def test_database(self):
if find_cmd("sqlite3"):
with make_workdir() as workdir:
arg = namedtuple('args', 'debug print_debug json ')
args = arg(True, True, "../../data/examples/seqcluster.json")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Create database")
make_database(data)
def test_database(self):
if find_cmd("sqlite3"):
with make_workdir() as workdir:
arg = namedtuple('args', 'debug print_debug json ')
args = arg(True, True, "../../data/examples/seqcluster.json")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Create database")
make_database(data)
def test_databse(self):
if find_cmd("sqlite3"):
mod_dir = os.path.dirname(inspect.getfile(seqcluster)).replace("seqcluster/", "")
os.chdir(os.path.join(mod_dir, "data/examples"))
arg = namedtuple('args', 'debug print_debug json ')
args = arg(True, True, "seqcluster.json")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Create databse")
make_database(data)
def test_predict(self):
if find_cmd("tRNAscan-SE"):
mod_dir = os.path.dirname(inspect.getfile(seqcluster)).replace("seqcluster/", "")
os.chdir(os.path.join(mod_dir, "data/examples"))
out_dir = os.path.join(mod_dir, "data/examples/predictions")
arg = namedtuple('args', 'debug print_debug MIN_SEQ json reference')
args = arg(True, True, 1, "seqcluster.json", "../genomes/genome.fa")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Start prediction")
make_predictions(data, out_dir, args)
def get_loci_fasta(loci, out_fa, ref):
"""get fasta from precursor"""
if not find_cmd("bedtools"):
raise ValueError("Not bedtools installed")
with make_temp_directory() as temp:
bed_file = os.path.join(temp, "file.bed")
for nc, loci in loci.iteritems():
for l in loci:
with open(bed_file, 'w') as bed_handle:
logger.debug("get_fasta: loci %s" % l)
nc, c, s, e, st = l
print("{0}\t{1}\t{2}\t{3}\t{3}\t{4}".format(c, s, e, nc, st), file=bed_handle)
get_fasta(bed_file, ref, out_fa)
return out_fa
def test_databse(self):
if find_cmd("sqlite3"):
mod_dir = os.path.dirname(inspect.getfile(seqcluster)).replace(
"seqcluster/", "")
os.chdir(os.path.join(mod_dir, "data/examples"))
arg = namedtuple('args', 'debug print_debug json ')
args = arg(True, True, "seqcluster.json")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Create databse")
make_database(data)
def get_loci_fasta(loci, out_fa, ref):
"""get fasta from precursor"""
if not find_cmd("bedtools"):
raise ValueError("Not bedtools installed")
with make_temp_directory() as temp:
bed_file = os.path.join(temp, "file.bed")
for nc, loci in loci.iteritems():
for l in loci:
with open(bed_file, 'w') as bed_handle:
logger.debug("get_fasta: loci %s" % l)
nc, c, s, e, st = l
print("{0}\t{1}\t{2}\t{3}\t{3}\t{4}".format(
c, s, e, nc, st),
file=bed_handle)
get_fasta(bed_file, ref, out_fa)
return out_fa
def map_to_precursors(seqs, names, loci, out_file, args):
"""map sequences to precursors with razers3"""
with make_temp_directory() as temp:
pre_fasta = os.path.join(temp, "pre.fa")
seqs_fasta = os.path.join(temp, "seqs.fa")
out_sam = os.path.join(temp, "out.sam")
pre_fasta = get_loci_fasta(loci, pre_fasta, args.ref)
out_precursor_file = out_file.replace("tsv", "fa")
seqs_fasta = get_seqs_fasta(seqs, names, seqs_fasta)
