def getSpliceEvent(defString, location):
defString = defString.lower()
# grab location information
try:
locationList = [int(location["GBInterval_from"])-1, int(location["GBInterval_to"])]
except KeyError:
# some regions are defined as a single point
locationList = [int(location["GBInterval_point"])-1, int(location["GBInterval_point"])]
# if this defString defines a missing event
if defString.find("missing") > -1:
return["missing", locationList[0], locationList[1], "", ""]
# if the defstring defines a replacement event
if defString.find("->") > -1:
# to get the string of AA either side of -> we have to do some
# quick regexs/reformatting
defString = defString.replace(" ","")
old = str(defString.split("->")[0])
new = re.search("[a-zA-Z]*", defString.split("->")[1]).group()
return ["replacement", locationList[0], locationList[1], old, new]
########################################
# if we get here this XML is, technically, maformed. HOWEVER, we can try and salvage by adding
# 'exception' handling
show_warning("The isform line \n\n[" + defString + "] \nis malformed\n\nTrying to parse regardless...\n")
# Exception 1 use of ">" instead of "->"
if defString.find(">") > -1:
# to get the string of AA either side of -> we have to do some
# quick regexs/reformatting
defString = defString.replace(" ","")
old = str(defString.split(">")[0])
new = re.search("[a-zA-Z]*", defString.split(">")[1]).group()
return ["replacement", locationList[0], locationList[1], old, new]
raise ProteinObjectException("Isoform 'note string' had neither 'missing' nor '->' in it, and\neven specific known exception handling couldn't help!\nParse error!")
def _xml_is_OK(self, proteinxml):
if len(proteinxml) > 1:
show_warning(" [ProteinObject._xml_is_ok()] - ProteinXML detected more than one record associated with this GI.\nThis should never happen.")
return False
# Nothing in XML - so return an empty-initiailized object with exists = 0
if len(proteinxml) == 0:
return False
# Check that we're really dealing with protein (despite specifying db="protein"
# on the efetch call, when a GI is used other databases seem to be searched too)
try:
if not (proteinxml[0]["GBSeq_moltype"] == "AA"):
return False
except TypeError:
return False
return True
isoformSequenceList_keys = isoformSequenceList.keys()
nametoIsoID_keys = nametoIsoID.keys()
# remove 1
try:
nametoIsoID_keys.remove("1")
except Exception:
pass
if len(set(isoformSequenceList_keys + nametoIsoID_keys))-len(nametoIsoID_keys) == 1:
for i in nametoIsoID_keys:
if i not in isoformSequenceList_keys:
try:
isoformReturnVal[nametoIsoID[i]] = missing_data
show_warning("Malformed isoform data lead to an inconsistency, but the gremlins think they fixed it... [ID=" + self.accession + "]")
except KeyError:
show_warning("Malformed isoform data has lead to an inconsistency - skipping that isoform [ID=" + self.accession + "]")
# OH COME ON!!
if salvage > 1:
show_warning("Malformed isoform data has lead to an inconsistency - skipping several isoforms [ID=" + self.accession + "]")
return isoformReturnVal
