def test_dory_extract_reads_fq(location):
# check that FASTQ is passed through properly.
copy_dory_catlas()
copy_dory_catlas_search()
copy_dory_subset()
# run make_bgzf - FIXTURE
print('** running make_bgzf')
args = ['dory-subset.fq', '-o', relative_file('dory/dory.reads.bgz')]
make_bgzf.main(args)
# run label_cdbg - FIXTURE
print('** running label_cdbg')
args = ['dory_k21_r1',
relative_file('dory/dory.reads.bgz'),
'dory_k21_r1/reads.bgz.labels']
label_cdbg.main(args)
# run extract_reads
print('** running extract_reads')
args = [relative_file('dory/dory.reads.bgz'),
'dory_k21_r1/reads.bgz.labels',
'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.txt.gz',
'-o',
'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.reads.fa.gz']
extract_reads.main(args)
reads_filename = 'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.reads.fa.gz'
reads = [ record for record in screed.open(reads_filename) ]
assert len(reads) == 2
assert len(reads[0].quality) # FASTQ preserved!
def test_dory_extract_reads(location):
copy_dory_catlas()
copy_dory_catlas_search()
copy_dory_subset()
# run make_bgzf - FIXTURE
print('** running make_bgzf')
args = ['dory-subset.fa', '-o', relative_file('dory/dory.reads.bgz')]
make_bgzf.main(args)
# run label_cdbg - FIXTURE
print('** running label_cdbg')
args = ['dory_k21_r1',
relative_file('dory/dory.reads.bgz'),
'dory_k21_r1/reads.bgz.labels']
label_cdbg.main(args)
# run extract_reads
print('** running extract_reads')
args = [relative_file('dory/dory.reads.bgz'),
'dory_k21_r1/reads.bgz.labels',
'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.txt.gz',
'-o',
'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.reads.fa.gz']
extract_reads.main(args)
reads_filename = 'dory_k21_r1_search_oh0/dory-head.fa.cdbg_ids.reads.fa.gz'
reads = [ record for record in screed.open(reads_filename) ]
assert len(reads) == 2
def test_dory_label_cdbg(location):
copy_dory_catlas()
copy_dory_subset()
# run make_bgzf - FIXTURE
print('** running make_bgzf')
args = ['dory-subset.fa', '-o', relative_file('dory/dory.reads.bgz')]
make_bgzf.main(args)
# run label_cdbg
print('** running label_cdbg')
args = ['dory_k21_r1',
relative_file('dory/dory.reads.bgz'),
'dory_k21_r1/reads.bgz.labels']
label_cdbg.main(args)
def test_dory_make_bgzf(location):
copy_dory_subset()
# run make_bgzf
print('** running make_bgzf')
args = ['dory-subset.fa', '-o', relative_file('dory/dory.reads.bgz')]
make_bgzf.main(args)
def test_dory_query_workflow(location):
from spacegraphcats.cdbg import bcalm_to_gxt
copy_dory_head()
copy_dory_subset()
# make the output directory
try:
os.mkdir('dory_k21_r1')
except FileExistsError:
pass
# convert the bcalm file to gxt
args = ['-k', '21', '-P',
relative_file('dory/bcalm.dory.k21.unitigs.fa'),
'dory_k21_r1/cdbg.gxt',
'dory_k21_r1/contigs.fa.gz']
bcalm_to_gxt.main(args)
# build catlas
args = pytest_utils.Args()
args.no_checkpoint = True
args.level = 0
args.radius = 1
args.project = 'dory_k21_r1'
print('** running catlas')
catlas.main(args)
# make k-mer search index
args = '-k 21 dory_k21_r1'.split()
print('** running index_contigs_by_kmer')
index_contigs_by_kmer.main(args)
# do search!!
args='dory_k21_r1 dory_k21_r1_search_oh0 --query dory-head.fa -k 21'.split()
try:
query_by_sequence.main(args)
except SystemExit as e:
assert e.code == 0, str(e)
# check output!
output_path = 'dory_k21_r1_search_oh0/'
assert os.path.exists(output_path + 'command.txt')
assert os.path.exists(output_path + 'dory-head.fa.frontier.txt.gz')
assert os.path.exists(output_path + 'dory-head.fa.cdbg_ids.txt.gz')
assert os.path.exists(output_path + 'dory-head.fa.response.txt')
assert os.path.exists(output_path + 'dory-head.fa.contigs.sig')
assert os.path.exists(output_path + 'results.csv')
with open(output_path + 'results.csv') as fp:
lines = fp.readlines()
assert len(lines) == 2
last_line = lines[-1].strip()
assert last_line == 'dory-head.fa,1.0,1.0,1671,2,21,1631,1.0,0.0,0.0,dory_k21_r1'
def test_dory_query_by_hashval(location):
testdata = relative_file('data/dory-k31-hashval-queries.txt')
shutil.copyfile(testdata, 'dory-k31-hashval-queries.txt')
copy_dory_catlas()
# index by hashval
args = '-k 31 dory_k21_r1/contigs.fa.gz dory_k21_r1_mh.pickle'
index_cdbg_by_minhash.main(args.split())
args = '-k 31 dory_k21_r1 dory_k21_r1_mh.pickle dory-k31-hashval-queries.txt dory_k21_r1_hashval_k31'
query_by_hashval.main(args.split())
assert os.path.exists('dory_k21_r1_hashval_k31/hashval_results.csv')
def test_dory_search_nomatch(location):
# test situations where zero k-mers match - should not fail.
copy_dory_catlas()
testdata = relative_file('data/random-query-nomatch.fa')
shutil.copyfile(testdata, 'random-query.fa')
# make k-mer search index
args = '-k 21 dory_k21_r1'.split()
print('** running index_contigs_by_kmer')
index_contigs_by_kmer.main(args)
# do search!!
args='dory_k21_r1 dory_k21_r1_search_oh0 --query random-query.fa -k 21'.split()
try:
query_by_sequence.main(args)
except SystemExit as e:
assert e.code == 0, str(e)
def copy_dory_subset():
testdata = relative_file('data/dory-subset.fa')
shutil.copyfile(testdata, 'dory-subset.fa')
testdata = relative_file('data/dory-subset.fq')
shutil.copyfile(testdata, 'dory-subset.fq')
def copy_dory_head():
testdata = relative_file('data/dory-head.fa')
shutil.copyfile(testdata, 'dory-head.fa')
def copy_dory_sig():
testdata = relative_file('data/dory-subset.fq.sig')
shutil.copyfile(testdata, 'dory-subset.fq.sig')