def test_nodefault(monkeypatch):
G.create_directory(tmpdir)
dbpath_tmp = os.path.join(tmpdir, "customdb.fas.nsq")
with open(dbpath_tmp, "w") as f:
f.write(">mockDB")
try:
monkeypatch.setattr('builtins.input', nodef_input)
conf_from_file = Config()
configpath = get_config_from_file(conf_from_file)[0]
compare_configfiles(reference_dict, configpath, "3")
if os.path.isfile(configpath):
os.remove(configpath)
configpath = get_config_from_file(conf_from_file)[1]
compare_configfiles(reference_dict, configpath, "4")
if os.path.isfile(configpath):
os.remove(configpath)
monkeypatch.setattr('builtins.input', nodef_input2)
conf_from_file = Config()
configpath = get_config_from_file(conf_from_file)[0]
compare_configfiles(reference_dict, configpath, "8")
configpath = get_config_from_file(conf_from_file)[1]
compare_configfiles(reference_dict, configpath, "9")
finally:
if os.path.isfile(dbpath_tmp):
os.remove(dbpath_tmp)
def prepare_blastdb(config):
if os.path.isdir(tmpdir):
shutil.rmtree(tmpdir)
G.create_directory(tmpdir)
config.customdb = os.path.join(tmpdir, "primer_customdb.fas")
config.blastdbv5 = False
def dbinputfiles():
filenames = [
"GCF_004088235v1_20191001.fna",
"GCF_002224565.1_ASM222456v1_genomic.fna"
]
dbfile = os.path.join(tmpdir, "primer_customdb.fas")
with open(dbfile, "w") as f:
for filename in filenames:
filepath = os.path.join(testfiles_dir, filename)
records = SeqIO.parse(filepath, "fasta")
for record in records:
if record.id == record.description:
description = (
record.id +
" Lactobacillus curvatus strain SRCM103465")
record.description = description
SeqIO.write(record, f, "fasta")
return dbfile
def create_customblastdb(config, infile):
cmd = [
"makeblastdb", "-in", infile, "-parse_seqids", "-title",
"mockconservedDB", "-dbtype", "nucl", "-out", config.customdb
]
G.run_subprocess(cmd, printcmd=False, logcmd=False, printoption=False)
dbfile = dbinputfiles()
create_customblastdb(config, dbfile)
def test_commandline_fail_correct_settings():
test = os.path.join("/", "primerdesign", "test")
if os.path.isdir(test):
shutil.rmtree(test)
G.create_directory(test)
G.create_directory(tmpdir)
os.chdir(test)
CONFFILE = os.path.join(testfiles_dir, "adconfig", "advanced_config3.json")
reffile = os.path.join(dict_path, "p3parameters")
wrong_ext = os.path.join(tmpdir, "p3parameters.txt")
shutil.copy(reffile, wrong_ext)
cmd = [
"speciesprimer.py", "-t", "Lactobacillus_curvatus", "--configfile",
CONFFILE, "-p", test
]
outputlist = G.read_shelloutput(cmd)
assert outputlist[-2::] == [
'/primerdesign/tmp/genus_abbrev.csv',
'A list containing the settings or a complete settings file'
' is required'
]
os.chdir(pipe_dir)
shutil.rmtree(test)
def prepare_files():
if os.path.isdir(genomic_dir):
shutil.rmtree(genomic_dir)
G.create_directory(genomic_dir)
G.create_directory(config_dir)
shutil.copy(conffile, testconfig)
files = ["GCF_001981925v1_20190923.ffn", "GCF_003410375v1_20190923.ffn"]
ffn_files_dir = os.path.join(testfiles_dir, "ffn_files")
for filename in files:
filepath = os.path.join(ffn_files_dir, filename)
sequences = []
with open(filepath) as f:
records = list(SeqIO.parse(f, "fasta"))
for record in records:
seq = str(record.seq)
sequences.append(seq)
newfilename = ".".join(
filename.split(".ffn")[0].split("v")) + "_genomic.fna"
outpath = os.path.join(genomic_dir, newfilename)
mockfna = SeqRecord(Seq("".join(sequences)),
id="MOCK_" + filename.split("_")[1],
name="MOCK_" + filename.split("_")[1],
description="Lactobacillus curvatus")
with open(outpath, "w") as o:
SeqIO.write(mockfna, o, "fasta")
maxcontig = os.path.join(genomic_dir, "GCF_007MOCKv1_genomic.fna")
with open(maxcontig, "w") as f:
f.write(">GCF_007MOCKv1_file\n")
for i in range(0, 500):
f.write(">GCF_007MOCKv1_" + str(i) + "\n")
def test_compare_corrupted_md5(delete_config):
chmdfile = "ref_prok_rep_genomes.05.tar.gz.md5"
create_mock_archives()
shutil.rmtree("md5_files")
G.create_directory("md5_files")
os.chdir(tmpdir)
for files in os.listdir(tmpdir):
if (
files.startswith("ref_prok_rep_genomes.05")
and "tar.gz" not in files
):
os.remove(files)
with open(chmdfile, "r") as f:
for line in f:
info = (
line.strip().split(" ")[0][0:-1] + "q67 " +
line.strip().split(" ")[1])
with open(chmdfile, "w") as f:
f.write(info)
getblastdb.download_from_ftp([chmdfile], delete_config)
check_part_five(outcome=True)
remove_tmp_db()
os.chdir("..")
