def getProteinGenbankDescription(self,directory,proteinID):
fname = "%s/%s.gbk"%(directory,proteinID)
seq_record = SeqIO.parse(open(fname), "genbank").next()
note = ""
for feature in seq_record.features:
try:
if "note" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["note"][0])
if "function" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["function"][0])
if "product" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["product"][0])
except KeyError as k:
continue
return note
def buildProteinTable(self):
for genbank_file in self.genbank_files:
seq_record = SeqIO.parse(open(genbank_file), "genbank").next()
for feature in seq_record.features:
try:
proteinID = feature.qualifiers["protein_id"][0]
note = ""
if "note" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["note"][0])
if "function" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["function"][0])
if "product" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["product"][0])
self.proteinDict[proteinID] = note
except KeyError as k:
continue
def insertGenbankProteins(genbank_files,dbout):
db = sqlite3.connect(dbout)
cursor = db.cursor()
for genbank_file in genbank_files:
seq_record = SeqIO.parse(open(genbank_file), "genbank").next()
for feature in seq_record.features:
try:
proteinID = feature.qualifiers["protein_id"][0]
note = ""
if "note" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["note"][0])
if "function" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["function"][0])
if "product" in feature.qualifiers:
note+= text.formatText(feature.qualifiers["product"][0])
cursor.execute('''INSERT INTO protein_text(protein_id,note)
VALUES(?,?)''',(proteinID,note))
except KeyError as k:
continue
db.commit()
db.close()
def filterWords(self, words):
# Remove all single letter words
return [text.formatText(w) for w in words if len(w) > 1]