msfiddle 2.2.1__tar.gz → 2.4.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (58) hide show
  1. {msfiddle-2.2.1 → msfiddle-2.4.0}/PKG-INFO +18 -10
  2. {msfiddle-2.2.1 → msfiddle-2.4.0}/README.md +15 -7
  3. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/__init__.py +2 -0
  4. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/api.py +485 -125
  5. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_casmi.yml +11 -10
  6. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_demo.yml +12 -10
  7. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_demo_wo_da.yml +12 -10
  8. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_embl.yml +29 -22
  9. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_orbitrap.yml +18 -19
  10. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/config/fiddle_tcn_qtof.yml +24 -26
  11. msfiddle-2.4.0/msfiddle/config/fiddle_tcn_timstof.yml +142 -0
  12. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/dataset.py +13 -5
  13. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/demo/input_msms.mgf +21 -21
  14. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/download.py +59 -81
  15. msfiddle-2.4.0/msfiddle/evaluate.py +378 -0
  16. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/external_results.py +42 -6
  17. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/main.py +98 -10
  18. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/model_tcn.py +17 -0
  19. msfiddle-2.4.0/msfiddle/settings.py +75 -0
  20. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/utils/__init__.py +6 -0
  21. msfiddle-2.4.0/msfiddle/utils/casmi2026.py +231 -0
  22. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/utils/mol_utils.py +0 -23
  23. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/utils/msms_utils.py +220 -74
  24. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/utils/pkl_utils.py +59 -48
  25. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/utils/refine_utils.py +1 -49
  26. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/PKG-INFO +18 -10
  27. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/SOURCES.txt +7 -1
  28. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/requires.txt +2 -2
  29. {msfiddle-2.2.1 → msfiddle-2.4.0}/setup.py +3 -3
  30. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_api.py +122 -1
  31. msfiddle-2.4.0/tests/test_casmi2026.py +305 -0
  32. msfiddle-2.4.0/tests/test_config.py +35 -0
  33. msfiddle-2.4.0/tests/test_download.py +93 -0
  34. msfiddle-2.4.0/tests/test_evaluate.py +185 -0
  35. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_external_results.py +46 -0
  36. msfiddle-2.4.0/tests/test_main.py +79 -0
  37. msfiddle-2.4.0/tests/test_msms_utils.py +208 -0
  38. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_pkl_utils.py +22 -1
  39. msfiddle-2.4.0/tests/test_validate.py +154 -0
  40. msfiddle-2.2.1/msfiddle/evaluate.py +0 -152
  41. msfiddle-2.2.1/tests/test_download.py +0 -27
  42. msfiddle-2.2.1/tests/test_evaluate.py +0 -62
  43. msfiddle-2.2.1/tests/test_main.py +0 -43
  44. msfiddle-2.2.1/tests/test_msms_utils.py +0 -53
  45. {msfiddle-2.2.1 → msfiddle-2.4.0}/LICENSE +0 -0
  46. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/__main__.py +0 -0
  47. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/demo/output_buddy.csv +0 -0
  48. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/demo/output_fiddle.csv +0 -0
  49. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/demo/output_fiddle_all.csv +0 -0
  50. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/demo/output_sirius.csv +0 -0
  51. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle/optim.py +0 -0
