def run(self):
gff3Data = gff3.Gff3Parser(self.gff3File).parse()
dbconn = sqlite3.connect(self.dbFile)
dbcur = dbconn.cursor()
dbcur.execute(_dbTableSQL) # create table
dbcur.execute("PRAGMA journal_mode=WAL")
dbconn.commit()
for i, featureName in enumerate(gff3Data.byFeatureName.keys()):
for feature in gff3Data.byFeatureName[featureName]:
# Ignores any parent IDs besides the first one.
parentIds = [parent.uniqueId for parent in feature.parents]
parentId = parentIds[0] if len(parentIds) > 0 else ''
# FIXME: No current code to ensure childIDs in correct order.
childIds = [child.uniqueId for child in feature.children]
values = (feature.uniqueId, parentId, _db_serialize(childIds),
feature.seqname, feature.source, feature.type,
feature.start, feature.end, feature.score,
feature.strand, feature.featureName,
feature.attributes.get("gene_name", [None])[0],
feature.attributes.get("transcript_name", [None])[0],
_db_serialize(feature.attributes))
self._batchInsertValues(values, dbcur, dbconn)
self._insertValues(dbcur, dbconn)
dbcur.execute(("create INDEX idx1 "
"on feature(start, end, reference_name)"))
dbcur.execute(
"CREATE INDEX name_type_index ON FEATURE (gene_name, type)")
dbcur.execute("PRAGMA INDEX_LIST('feature')")
dbcur.close()
dbconn.close()
def setUp(self):
testDataFile = _testDataDir + "specialCasesTest.gff3"
self.gff3Parser = gff3.Gff3Parser(testDataFile)
self.gff3Data = self.gff3Parser.parse()
def setUp(self):
testDataFile = _testDataDir + "sacCerTest.gff3"
self.gff3Parser = gff3.Gff3Parser(testDataFile)
self.gff3Data = self.gff3Parser.parse()
def setUp(self):
testDataFile = _testDataDir + "discontinuous.gff3"
self.gff3Parser = gff3.Gff3Parser(testDataFile)
self.gff3Data = self.gff3Parser.parse()
def setUp(self):
testDataFile = _testDataDir + "gencodeV21Set1.gff3"
self.gff3Parser = gff3.Gff3Parser(testDataFile)
self.gff3Data = self.gff3Parser.parse()