Mercurial > repos > recetox > rem_complex

annotate rem_complex.py @ 0:a0e07a0bc047 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
author recetox
date Mon, 27 Nov 2023 09:04:04 +0000
parents
children 567327a97ad2
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
1 import argparse
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
2
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
3 import pandas as pd
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
4 from openbabel import openbabel, pybel
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
5 openbabel.obErrorLog.SetOutputLevel(1) # 0: suppress warnings; 1: warnings
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
6
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
7
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
8 def parse_arguments() -> argparse.Namespace:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
9 parser = argparse.ArgumentParser()
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
10 parser.add_argument('-iformat', '--input_format', help='Input file format')
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
11 parser.add_argument('-i', '--input_filename', type=str, required=True, help='Input file name')
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
12 parser.add_argument('-o', '--output_filename', type=str, required=True, help='Outout file name')
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
13 args = parser.parse_args()
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
14 return args
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
15
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
16
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
17 def filter_csv_molecules(file_name: str, output_file_name: str) -> None:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
18 """Removes molecules with '.' in SMILES string from csv file.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
19
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
20 Args:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
21 file_name (str): Path to csv file that contains metadata.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
22 output_file_name (str): Path to destination file, in csv format.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
23 """
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
24 df = pd.read_csv(file_name)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
25 mask = df['smiles'].str.contains(".", na=False, regex=False)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
26 mask = mask.apply(lambda x: not x)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
27 df[mask].to_csv(output_file_name, index=False)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
28
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
29
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
30 def filter_other_format_molecules(file_name: str, output_file_name: str, input_format: str) -> None:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
31 """Removes molecules with '.' in SMILES string from smi or inchi files.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
32
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
33 Args:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
34 file_name (str): Path to smi or inchi files.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
35 output_file_name (str): Path to destination files, in smi or inchi formats.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
36 input_format (str): Input file format.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
37 """
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
38 molecules = list(pybel.readfile(input_format, file_name))
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
39 filtered_molecules = [mol for mol in molecules if "." not in mol.write('smi').strip()]
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
40
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
41 with open(output_file_name, 'w') as f:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
42 for mol in filtered_molecules:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
43 f.write(mol.write(input_format))
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
44
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
45
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
46 def filter_complex_molecules(file_name: str, output_file_name: str, input_format: str) -> None:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
47 """Removes molecular complexes depending on the input format.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
48
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
49 Args:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
50 file_name (str): Path to csv, smi or inchi files
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
51 output_file_name (str): Path to destination files, in csv. smi or inchi formats.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
52 input_format (str): Input file formats.
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
53 """
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
54 if input_format == 'csv':
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
55 filter_csv_molecules(file_name, output_file_name)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
56 else:
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
57 filter_other_format_molecules(file_name, output_file_name, input_format)
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
58
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
59
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
60 if __name__ == "__main__":
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
61 args = parse_arguments()
a0e07a0bc047 planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/rem_complex commit 833f7671d1e1b713d52ba5c7e59d28be38b92b1e
recetox
parents:
diff changeset
62 filter_complex_molecules(args.input_filename, args.output_filename, args.input_format)