  52. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/dependency_links.txt +0 -0
  53. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/entry_points.txt +0 -0
  54. {msfiddle-2.2.1 → msfiddle-2.4.0}/msfiddle.egg-info/top_level.txt +0 -0
  55. {msfiddle-2.2.1 → msfiddle-2.4.0}/setup.cfg +0 -0
  56. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_mol_utils.py +0 -0
  57. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_muon.py +0 -0
  58. {msfiddle-2.2.1 → msfiddle-2.4.0}/tests/test_refine_utils.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: msfiddle
3
- Version: 2.2.1
3
+ Version: 2.4.0
4
4
  Summary: A package for predicting chemical formulas from tandem mass spectra
5
5
  Home-page: https://github.com/JosieHong/FIDDLE
6
6
  Author: Yuhui Hong
@@ -15,14 +15,14 @@ Classifier: Topic :: Scientific/Engineering :: Medical Science Apps.
15
15
  Requires-Python: >=3.8
16
16
  Description-Content-Type: text/markdown
17
17
  License-File: LICENSE
18
- Requires-Dist: numpy<2.0.0,>=1.20.0
18
+ Requires-Dist: numpy>=1.20.0
19
19
  Requires-Dist: pandas>=2.0.0
20
20
  Requires-Dist: tqdm>=4.60.0
21
21
  Requires-Dist: pyyaml>=6.0
22
22
  Requires-Dist: scikit-learn>=1.0.0
23
23
  Requires-Dist: scipy>=1.8.0
24
24
  Requires-Dist: pyarrow>=10.0.0
25
- Requires-Dist: rdkit>=2022.03.5
25
+ Requires-Dist: rdkit!=2026.9.2,>=2022.03.5
26
26
  Requires-Dist: molmass
27
27
  Requires-Dist: pyteomics
28
28
  Provides-Extra: inference
@@ -69,11 +69,13 @@ msfiddle predict --demo --result_path demo.csv # 21 bundled example spect
69
69
  msfiddle predict --test_data spectra.mgf --instrument_type orbitrap --result_path results.csv
70
70
  ```
71
71
 
72
- Use `--instrument_type qtof` for Q-TOF data. Each spectrum needs `TITLE`, `PRECURSOR_MZ`, `PRECURSOR_TYPE` and `COLLISION_ENERGY`. The results list ranked candidates per spectrum in `Refined Formula (0)`, `Refined Formula (1)`, … with their scores in `Rescore (k)`.
72
+ The `inference` extra installs PyTorch. For a particular PyTorch build (another CUDA version, CPU only), install it [from pytorch.org](https://pytorch.org/get-started/locally/) first, then `pip install msfiddle`.
73
73
 
74
- When the true formulas are known (a `SMILES` or `FORMULA` field in each spectrum), add `--eval` to also report top-k accuracy, saved as `results_eval.csv`.
74
+ Use `--instrument_type qtof` for Q-TOF data and `--instrument_type timstof` for timsTOF data. Each spectrum needs `TITLE`, `PRECURSOR_MZ`, `PRECURSOR_TYPE` and `COLLISION_ENERGY`. The results list ranked candidates per spectrum in `Refined Formula (0)`, `Refined Formula (1)`, … with their scores in `Rescore (k)`.
75
75
 
76
- The [documentation](https://msfiddle.readthedocs.io) covers the Python API, BUDDY and SIRIUS integration, custom checkpoints, and all output columns.
76
+ When the true formulas are known (a `SMILES` or `FORMULA` field in each spectrum), add `--eval` to also report top-k accuracy, saved as `results_eval.csv`. `msfiddle validate --test_data your.mgf` checks a file without the models and says why any spectrum would be skipped; if accuracy is lower than expected, see [troubleshooting](https://msfiddle.readthedocs.io/en/latest/troubleshooting.html).
77
+
78
+ The [documentation](https://msfiddle.readthedocs.io) covers the Python API, BUDDY and SIRIUS integration, custom checkpoints, all output columns, and [benchmark results](https://msfiddle.readthedocs.io/en/latest/benchmarks.html).
77
79
 
78
80
  ## Reproduce the paper
79
81
 
@@ -85,7 +87,12 @@ This repository also holds the research code for training and evaluation. Run ev
85
87
  conda env create -f environment.yml
86
88
  ```
87
89
 
