coolneng
/
adapter-parser
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Compare commits

base: coolneng:7ed975c7ce5688901a60667f34e42d4b60dde738
Branches Tags
coolneng:master
..
compare: coolneng:e826d6f92b53a3e9f57f1327d54c6ea9cadcb310
Branches Tags
coolneng:master

No commits in common. "7ed975c7ce5688901a60667f34e42d4b60dde738" and "e826d6f92b53a3e9f57f1327d54c6ea9cadcb310" have entirely different histories.
7ed975c7ce ... e826d6f92b

1 changed files with 6 additions and 5 deletions
Show Stats Expand all files Collapse all files

11
parser.py
View File

@ -6,9 +6,9 @@ from re import sub
from pandas import DataFrame, read_html, Series from pandas import DataFrame, read_html, Series
def remove_parenthesis(identifier): def find_html_files(path) -> List:
sequence_str = "".join(map(str, identifier)) file_list = glob(path + "/*fastqc.html")
return sub(r"[()]", "", sequence_str) return file_list
def extract_adapters(files) -> Tuple[Series, List]: def extract_adapters(files) -> Tuple[Series, List]:
@ -36,7 +36,8 @@ def preprocess_dataframe(adapters) -> Series:
def save_to_file(filename, adapters) -> None: def save_to_file(filename, adapters) -> None:
with open(filename, "w") as f: with open(filename, "w") as f:
for index, value in adapters.iteritems(): for index, value in adapters.iteritems():
sequence_id = remove_parenthesis(index) sequence_str = "".join(map(str, index))
sequence_id = sub(r"[()]", "", sequence_str)
fasta_entry = f">{sequence_id}\n{value}\n" fasta_entry = f">{sequence_id}\n{value}\n"
f.write(fasta_entry) f.write(fasta_entry)
@ -50,7 +51,7 @@ def parse_arguments():
def main(): def main():
args = parse_arguments() args = parse_arguments()
file_list = glob(args.input + "/*fastqc.html") file_list = find_html_files(args.input)
adapters, stats = extract_adapters(file_list) adapters, stats = extract_adapters(file_list)
save_to_file(args.output, adapters) save_to_file(args.output, adapters)
print( print(