if os.path.isdir(tmpdir):
shutil.rmtree(tmpdir)
def prepare_testfiles():
def prepare_tmp_db():
t = os.path.join(BASE_PATH, "testfiles", "tmp_config.json")
# Docker only
tmp_path = os.path.join(pipe_dir, "tmp_config.json")
if os.path.isfile(tmp_path):
os.remove(tmp_path)
shutil.copy(t, tmp_path)
def change_tmp_db():
tmp_path = os.path.join(pipe_dir, "tmp_config.json")
with open(tmp_path) as f:
for line in f:
tmp_dict = json.loads(line)
tmp_dict["new_run"].update({'modus': "continue", "targets": None})
with open(tmp_path, "w") as f:
f.write(json.dumps(tmp_dict))
def dbinputfiles():
filenames = [
"GCF_004088235v1_20191001.fna",
"GCF_002224565.1_ASM222456v1_genomic.fna"
]
with open(dbpath, "w") as f:
for filename in filenames:
filepath = os.path.join(testfiles_dir, filename)
records = SeqIO.parse(filepath, "fasta")
for record in records:
if record.id == record.description:
description = (
record.id +
" Lactobacillus curvatus strain SRCM103465")
record.description = description
SeqIO.write(record, f, "fasta")
mockseqs = os.path.join(testfiles_dir, "mocktemplate.seqs")
records = list(SeqIO.parse(mockseqs, "fasta"))
SeqIO.write(records, f, "fasta")
return dbpath
def create_customblastdb():
cmd = [
"makeblastdb", "-in", dbpath, "-parse_seqids", "-title",
"mockconservedDB", "-dbtype", "nucl", "-out", dbpath
]
G.run_subprocess(cmd, printcmd=False, logcmd=False, printoption=False)
if os.path.isdir(os.path.dirname(dbpath)):
shutil.rmtree(os.path.dirname(dbpath))
test = os.path.join("/", "primerdesign", "test")
if os.path.isdir(test):
shutil.rmtree(test)
G.create_directory(os.path.dirname(dbpath))
dbinputfiles()
create_customblastdb()
prepare_tmp_db()
change_tmp_db()
def primerBLAST(pqc, config):
from speciesprimer import BlastPrep
prepare_QC_testfiles(pqc, config)
G.create_directory(pqc.primerblast_dir)
prep = BlastPrep(pqc.primerblast_dir, pqc.primerlist, "primer",
pqc.config.blastseqs)
use_cores, inputseqs = prep.run_blastprep()
reffile = os.path.join(testfiles_dir, "primer_nontargethits.json")
tofile = os.path.join(pqc.primerblast_dir, "nontargethits.json")
shutil.copy(reffile, tofile)
return inputseqs
def test_compare_md5_archive(delete_config):
chmdfile = "ref_prok_rep_genomes.05.tar.gz.md5"
create_mock_archives()
os.chdir(tmpdir)
for files in os.listdir(tmpdir):
if (
files.startswith("ref_prok_rep_genomes.05")
and "tar.gz" not in files
):
os.remove(files)
shutil.rmtree("md5_files")
G.create_directory("md5_files")
getblastdb.download_from_ftp([chmdfile], delete_config)
check_part_five(outcome=True)
remove_tmp_db()
os.chdir("..")
