def main(argv):
out = sys.stdout
cds_d = {}
try:
flanksize = int(argv[1])
except:
flanksize = 300
for line in open(argv[0]):
line = line.strip().split('\t')
attributes = get_attributes(line[8])
if line[2] == 'CDS':
id_ = attributes['Parent'].split(',')[0]
cds_d[id_] = cds_d.get(id_, []) + [int(line[3]), int(line[4])]
for id_, coding_sequences in sorted(cds_d.items()):
p5, p3 = min(coding_sequences), max(coding_sequences)
line = ['Chr%c' % id_[2], 'TAIR10', 'five_prime_flank', str(max(1, p5 - flanksize)),
str(p5), '.', '.', '.', 'Parent=%s' % id_]
out.write('%s\n' % '\t'.join(line))
line = ['Chr%c' % id_[2], 'TAIR10', 'three_prime_flank', str(p3),
str(p3 + flanksize), '.', '.', '.', 'Parent=%s' % id_]
out.write('%s\n' % '\t'.join(line))
pass
def process_intersected_gff(open_fn, transcript_info, out=sys.stdout):
header = ['Contig',
'AGI', 'tRNA_located_in', 'Start_feature', 'End_feature', 'Strand',
'AGI_tRNA', 'Start_tRNA', 'End_tRNA', 'Strand_tRNA', 'Length_tRNA',
'Overlap[nt]', '%covered_tRNA', 'Transcript_found', 'Transcript_isMobile',
'Same_strand']
out.write(','.join(header) + '\n')
for line in open_fn:
line = line.strip().split('\t')
contig = line[0]
start, end = map(int, line[3:5])
strand = line[6]
attributes = get_attributes(line[8])
type_region = line[11]
start_region, end_region = map(int, line[12:14])
strand_region = line[15]
attributes_region = get_attributes(line[17])
overlap = int(line[18])
trna_length = end - start + 1
agi = attributes_region['Parent']
if agi[:-2] in transcript_info:
transcript_found = 'yes'
if transcript_info[agi[:-2]]:
transcript_is_mobile = 'yes'
else:
transcript_is_mobile = 'no'
else:
transcript_found = 'no'
transcript_is_mobile = 'unknown'
if strand == strand_region:
same_strand = 'yes'
else:
same_strand = 'no'
row = [contig,
agi, type_region, start_region, end_region, strand_region,
attributes['ID'], start, end, strand, trna_length,
overlap, int(float(overlap)/trna_length * 100 + 0.5),
transcript_found, transcript_is_mobile, same_strand]
out.write(','.join(map(str, row)) + '\n')
return None
