report_apc

#!/usr/bin/env python3

import argparse
import gzip
import isoform

# CLI

parser = argparse.ArgumentParser(description='apc gene reporter')
parser.add_argument('apc', help='apc directory')
parser.add_argument('models', help='models directory')
parser.add_argument('isos', help='file of isoformer performances')
parser.add_argument('opts', help='file of optiso performances')
arg = parser.parse_args()

# Collect data from files
isos = []
with gzip.open(arg.isos, 'rt') as fp:
	for line in fp: isos.append(round(float(line), 4))


opts = []
perfs = []
dons = []
accs = []
emms = []
imms = []
elens = []
ilens = []
icosts = []
names = []
with gzip.open(arg.opts, 'rt') as fp:
	for line in fp:
		perf, don, acc, emm, imm, elen, ilen, icost, name = line.split()
		opts.append(float(perf))
		dons.append(float(don))
		accs.append(float(acc))
		emms.append(float(emm))
		imms.append(float(imm))
		elens.append(float(elen))
		ilens.append(float(ilen))
		icosts.append(float(icost))
		names.append(name)

lens = []
for name in names:
	defline, seq = next(isoform.read_fasta(f'{arg.apc}/{name}.fa'))
	lens.append(len(seq))

for name, le, opt, iso in zip(names, lens, opts, isos):
	print(name, le, opt, iso, sep='\t')



"""

params = {
	'--min_exon': 25,
	'--min_intron': 35,
	'--flank': 99,
	'--limit': 100,
	'--apwm': f'{os.path.abspath(arg.models)}/acc.pwm',
	'--dpwm': f'{os.path.abspath(arg.models)}/don.pwm',
	'--emm': f'{os.path.abspath(arg.models)}/exon.mm',
	'--imm': f'{os.path.abspath(arg.models)}/intron.mm',
	'--elen': f'{os.path.abspath(arg.models)}/exon.len',
	'--ilen': f'{os.path.abspath(arg.models)}/intron.len',
}
"""