if os.path.isdir(tmpdir):
shutil.rmtree(tmpdir)
def test_no_old_md5(keep_config):
chmdfile = "ref_prok_rep_genomes.05.tar.gz.md5"
create_mock_archives()
os.chdir(tmpdir)
for files in os.listdir(tmpdir):
if (
files.startswith("ref_prok_rep_genomes.05")
and not files.endswith(".md5")
):
os.remove(files)
shutil.rmtree("md5_files")
G.create_directory("md5_files")
getblastdb.download_from_ftp([chmdfile], keep_config)
check_part_five(outcome=True)
assert os.path.isfile("ref_prok_rep_genomes.05.tar.gz") is True
remove_tmp_db()
os.chdir("..")
if os.path.isdir(tmpdir):
shutil.rmtree(tmpdir)
def prepare_QC_testfiles(pqc, config):
targetdir = os.path.join(config.path, config.target)
fna_dir = os.path.join(targetdir, "fna_files")
qc_file = os.path.join(testfiles_dir, "Lb_curva_qc_sequences_ignoreqc.csv")
qc_out = os.path.join(pqc.summ_dir, "Lb_curva_qc_sequences.csv")
G.create_directory(fna_dir)
G.create_directory(pqc.primer_qc_dir)
if os.path.isfile(qc_out):
os.remove(qc_out)
if os.path.isdir(pqc.summ_dir):
shutil.rmtree(pqc.summ_dir)
G.create_directory(pqc.summ_dir)
shutil.copy(qc_file, qc_out)
files = ["GCF_001981925v1_20190923.ffn", "GCF_003410375v1_20190923.ffn"]
ffn_files_dir = os.path.join(testfiles_dir, "ffn_files")
for filename in files:
filepath = os.path.join(ffn_files_dir, filename)
sequences = []
with open(filepath) as f:
records = list(SeqIO.parse(f, "fasta"))
for record in records:
seq = str(record.seq)
sequences.append(seq)
newfilename = filename.split(".ffn")[0] + ".fna"
outpath = os.path.join(fna_dir, newfilename)
mockfna = SeqRecord(Seq("".join(sequences)),
id="MOCK_" + filename.split("_")[1],
name="MOCK_" + filename.split("_")[1],
description="Lactobacillus curvatus")
with open(outpath, "w") as o:
SeqIO.write(mockfna, o, "fasta")
def test_qc_nottrue():
from speciesprimer import CLIconf
confargs["ignore_qc"] = False
args = AttrDict(confargs)
nontargetlist = []
config = CLIconf(args.minsize, args.maxsize, args.mpprimer, args.exception,
args.target, args.path, args.intermediate, args.qc_gene,
args.mfold, args.skip_download, args.assemblylevel,
nontargetlist, args.skip_tree, args.nolist, args.offline,
args.ignore_qc, args.mfethreshold, args.customdb,
args.blastseqs, args.probe, args.blastdbv5)
reffile = os.path.join(testfiles_dir, "ref_primer3_summary.json")
with open(reffile) as f:
for line in f:
primer3dict = json.loads(line)
pqc = PrimerQualityControl(config, primer3dict)
G.create_directory(pqc.summ_dir)
qc_file = os.path.join(testfiles_dir, "Lb_curva_qc_sequences_ignoreqc.csv")
qc_out = os.path.join(pqc.summ_dir, "Lb_curva_qc_sequences.csv")
shutil.copy(qc_file, qc_out)
genDB = os.path.join(pqc.primer_qc_dir, "Lb_curva.genomic")
tempDB = os.path.join(pqc.primer_qc_dir, "template.sequences")
primer_qc_list = []
pqc.create_template_db_file(primer_qc_list)
pqc.create_assembly_db_file()
assemblyfilepath = os.path.join(pqc.primer_qc_dir,
H.abbrev(pqc.target) + ".genomic")
templatefilepath = os.path.join(pqc.primer_qc_dir, "template.sequences")
dblist = [assemblyfilepath, templatefilepath]
for dbpath in dblist:
P.index_database(dbpath)
assert os.path.isfile(genDB) is False
assert os.path.isfile(tempDB) is False
os.chdir(pipe_dir)
shutil.rmtree(config.path)
if os.path.isdir(pqc.summ_dir):
shutil.rmtree(pqc.summ_dir)
def write_config_file(self, target):
path = self.config_dict[target]["path"]
target_dir = os.path.join(path, target)
genomic_dir = os.path.join(target_dir, "genomic_fna")
config_dir = os.path.join(target_dir, "config")
