def build_ensembl_transcripts_by_genomic_location_index(self, ensembl_transcript_index_fname, output_filename, protocol="file"):
"""Create an index for genomic position to transcripts index, using a transcript index created in
build_ensembl_transcript_index
"""
transcript_db = Shove(protocol + "://" + ensembl_transcript_index_fname)
output_db = Shove(protocol + "://" + output_filename, optimize=False)
transcript_keys = transcript_db.keys()
for i,tx_id in enumerate(transcript_keys):
tx = transcript_db[tx_id]
start = tx.get_start()
end = tx.get_end()
genomic_location_bin = region2bin(start, end)
key = tx.get_contig() + "_" + str(genomic_location_bin)
try:
tmpList = output_db[key]
except KeyError:
output_db[key] = []
tmpList = output_db[key]
tmpList.append(tx)
output_db[key] = tmpList
if (i+1) % 10000 == 0:
logging.getLogger(__name__).info("Genomic position index added " + str(i) + " transcripts so far.")
output_db.close()
transcript_db.close()
def test_region2bin(self):
"""Simple test that the region2bin works for genomic position indexing """
# Footprint for PIK3CA transcript chr3:178,866,311-178,952,497 uc003fjk.3
guess = region2bin(178866311, 178952497)
self.assertTrue(guess == 243)
def test_region2bin(self):
"""Simple test that the region2bin works for genomic position indexing """
# Footprint for PIK3CA transcript chr3:178,866,311-178,952,497 uc003fjk.3
guess = region2bin(178866311, 178952497)
self.assertTrue(guess == 243)
def build_ensembl_transcripts_by_genomic_location_index(
self,
ensembl_transcript_index_fname,
output_filename,
protocol="file"):
"""Create an index for genomic position to transcripts index, using a transcript index created in
build_ensembl_transcript_index
"""
transcript_db = Shove(protocol + "://" +
ensembl_transcript_index_fname)
output_db = Shove(protocol + "://" + output_filename, optimize=False)
transcript_keys = transcript_db.keys()
for i, tx_id in enumerate(transcript_keys):
tx = transcript_db[tx_id]
start = tx.get_start()
end = tx.get_end()
genomic_location_bin = region2bin(start, end)
key = tx.get_contig() + "_" + str(genomic_location_bin)
try:
tmpList = output_db[key]
except KeyError:
output_db[key] = []
tmpList = output_db[key]
tmpList.append(tx)
output_db[key] = tmpList
if (i + 1) % 10000 == 0:
logging.getLogger(
__name__).info("Genomic position index added " + str(i) +
" transcripts so far.")
output_db.close()
transcript_db.close()