genmut

#!/usr/bin/env python3
import argparse
import os
from readfasta import read_record

parser = argparse.ArgumentParser(description='Simulate splice-site-generating point mutaitons')
parser.add_argument('fasta', type=str, metavar='<fasta>', help='input path to fasta file')
parser.add_argument('-o', type=str, metavar='<out_dir>', required=False, help='output dir')
arg = parser.parse_args()


for idn, seq in read_record(arg.fasta):
	file_name = os.path.basename(arg.fasta)
	base_name = os.path.splitext(file_name)[0]
	OUT_DIR='.'
	if arg.o: OUT_DIR = arg.o
	os.makedirs(f'{OUT_DIR}/{base_name}', exist_ok=True)
	count = 1
	for i in range(0, len(seq)-1, 2):
		# donor site
			cur_id  = None
			cur_seq = None
			if seq[i] == 'G' and seq[i+1] != 'T':
				cur_id  = f'{idn} {i+1} d {seq[i+1]}~T'
				cur_seq = f'{seq[:i+1]}T{seq[i+2:]}'
			if seq[i] != 'G' and seq[i+1] == 'T':
				cur_id  = f'{idn} {i} d {seq[i]}~G'
				cur_seq = f'{seq[:i]}G{seq[i+1:]}'
		# acceptor site
			if seq[i] == 'A' and seq[i+1] != 'G':
				cur_id  = f'{idn} {i+1} a {seq[i+1]}~G'
				cur_seq = f'{seq[:i+1]}G{seq[i+2:]}'
			if seq[i] != 'A' and seq[i+1] == 'G':
				cur_id  = f'{idn} {i} a {seq[i]}~A'
				cur_seq = f'{seq[:i]}A{seq[i+1:]}'
		# out
			if cur_id is not None:
				with open(f'{OUT_DIR}/{base_name}/mut_{count}_{base_name}.fa', 'w') as fh:
					fh.write(f'>{cur_id}\n')
					for i in range(0, len(cur_seq), 80):
						fh.write(cur_seq[i:i+80])
				count+=1
				
				#print(f'>{cur_id}')
				#for i in range(0, len(cur_seq), 80):
					#print(cur_seq[i:i+80])