Mercurial > repos > recetox > matchms_filtering
annotate matchms_split.py @ 23:83c4b4359ac4 draft default tip
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 129488bfd91735fc55fe32edb1079c24eee7b2d3
author | recetox |
---|---|
date | Thu, 14 Nov 2024 15:08:10 +0000 |
parents | 6161895506ff |
children |
rev | line source |
---|---|
5
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
1 import argparse |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
2 import itertools |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
3 import os |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
4 |
15
6161895506ff
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit da193865f41a3a840ecc4ba0afab1d358554998a
recetox
parents:
5
diff
changeset
|
5 import matchms |
5
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
6 from matchms.exporting import save_as_msp |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
7 from matchms.importing import load_from_msp |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
8 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
9 |
15
6161895506ff
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit da193865f41a3a840ecc4ba0afab1d358554998a
recetox
parents:
5
diff
changeset
|
10 matchms.Metadata.set_key_replacements({}) |
5
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
11 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
12 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
13 def make_outdir(outdir: str): |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
14 """Create destination directory. |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
15 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
16 Args: |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
17 outdir (str): Path to destination directory where split spectra files are generated. |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
18 """ |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
19 return os.mkdir(outdir) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
20 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
21 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
22 def write_spectra(spectra, outdir): |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
23 """Generates MSP files of individual spectra. |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
24 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
25 Args: |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
26 spectra (List[Spectrum]): Spectra to write to file |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
27 outdir (str): Path to destination directory. |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
28 """ |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
29 for i in range(len(spectra)): |
15
6161895506ff
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit da193865f41a3a840ecc4ba0afab1d358554998a
recetox
parents:
5
diff
changeset
|
30 save_as_msp(spectra[i], os.path.join(outdir, f"{i}.msp")) |
5
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
31 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
32 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
33 def split_round_robin(iterable, num_chunks): |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
34 chunks = [list() for _ in range(num_chunks)] |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
35 index = itertools.cycle(range(num_chunks)) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
36 for value in iterable: |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
37 chunks[next(index)].append(value) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
38 chunks = filter(lambda x: len(x) > 0, chunks) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
39 return chunks |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
40 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
41 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
42 listarg = argparse.ArgumentParser() |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
43 listarg.add_argument('--filename', type=str) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
44 listarg.add_argument('--method', type=str) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
45 listarg.add_argument('--outdir', type=str) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
46 listarg.add_argument('--parameter', type=int) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
47 args = listarg.parse_args() |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
48 outdir = args.outdir |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
49 filename = args.filename |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
50 method = args.method |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
51 parameter = args.parameter |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
52 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
53 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
54 if __name__ == "__main__": |
15
6161895506ff
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit da193865f41a3a840ecc4ba0afab1d358554998a
recetox
parents:
5
diff
changeset
|
55 spectra = load_from_msp(filename, metadata_harmonization=False) |
5
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
56 make_outdir(outdir) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
57 |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
58 if method == "one-per-file": |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
59 write_spectra(list(spectra), outdir) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
60 else: |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
61 if method == "chunk-size": |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
62 chunks = iter(lambda: list(itertools.islice(spectra, parameter)), []) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
63 elif method == "num-chunks": |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
64 chunks = split_round_robin(spectra, parameter) |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
65 for i, x in enumerate(chunks): |
a177ac3c752c
planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit f79a5b51599254817727bc9028b9797ea994cb4e
recetox
parents:
diff
changeset
|
66 save_as_msp(x, os.path.join(outdir, f"chunk_{i}.msp")) |