable_regions(['chrM'], 100)
def test_generate_mappable_regions_hisat(tests_config, hisat2):
assert hisat2 is not None
pipeline = MappableRegionsPipeline(tests_config, hisat2)
pipeline.generate_mappable_regions(['chrM'], 100)
def test_generate_mappable_regions(tests_config):
pipeline = MappableRegionsPipeline(tests_config)
pipeline.generate_mappable_regions(['chrM'], 100)
def test_bowtie_command(tests_config):
pipeline = MappableRegionsPipeline(tests_config)
command = pipeline.genome.aligner.build_mappable_regions_command()
print(command)
assert command[0] == 'hisat2'