def create_fasta_from_fastq(fastq_filename: str):
out_lines = []
rules = FastDNARules()
with open(fastq_filename, "r") as in_file:
lines = in_file.readlines()[1:][::4]
for line in lines:
err = rules.apply_all_rules(line[18:182])
if err < 1.0:
out_lines.append(line[18:182])
with open("R:/out.fasta", "w") as out_file:
for line in out_lines:
out_file.write(">todo\n")
out_file.write(line + "\n")
return lines
def fix_(in_file, out_file_str):
correct = []
rule = FastDNARules()
with open(in_file, "r") as inf:
lines = inf.readlines()
for line in lines[1::2]:
line = line.strip()
if len(line) != 164:
continue
err_prob = rule.apply_all_rules(line)
if err_prob < 1.0:
correct.append((line, err_prob))
with open(out_file_str, "w") as out_file:
for line, err_prob in correct:
out_file.write(f">%s\n" % err_prob)
out_file.write(line.strip().replace("\n", "") + "\n")
cleaned = find_dup_ids(out_file_str)
with open(out_file_str, "w") as out_file:
for line in cleaned:
out_file.write(f">abc\n")
out_file.write(line.strip().replace("\n", "") + "\n")