def tab_parser(handle, quiet=False):
from Bio.GenBank import _FeatureConsumer
from Bio.GenBank.utils import FeatureValueCleaner
def Si_parse_tab_features(object, skip=False):
"""Return list of tuples for the features (if present)
Each feature is returned as a tuple (key, location, qualifiers)
where key and location are strings (e.g. "CDS" and
"complement(join(490883..490885,1..879))") while qualifiers
is a list of two string tuples (feature qualifier keys and values).
Assumes you have already read to the start of the features table.
"""
# if object.line.rstrip() not in object.FEATURE_START_MARKERS:
# if object.debug : print "Didn't find any feature table"
# return []
#
# while object.line.rstrip() in object.FEATURE_START_MARKERS:
# object.line = object.handle.readline()
features = []
line = object.line
while True:
if not line:
break
raise ValueError("Premature end of line during features table")
if line[:object.HEADER_WIDTH].rstrip() in object.SEQUENCE_HEADERS:
if object.debug: print "Found start of sequence"
break
line = line.rstrip()
if line == "//":
raise ValueError(
"Premature end of features table, marker '//' found")
if line in object.FEATURE_END_MARKERS:
if object.debug: print "Found end of features"
line = object.handle.readline()
break
if line[2:object.FEATURE_QUALIFIER_INDENT].strip() == "":
print line[2:object.FEATURE_QUALIFIER_INDENT].strip()
raise ValueError("Expected a feature qualifier in line '%s'" %
line)
if line.split()[0] in ["ID", "source"]:
line = object.handle.readline()
continue
if skip:
line = object.handle.readline()
while line[:object.
FEATURE_QUALIFIER_INDENT] == object.FEATURE_QUALIFIER_SPACER:
line = object.handle.readline()
else:
#Build up a list of the lines making up this feature:
feature_key = line[2:object.FEATURE_QUALIFIER_INDENT].strip()
feature_lines = [line[object.FEATURE_QUALIFIER_INDENT:]]
line = object.handle.readline()
while line and (
line[:object.FEATURE_QUALIFIER_INDENT]
== object.FEATURE_QUALIFIER_SPACER
or line.rstrip() == ""
): # cope with blank lines in the midst of a feature
feature_lines.append(
line[object.FEATURE_QUALIFIER_INDENT:].rstrip())
line = object.handle.readline()
if len(line) == 0:
break #EOF
feature_lines.append('/seq="N"')
sys.stdout.flush()
features.append(
object.parse_feature(feature_key, feature_lines))
object.line = line
return features
def Si_feed(object, handle, consumer, do_features=True):
"""Feed a set of data into the consumer.
This method is intended for use with the "old" code in Bio.GenBank
Arguments:
handle - A handle with the information to parse.
consumer - The consumer that should be informed of events.
do_features - Boolean, should the features be parsed?
Skipping the features can be much faster.
Return values:
true - Passed a record
false - Did not find a record
"""
#Should work with both EMBL and GenBank files provided the
#equivalent Bio.GenBank._FeatureConsumer methods are called...
# object.set_handle(handle)
# if not object.find_start():
# #Could not find (another) record
# consumer.data=None
# print "here"
# return False
#We use the above class methods to parse the file into a simplified format.
#The first line, header lines and any misc lines after the features will be
#dealt with by GenBank / EMBL specific derived classes.
#First line and header:
# object._feed_first_line(consumer, object.line)
# object._feed_header_lines(consumer, object.parse_header())
#Features (common to both EMBL and GenBank):
if do_features:
object._feed_feature_table(
consumer, Si_parse_tab_features(object, skip=False))
else:
Si_parse_tab_features(object, skip=True) # ignore the data
#Footer and sequence
# misc_lines, sequence_string = object.parse_footer()
# object._feed_misc_lines(consumer, misc_lines)
sequence_string = "N"
consumer.sequence(sequence_string)
# Calls to consumer.base_number() do nothing anyway
consumer.record_end("//")
length = 0
for record in consumer.data.features:
if record.location.nofuzzy_end > length:
length = record.location.nofuzzy_end
consumer.data.seq = "N" * length
# assert object.line == "//"
#And we are done
return True
myscanner = Scanner.InsdcScanner()
myscanner.set_handle(handle)
myscanner.line = myscanner.handle.readline()
myscanner.FEATURE_QUALIFIER_INDENT = 21
myscanner.FEATURE_QUALIFIER_SPACER = "FT" + " " * (
myscanner.FEATURE_QUALIFIER_INDENT - 2)
myscanner.debug = True
#featuretuples=Si_parse_tab_features(myscanner)
consumer = _FeatureConsumer(use_fuzziness=1,
feature_cleaner=FeatureValueCleaner())
Si_feed(myscanner, handle, consumer)
return consumer.data
def tab_parser(handle, quiet=False):
def Drawer_parse_tab_features(object, skip=False):
features = []
line = object.line
while True:
if not line:
break
raise ValueError("Premature end of line during features table")
if line[:object.HEADER_WIDTH].rstrip() in object.SEQUENCE_HEADERS:
if object.debug: print("Found start of sequence")
break
line = line.rstrip()
if line == "//":
raise ValueError(
"Premature end of features table, marker '//' found")
if line in object.FEATURE_END_MARKERS:
if object.debug: print("Found end of features")
line = object.handle.readline()
break
if line[2:object.FEATURE_QUALIFIER_INDENT].strip() == "":
print(line[2:object.FEATURE_QUALIFIER_INDENT].strip())
raise ValueError("Expected a feature qualifier in line '%s'" %
line)
if skip:
line = object.handle.readline()
while line[:object.
FEATURE_QUALIFIER_INDENT] == object.FEATURE_QUALIFIER_SPACER:
line = object.handle.readline()
else:
#Build up a list of the lines making up this feature:
feature_key = line[2:object.FEATURE_QUALIFIER_INDENT].strip()
feature_lines = [line[object.FEATURE_QUALIFIER_INDENT:]]
line = object.handle.readline()
while line[:object.
FEATURE_QUALIFIER_INDENT] == object.FEATURE_QUALIFIER_SPACER or line.rstrip(
) == "": # cope with blank lines in the midst of a feature
feature_lines.append(
line[object.FEATURE_QUALIFIER_INDENT:].rstrip())
line = object.handle.readline()
if len(line) == 0:
break #EOF
feature_lines.append('/seq="N"')
sys.stdout.flush()
features.append(
object.parse_feature(feature_key, feature_lines))
object.line = line
return features
def Drawer_feed(object, handle, consumer, do_features=True):
if do_features:
object._feed_feature_table(
consumer, Drawer_parse_tab_features(object, skip=False))
else:
Drawer_parse_tab_features(object, skip=True) # ignore the data
sequence_string = "N"
consumer.sequence(sequence_string)
consumer.record_end("//")
length = 0
for record in consumer.data.features:
if record.location.nofuzzy_end > length:
length = record.location.nofuzzy_end
consumer.data.seq = "N" * length
return True
myscanner = Scanner.InsdcScanner()
myscanner.set_handle(handle)
myscanner.line = myscanner.handle.readline()
myscanner.FEATURE_QUALIFIER_INDENT = 21
myscanner.FEATURE_QUALIFIER_SPACER = "FT" + " " * (
myscanner.FEATURE_QUALIFIER_INDENT - 2)
myscanner.debug = True
consumer = _FeatureConsumer(use_fuzziness=1,
feature_cleaner=FeatureValueCleaner())
Drawer_feed(myscanner, handle, consumer)
return consumer.data