def actual_phage_cds(gbkf, verbose=False):
"""
Read the genbank file and return a list of features that are actually phage regions
:param gbkf: the test genbank file with CDS marked with is_phage
:param verbose: more output
:return: a set of phage features
"""
if verbose:
sys.stderr.write(f"Reading {gbkf}\n")
phage = {}
nonphage = {}
for seq in genbank_seqio(gbkf):
for feat in seq.features:
if feat.type == 'CDS':
if 'product' not in feat.qualifiers:
feat.qualifiers['product'] = [
f"Hypothetical protein (not annotated in {gbkf})"
]
if 'is_phage' in feat.qualifiers:
phage[str(feat.translate(seq, cds=False).seq).upper(
)] = feat.qualifiers['product'][0]
else:
nonphage[str(feat.translate(seq, cds=False).seq).upper(
)] = feat.qualifiers['product'][0]
return phage, nonphage
def predicted_genbank(predf, verbose=False):
"""
Read the predictions from the genbank file and return
a set of features
:param predf: the predictions file
:param verbose: more output
:return: a set of predicted phage genes
"""
if verbose:
sys.stderr.write(f"Reading {predf}\n")
predicted = {}
for seq in genbank_seqio(predf):
for feat in seq.features:
if feat.type == 'CDS':
if 'product' in feat.qualifiers:
predicted[str(feat.translate(seq, cds=False).seq).upper(
)] = feat.qualifiers['product'][0]
else:
predicted[str(feat.translate(seq, cds=False).seq).upper()] = \
f"Hypothetical protein (not annotated in {predf})"
if verbose:
sys.stderr.write(
f"Found {len(predicted)} predicted prophage features\n")
return predicted
def predicted_regions(regf, gbkf, verbose):
"""
Pull the phage genes from the regions
:param regf: the regions file with contigs/start/stop
:param gbkf: the genbank file used to make those predictions
:param verbose: more output
:return: a set of predicted phage genes
"""
regions = {}
if verbose:
sys.stderr.write(f"Reading {regf}\n")
with open(regf, 'r') as f:
for l in f:
p = l.strip().split("\t")
assert (len(p) == 3
), f"Expected a tple of [contig, start, stop] in {regf}"
p[1] = int(p[1])
p[2] = int(p[2])
# print("Found a region predicted from {} to {} into {} ".format(p[1],p[2],p[0]))
if p[0] not in regions:
regions[p[0]] = []
if p[2] < p[1]:
regions[p[0]].append([p[2], p[1]])
else:
regions[p[0]].append([p[1], p[2]])
if verbose:
sys.stderr.write(f"Reading {gbkf} again to get the phage regions\n")
predicted = {}
for seq in genbank_seqio(gbkf):
# print("Now testing {}".format(seq))
if seq.id in regions:
for loc in regions[seq.id]:
if verbose:
sys.stderr.write(f"Getting from {loc[0]} to {loc[1]}\n")
for feat in seq[loc[0]:loc[1]].features:
if feat.type == 'CDS':
if 'product' in feat.qualifiers:
predicted[str(feat.translate(seq[loc[0]:loc[1]], cds=False).seq).upper()] = \
feat.qualifiers['product'][0]
else:
predicted[str(feat.translate(seq[loc[0]:loc[1]], cds=False).seq).upper()] = \
f"Hypothetical protein (not annotated in {gbkf})"
if verbose:
sys.stderr.write(
f"Found {len(predicted)} predicted prophage features\n")
return predicted
def genbank_to_pandas(gbkf,
mincontiglen,
ignorepartials=True,
convert_selenocysteine=False):
"""
This is a bit of a specific format used by phage_boost. its a simple dataframe with a couple of
additional columns:
['contig',
'id',
'start',
'stop',
'direction',
'partial',
'DNAseq',
'AAseq',
'header']
:param mincontiglen: minimum contig length
:param ignorepartials: Ignore any gene call with a frameshift (ie. a stop codon in the middle of the sequence)
:param convert_selenocysteine: PhageBoost crashes with a selenocysteine protein because it is not in Biopython
:param gbkf: Genbank file to parse
:return: a pandas data frame
"""
c = 0
genes = []
for seq in genbank_seqio(gbkf):
if len(seq) < mincontiglen:
sys.stderr.write(
f"Skipped {seq.id} because it's length ({len(seq)}) is less than the "
+ "minimum contig length ({mincontiglen})\n")
continue
for feat in seq.features:
if feat.type != 'CDS':
continue
tid = seq.id + "_" + str(c)
partial = 0
# I don't think this is exactly right
if 'truncated' in feat.qualifiers:
partial = 1
dnaseq = str(feat.extract(seq).seq)
if len(dnaseq) == 0:
sys.stderr.write(
f"The DNA sequence for {feature_id(seq, feat)} was zero, so skipped\n"
)
continue
# we just do a de novo translation rather than relying on the translation provided
# in the genbank file that is often wrong
trans = str(feat.extract(seq).translate().seq)
while trans.endswith('*'):
trans = trans[:-1]
# Partial amino acid codes we should ignore. These are not present in BioPython's SeqUtils::ProtParam
# and it crashes the system
paa = {'B', 'Z', 'J', 'X', '*'}
keeporf = True
if ignorepartials:
for aa in paa:
if aa in trans:
sys.stderr.write(
f"There is a {aa} in {feature_id(seq, feat)} so skipped.\n"
)
keeporf = False
if not keeporf:
continue
if len(trans) == 0:
sys.stderr.write(
f"The translation for {feature_id(seq, feat)} was zero, so skipped.\n"
)
continue
if convert_selenocysteine:
trans = trans.replace('U', 'C')
row = [
seq.id, c, feat.location.start.position,
feat.location.end.position, feat.strand, partial, dnaseq,
trans, tid
]
c += 1
genes.append(row)
gc = pd.DataFrame(genes,
columns=[
'contig', 'id', 'start', 'stop', 'direction',
'partial', 'DNAseq', 'AAseq', 'header'
])
return gc