config_file = os.path.join(config_dir, "config.json")
G.create_directory(target_dir)
G.create_directory(config_dir)
G.create_directory(genomic_dir)
self.config_paths.append(config_file)
with open(config_file, "w") as f:
f.write(json.dumps(self.config_dict[target]))
def get_DB(mode=False):
today = time.strftime("%Y_%m_%d", time.localtime())
if mode == "auto":
signal.signal(signal.SIGTERM, exitatsigterm)
with open(tmp_db_path, 'r') as f:
for line in f:
tmp_db = json.loads(line)
delete = tmp_db['BLAST_DB']['delete']
db = tmp_db['BLAST_DB']['db']
try:
dbdir = tmp_db['BLAST_DB']['path']
blastdb_dir = dbdir
except KeyError:
blastdb_dir = "/blastdb"
try:
if tmp_db['BLAST_DB']['test'] is True:
test = True
except KeyError:
test = False
logging.basicConfig(
filename=os.path.join(
"/", "primerdesign",
"speciesprimer_" + today + ".log"),
level=logging.DEBUG, format="%(message)s")
else:
parser = commandline()
args = parser.parse_args()
delete = args.delete
test = args.test
if args.dbpath:
if args.dbpath.endswith("/"):
blastdb_dir = args.dbpath
else:
blastdb_dir = args.dbpath + "/"
else:
blastdb_dir = os.getcwd() + "/"
db = args.database
logging.basicConfig(
filename=os.path.join(
blastdb_dir, "blastdb_download_" + today + ".log"),
level=logging.DEBUG, format="%(message)s")
G.create_directory(os.path.join(blastdb_dir, "md5_files"))
conf = config(db, blastdb_dir, delete, test)
logger("Start Download of NCBI " + db + " BLAST database")
get_md5files(conf)
filelist = get_filelist(conf)
download_from_ftp(filelist, conf)
nal_filepath = os.path.join(blastdb_dir, db + ".nal")
if os.path.isfile(nal_filepath):
logger("NCBI " + db + " BLAST database is ready")
else:
logger(
"Error nal file for NCBI " + db + " BLAST database is missing")
def test_advancedconfig_fromfile():
CONFFILE = os.path.join(testfiles_dir, "adconfig", "advanced_config.json")
H.advanced_pipe_config(CONFFILE)
filenames = [
"no_blast.gi", "genus_abbrev.csv", "p3parameters", "species_list.txt"]
for filename in filenames:
ref = os.path.join(ref_data, "advanced_config", filename)
newfile = os.path.join(dict_path, filename)
assert filecmp.cmp(newfile, ref) is True
deffile = os.path.join(dict_path, "default", filename)
os.remove(newfile)
shutil.copy(deffile, newfile)
G.create_directory(tmpdir)
for filename in filenames:
ref = os.path.join(ref_data, "advanced_config", filename)
tmp = os.path.join(tmpdir, filename)
shutil.copy(ref, tmp)
CONFFILE = os.path.join(testfiles_dir, "adconfig", "advanced_config2.json")
H.advanced_pipe_config(CONFFILE)
for filename in filenames:
ref = os.path.join(ref_data, "advanced_config", filename)
newfile = os.path.join(dict_path, filename)
assert filecmp.cmp(newfile, ref) is True
deffile = os.path.join(dict_path, "default", filename)
os.remove(newfile)
shutil.copy(deffile, newfile)
CONFFILE = os.path.join(testfiles_dir, "adconfig","advanced_config3.json")
tmp = os.path.join(tmpdir, "p3parameters")
wrong_ext = os.path.join(tmpdir, "p3parameters.txt")
shutil.copy(tmp, wrong_ext)
exitstat = H.advanced_pipe_config(CONFFILE)
assert exitstat == 1
CONFFILE = os.path.join(testfiles_dir, "adconfig", "advanced_config4.json")
exitstat = H.advanced_pipe_config(CONFFILE)
assert exitstat == 1
for filename in filenames:
ref = os.path.join(ref_data, "advanced_config", filename)
newfile = os.path.join(dict_path, filename)
deffile = os.path.join(dict_path, "default", filename)
os.remove(newfile)
shutil.copy(deffile, newfile)
CERTFILE = os.path.join(testfiles_dir, "adconfig", "mock_certificate.crt")
