def _parse_feature(element):
feature = SeqFeature.SeqFeature()
for k, v in element.attrib.items():
feature.qualifiers[k] = v
feature.type = element.attrib.get('type', '')
if 'id' in element.attrib:
feature.id = element.attrib['id']
for feature_element in element:
if feature_element.tag == NS + 'location':
position_elements = feature_element.findall(NS + 'position')
if position_elements:
element = position_elements[0]
start_position = _parse_position(element, -1)
end_position = _parse_position(element)
else:
element = feature_element.findall(NS + 'begin')[0]
start_position = _parse_position(element, -1)
element = feature_element.findall(NS + 'end')[0]
end_position = _parse_position(element)
feature.location = SeqFeature.FeatureLocation(start_position, end_position)
else:
try:
feature.qualifiers[feature_element.tag.replace(NS, '')] = feature_element.text
except:
pass # skip unparsable tag
self.ParsedSeqRecord.features.append(feature)
def _make_seqfeature(name, from_res, to_res, description, ft_id):
"""Construct SeqFeature from feature data from parser (PRIVATE)."""
loc = SeqFeature.FeatureLocation(_make_position(from_res, -1),
_make_position(to_res, 0))
if not ft_id:
ft_id = "<unknown id>" # The default in SeqFeature object
return SeqFeature.SeqFeature(loc,
type=name,
id=ft_id,
qualifiers={"description": description})
def _parse_dbReference(element):
self.ParsedSeqRecord.dbxrefs.append(element.attrib['type'] + ':' + element.attrib['id'])
#e.g.
# <dbReference type="PDB" key="11" id="2GEZ">
# <property value="X-ray" type="method"/>
# <property value="2.60 A" type="resolution"/>
# <property value="A/C/E/G=1-192, B/D/F/H=193-325" type="chains"/>
# </dbReference>
if 'type' in element.attrib:
if element.attrib['type'] == 'PDB':
method = ""
resolution = ""
for ref_element in element:
if ref_element.tag == NS + 'property':
dat_type = ref_element.attrib['type']
if dat_type == 'method':
method = ref_element.attrib['value']
if dat_type == 'resolution':
resolution = ref_element.attrib['value']
if dat_type == 'chains':
pairs = ref_element.attrib['value'].split(',')
for elem in pairs:
pair = elem.strip().split('=')
if pair[1] != '-':
#TODO - How best to store these, do SeqFeatures make sense?
feature = SeqFeature.SeqFeature()
feature.type = element.attrib['type']
feature.qualifiers['name'] = element.attrib['id']
feature.qualifiers['method'] = method
feature.qualifiers['resolution'] = resolution
feature.qualifiers['chains'] = pair[0].split('/')
start = int(pair[1].split('-')[0]) - 1
end = int(pair[1].split('-')[1])
feature.location = SeqFeature.FeatureLocation(start, end)
#self.ParsedSeqRecord.features.append(feature)
for ref_element in element:
if ref_element.tag == NS + 'property':
pass # this data cannot be fitted in a seqrecord object with a simple list. however at least ensembl and EMBL parsing can be improved to add entries in dbxrefs