def parse_ref_exons(self):
""" Return fasta reference with only the sequences needed"""
ens_db = EnsemblRelease(75)
try:
exons = ens_db.exons_at_locus(self.chrom, self.start, self.stop)
except ValueError as e:
# Load pyensembl db
raise e
exon_array = np.zeros(self.stop - self.start)
exon_numbers = self.get_exon_numbers(ens_db, exons[0].gene_name)
for exobj in exons:
start = exobj.start - self.start
stop = exobj.end - self.start
i = start
while i < stop:
exon_array[i] = 1
i += 1
# 2:29,448,326-29,448,432 exon 19
# exon 22 start: 29445210
# exon 18 end: 29449940
# intron 19: 29446395-29448326
# ATI initiation 29446768-29448326
return exon_array, exon_numbers[(exon_numbers['start'] > self.hist[0][0]) &
(exon_numbers['start'] < self.hist[0][-1])]