CERT_REF = os.path.join(tmpdir, "mock_certificate.crt")
CONFFILE = os.path.join(tmpdir, "adconfig.json")
CERT_EXT = os.path.join(tmpdir, "mock_certificate.txt")
shutil.copy(CERTFILE, CERT_REF)
shutil.copy(CERT_REF, CERT_EXT)
conf = {"certificate": CERT_REF}
with open(CONFFILE, "w") as f:
f.write(json.dumps(conf))
exitstat = H.advanced_pipe_config(CONFFILE)
assert exitstat == 0
os.remove(CONFFILE)
conf = {"certificate": CERT_EXT}
with open(CONFFILE, "w") as f:
f.write(json.dumps(conf))
exitstat = H.advanced_pipe_config(CONFFILE)
assert exitstat == 1
os.remove(CONFFILE)
conf = {"certificate": tmpdir}
with open(CONFFILE, "w") as f:
f.write(json.dumps(conf))
exitstat = H.advanced_pipe_config(CONFFILE)
assert exitstat == 1
shutil.rmtree(tmpdir)
def test_rollback():
testdir = os.path.join("/", "primerdesign", "test")
target_dir = os.path.join(testdir, "Lactobacillus_curvatus")
pangenome_dir = os.path.join(target_dir, "Pangenome")
results_dir = os.path.join(pangenome_dir, "results")
alignments_dir = os.path.join(results_dir, "alignments")
consensus_dir = os.path.join(results_dir, "consensus")
primer_dir = os.path.join(results_dir, "primer")
primer_qc_dir = os.path.join(primer_dir, "primer_QC")
blast_dir = os.path.join(results_dir, "blast")
primerblast_dir = os.path.join(primer_dir, "primerblast")
G.create_directory(primerblast_dir)
G.create_directory(alignments_dir)
G.create_directory(consensus_dir)
G.create_directory(blast_dir)
G.create_directory(primer_qc_dir)
# annotation
annotdir = "GCF_902362325v1_20191114"
G.create_directory(os.path.join(target_dir, annotdir))
outdir = annotdir
assert os.path.isdir(os.path.join(target_dir, annotdir)) is True
G.keyexit_rollback("annotation", dp=os.path.join(target_dir, outdir))
assert os.path.isdir(os.path.join(target_dir, annotdir)) is False
# p3
p3_file = os.path.join(primer_dir, "primer3_output")
output_file = p3_file
with open(p3_file, "w") as f:
f.write("PRIMER3MOCK")
assert os.path.isfile(p3_file) is True
G.keyexit_rollback("primer3 run", fp=output_file)
assert os.path.isfile(p3_file) is False
# alignments
run_file = os.path.join(results_dir, "run_prank")
with open(run_file, "w") as f:
f.write("PRANK MSA MOCK")
assert os.path.isfile(run_file) is True
G.keyexit_rollback("Prank MSA run", dp=alignments_dir, fp=run_file)
assert os.path.isfile(run_file) is False
assert os.path.isdir(alignments_dir) is False
# consensus
run_file = os.path.join(results_dir, "run_consensus")
with open(run_file, "w") as f:
f.write("CONSAMBIG MOCK")
assert os.path.isfile(run_file) is True
G.keyexit_rollback("consensus run", dp=consensus_dir, fp=run_file)
assert os.path.isfile(run_file) is False
assert os.path.isdir(consensus_dir) is False
# dp only
run_file = os.path.join(results_dir, "run_consensus")
G.create_directory(consensus_dir)
assert os.path.isfile(run_file) is False
G.keyexit_rollback("consensus run", dp=consensus_dir, fp=run_file)
assert os.path.isfile(run_file) is False
assert os.path.isdir(consensus_dir) is False
# fp only
run_file = os.path.join(results_dir, "run_consensus")
with open(run_file, "w") as f:
f.write("CONSAMBIG MOCK")
G.keyexit_rollback("consensus run", dp=consensus_dir, fp=run_file)
assert os.path.isfile(run_file) is False
assert os.path.isdir(consensus_dir) is False
# DB extraction
filename = "BLASTnontarget0.sequences"
filepath = os.path.join(primer_qc_dir, filename)
with open(filepath, "w") as f:
f.write("DB extract Mock")
assert os.path.isfile(filepath) is True
