本文整理汇总了Python中Bio.SeqFeature.SeqFeature.qualifiers[key]方法的典型用法代码示例。如果您正苦于以下问题：Python SeqFeature.qualifiers[key]方法的具体用法？Python SeqFeature.qualifiers[key]怎么用？Python SeqFeature.qualifiers[key]使用的例子？那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在类Bio.SeqFeature.SeqFeature的用法示例。
在下文中一共展示了SeqFeature.qualifiers[key]方法的1个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: parse_gff
# 需要导入模块: from Bio.SeqFeature import SeqFeature [as 别名]
# 或者: from Bio.SeqFeature.SeqFeature import qualifiers[key] [as 别名]
def parse_gff(handle):
"""Quick hack to parse Bacterial GFF files from Prokka etc.
Does NOT support multi-line features (i.e. splicing and
multiple exons). Will load EVERYTHING into memory!
Iterator yielding SeqRecord objects, intended to fit into the
Biopython SeqIO structure.
"""
line = handle.readline()
assert line.startswith("##gff-version 3"), line
# print("Parsing GFF3")
references = OrderedDict()
for line in handle:
# print(line)
if line.startswith("##sequence-region "):
_, name, start, end = line.split()
assert start == "1"
references[name] = SeqRecord(
UnknownSeq(int(end)), id=name, name=name)
elif line.strip() == "##FASTA":
break
elif line.startswith("#"):
raise NotImplementedError(line)
elif line.count("\t") == 8:
seqid, source, ftype, start, end, score, strand, phase, attributes = line.split(
"\t")
assert seqid in references, "Reference %r not declared with ##sequence-region line:\n%r" % (
seqid, line)
start = int(start) - 1
end = int(end)
assert 0 <= start < end < len(references[seqid])
if ftype in FEATURE_TYPE_TO_IGNORE:
continue
if FEATURE_TYPE_WANTED and ftype not in FEATURE_TYPE_WANTED:
continue
if strand == "+":
loc = FeatureLocation(start, end, +1)
elif strand == "-":
loc = FeatureLocation(start, end, -1)
elif strand == ".":
# Unstranded - should use zero but +1 to match EMBL/GB
loc = FeatureLocation(start, end, +1)
elif strand == "?":
# Stranded by missing - should use None but +1 to match EMBL/GB
loc = FeatureLocation(start, end, +1)
else:
raise ValueError("Bad strand %r in line:\n%r" % (strand, line))
f = SeqFeature(loc, type=ftype)
for part in attributes.strip().split(";"):
if not part:
assert ";;" in line, line
sys.stderr.write(
"Warning - missing key=value or double semi-colon in line:\n%r\n" % line)
continue
if "=" not in part:
sys.exit("Bad key=value entry %r in line:\n%r" %
(part, line))
key, value = part.split("=", 1)
if key in MISSING_QUALIFIERS_TO_IGNORE:
continue
if key == "eC_number":
key = "EC_number"
value = value.replace("%2C", ",")
try:
f.qualifiers[key].append(value)
except KeyError:
f.qualifiers[key] = [value]
references[seqid].features.append(f)
else:
raise NotImplementedError(line)
# Deal with any FASTA block
name = None
seqs = []
for line in handle:
if line.startswith(">"):
if name and seqs:
seq = "".join(seqs)
assert len(seq) == len(references[name]), \
"FASTA entry for %s was %i long, expected %i" % (
name, len(seq), len(references[name]))
references[name].seq = Seq(seq)
name = line[1:].split(None, 1)[0]
seqs = []
elif name:
seqs.append(line.strip())
elif line.strip():
raise NotImplementedError(line)
if name and seqs:
seq = "".join(seqs)
assert len(seq) == len(references[name]), \
"FASTA entry for %s was %i long, expected %i" % (
name, len(seq), len(references[name]))
references[name].seq = Seq(seq)
# Return results
for name, record in references.items():
# print("%s length %i with %i features" % (name, len(record), len(record.seq)))
yield record