# print(open(pre_fasta).read().split("\n")[1])
if find_cmd("razers3"):
cmd = "razers3 -dr 2 -i 80 -rr 90 -f -o {out_sam} {temp}/pre.fa {seqs_fasta}"
run(cmd.format(**locals()))
out_file = read_alignment(out_sam, loci, seqs, out_file)
shutil.copy(pre_fasta, out_precursor_file)
return out_file
def deprecated_map_to_precursors(seqs, names, loci, out_file, args):
"""map sequences to precursors with bowtie"""
with make_temp_directory() as temp:
pre_fasta = os.path.join(temp, "pre.fa")
seqs_fasta = os.path.join(temp, "seqs.fa")
out_sam = os.path.join(temp, "out.sam")
pre_fasta = get_loci_fasta(loci, pre_fasta, args.ref)
out_precursor_file = out_file.replace("tsv", "fa")
seqs_fasta = get_seqs_fasta(seqs, names, seqs_fasta)
if find_cmd("bowtie2-build"):
cmd = "bowtie2-build -f {pre_fasta} {temp}/pre"
run(cmd.format(**locals()))
cmd = "bowtie2 -a --rdg 7,3 --mp 4 --end-to-end -D 20 -R 3 -N 0 -i S,1,0.8 -L 3 -f -x {temp}/pre -U {seqs_fasta} -S {out_sam}"
run(cmd.format(**locals()))
out_file = read_alignment(out_sam, loci, seqs, out_file)
shutil.copy(pre_fasta, out_precursor_file)
return out_file
def map_to_precursors(seqs, names, loci, out_file, args):
"""map sequences to precursors with razers3"""
with make_temp_directory() as temp:
pre_fasta = os.path.join(temp, "pre.fa")
seqs_fasta = os.path.join(temp, "seqs.fa")
out_sam = os.path.join(temp, "out.sam")
pre_fasta = get_loci_fasta(loci, pre_fasta, args.ref)
out_precursor_file = out_file.replace("tsv", "fa")
seqs_fasta = get_seqs_fasta(seqs, names, seqs_fasta)
# print(open(pre_fasta).read().split("\n")[1])
if find_cmd("razers3"):
cmd = "razers3 -dr 2 -i 80 -rr 90 -f -o {out_sam} {temp}/pre.fa {seqs_fasta}"
run(cmd.format(**locals()))
out_file = read_alignment(out_sam, loci, seqs, out_file)
shutil.copy(pre_fasta, out_precursor_file)
return out_file
def deprecated_map_to_precursors(seqs, names, loci, out_file, args):
"""map sequences to precursors with bowtie"""
with make_temp_directory() as temp:
pre_fasta = os.path.join(temp, "pre.fa")
seqs_fasta = os.path.join(temp, "seqs.fa")
out_sam = os.path.join(temp, "out.sam")
pre_fasta = get_loci_fasta(loci, pre_fasta, args.ref)
out_precursor_file = out_file.replace("tsv", "fa")
seqs_fasta = get_seqs_fasta(seqs, names, seqs_fasta)
if find_cmd("bowtie2-build"):
cmd = "bowtie2-build -f {pre_fasta} {temp}/pre"
run(cmd.format(**locals()))
cmd = "bowtie2 -a --rdg 7,3 --mp 4 --end-to-end -D 20 -R 3 -N 0 -i S,1,0.8 -L 3 -f -x {temp}/pre -U {seqs_fasta} -S {out_sam}"
run(cmd.format(**locals()))
out_file = read_alignment(out_sam, loci, seqs, out_file)
shutil.copy(pre_fasta, out_precursor_file)
return out_file
def test_predict(self):
if find_cmd("tRNAscan-SE"):
mod_dir = os.path.dirname(inspect.getfile(seqcluster)).replace(
"seqcluster/", "")
os.chdir(os.path.join(mod_dir, "data/examples"))
out_dir = os.path.join(mod_dir,
"data/examples/test_out_predictions")
arg = namedtuple('args',
'debug print_debug MIN_SEQ json reference')
args = arg(True, True, 1, "seqcluster.json",
"../genomes/genome.fa")
initialize_logger(".", args.debug, args.print_debug)
logger = mylog.getLogger(__name__)
logger.info(args)
logger.info("Reading data")
data = load_data(args.json)
logger.info("Start prediction")
is_tRNA(data, out_dir, args)