G.keyexit_rollback("DB extraction", fp=filepath)
assert os.path.isfile(filepath) is False
# blast
directory = blast_dir
blastfile = "conserved_0_results.xml"
filepath = os.path.join(blast_dir, blastfile)
with open(filepath, "w") as f:
f.write("BLAST MOCK")
filename = blastfile
assert os.path.isfile(filepath) is True
G.keyexit_rollback("BLAST search", dp=directory, fn=filename)
assert os.path.isfile(filepath) is False
# primerblast
directory = primerblast_dir
primerblastfile = "primer_0_results.xml"
filepath = os.path.join(primerblast_dir, primerblastfile)
with open(filepath, "w") as f:
f.write("PRIMERBLAST MOCK")
filename = primerblastfile
assert os.path.isfile(filepath) is True
G.keyexit_rollback("BLAST search", dp=directory, fn=filename)
assert os.path.isfile(filepath) is False
# DB indexing
dbfiles = [
"Lb_curva.genomic", "Lb_curva.genomic.sqlite3.db",
"Lb_curva.genomic.uni", "Lb_curva.genomic.2bit"]
for files in dbfiles:
filepath = os.path.join(primer_qc_dir, files)
with open(filepath, "w") as f:
f.write(files + " Mock")
assert os.path.isfile(filepath) is True
db_name = "Lb_curva.genomic"
G.keyexit_rollback("DB indexing", dp=primer_qc_dir, search=db_name)
for files in dbfiles:
filepath = os.path.join(primer_qc_dir, files)
assert os.path.isfile(filepath) is False
# pangenome
assert os.path.isdir(pangenome_dir) is True
G.keyexit_rollback("pan-genome analysis", dp=pangenome_dir)
assert os.path.isdir(pangenome_dir) is False
shutil.rmtree(testdir)
def create_mock_archives():
G.create_directory(tmpdir)
G.create_directory(downdir)
refhtml = os.path.join(testfiles_dir, "download.html")
htmlfile = os.path.join(tmpdir, "mockfiles","download.html")
shutil.copy(refhtml, htmlfile)
with open(os.path.join(tmpdir, "ref_prok_rep_genomes.html.tmp"), "w") as f:
f.write("mock")
for i in range(0, 6):
num = "{:02d}".format(i)
filestart = "ref_prok_rep_genomes." + num
mockends = [
".nhr", ".nni",
".nsi", ".nin",
".nog", ".nsq",
".nnd", ".nsd"]
archive = "ref_prok_rep_genomes." + num + ".tar.gz"
os.chdir(tmpdir)
with tarfile.open(archive, "w:gz") as tar:
for ending in mockends:
filepath = filestart + ending
with open(filepath, "w") as f:
f.write("")
tar.add(filepath)
with open(archive + ".md5", "w") as f:
archmd5 = md5Checksum(archive)
f.write(archmd5 + " " + archive)
archfile = "ref_prok_rep_genomes.03.tar.gz"
tmpfile = os.path.join(downdir, "ref_prok_rep_genomes.03.tar.gz")
if os.path.isfile(archfile):
shutil.move(archfile, tmpfile)
archfile = "ref_prok_rep_genomes.05.tar.gz"
tmpfile = os.path.join(downdir, "ref_prok_rep_genomes.05.tar.gz")
if os.path.isfile(archfile):
shutil.copy(archfile, tmpfile)
for files in os.listdir(tmpdir):
if (
files.startswith("ref_prok_rep_genomes.03")
and not files.endswith(".md5")
):
os.remove(files)
md5_dir = os.path.join(tmpdir, "md5_files")
G.create_directory(md5_dir)
for files in os.listdir(tmpdir):
if files.endswith("tar.gz.md5"):
fromfile = os.path.join(tmpdir, files)
tofilepath = os.path.join(md5_dir, files)
downfiles = os.path.join(downdir, files)
shutil.copy(fromfile, tofilepath)
shutil.copy(fromfile, downfiles)
chmdfile = os.path.join(md5_dir, "ref_prok_rep_genomes.05.tar.gz.md5")
with open(chmdfile, "r") as f:
for line in f:
info = (
line.strip().split(" ")[0][0:-1] + "k " +
line.strip().split(" ")[1])
with open(chmdfile, "w") as f:
f.write(info)
nalfile = os.path.join(tmpdir, "ref_prok_rep_genomes.nal")
with open(nalfile, "w") as f:
f.write("")
os.remove("ref_prok_rep_genomes.04.tar.gz")