88
- 2. (optional) Install [BUDDY](https://github.com/Philipbear/msbuddy) and [SIRIUS](https://v6.docs.sirius-ms.io/) for the comparison experiments.
90
+ 2. (optional) Install [BUDDY](https://github.com/Philipbear/msbuddy) and [SIRIUS](https://v6.docs.sirius-ms.io/) for the comparison experiments. SIRIUS needs a free account; log in once as `install_sirius.sh` prints at the end.
91
+
92
+ ```bash
93
+ bash running_scripts/install_buddy.sh
94
+ bash running_scripts/install_sirius.sh
95
+ ```
89
96
 
90
97
  3. Download the models into `./check_point/`, where the scripts expect them:
91
98
 
@@ -95,15 +102,16 @@ msfiddle download -d ./check_point
95
102
 
96
103
  | Script | Description |
97
104
  |---|---|
98
- | `running_scripts/train_release_v2.2.sh` | Build the data and train the release models (TCN and rescore, Q-TOF and Orbitrap) |
105
+ | `running_scripts/train_release_v2.4.sh` | As `train_release_v2.3.sh`, but as in the paper the theoretical precursor m/z of NIST20 and NIST23 is replaced by the theoretical m/z plus a seeded Gaussian error within the instrument's tolerance; trains one instrument after another on two GPUs |
106
+ | `running_scripts/train_release_v2.3.sh` | Build the data, train the release models (TCN and rescore) for Q-TOF, Orbitrap and timsTOF in parallel and evaluate them; `INSTRUMENTS=timstof` runs one of them. timsTOF trains on the Enveda CASMI 2026 library at `data/origin/casmi2026/train.parquet` |
99
107
  | `running_scripts/experiments_test_benchmark.sh` | Evaluate on external benchmarks (CASMI 2016, CASMI 2017, EMBL-MCF 2.0) |
100
108
  | `running_scripts/experiments_test_nist23.sh` | Train and evaluate with compounds new in NIST23 held out |
101
109
  | `running_scripts/experiments_test_chimeric.sh` | Evaluate on chimeric spectra |
102
110
  | `running_scripts/experiments_test_noised.sh` | Evaluate on spectra with added noise peaks |
103
111
 
104
- The benchmark, chimeric and noise experiments use the released models; the chimeric and noise ones also need the data built by `train_release_v2.2.sh`.
112
+ The benchmark, chimeric and noise experiments use the released models; the chimeric and noise ones also need the data those models were trained on.
105
113
 
106
- To train from scratch, see `scripts/train_tcn_gpus.py`, `scripts/train_tcn_gpus_cl.py` and `scripts/train_rescore.py` with the configs in [`msfiddle/config/`](https://github.com/JosieHong/FIDDLE/tree/main/msfiddle/config).
114
+ To train from scratch, see `scripts/train_tcn_gpus_cl.py` and `scripts/train_rescore.py` with the configs in [`msfiddle/config/`](https://github.com/JosieHong/FIDDLE/tree/main/msfiddle/config).
107
115
 
108
116
  ## Citation
109
117
 
@@ -28,11 +28,13 @@ msfiddle predict --demo --result_path demo.csv # 21 bundled example spect
28
28
  msfiddle predict --test_data spectra.mgf --instrument_type orbitrap --result_path results.csv
29
29
  ```
30
30
 
31
- Use `--instrument_type qtof` for Q-TOF data. Each spectrum needs `TITLE`, `PRECURSOR_MZ`, `PRECURSOR_TYPE` and `COLLISION_ENERGY`. The results list ranked candidates per spectrum in `Refined Formula (0)`, `Refined Formula (1)`, … with their scores in `Rescore (k)`.
31
+ The `inference` extra installs PyTorch. For a particular PyTorch build (another CUDA version, CPU only), install it [from pytorch.org](https://pytorch.org/get-started/locally/) first, then `pip install msfiddle`.
32
32
 
33
- When the true formulas are known (a `SMILES` or `FORMULA` field in each spectrum), add `--eval` to also report top-k accuracy, saved as `results_eval.csv`.
33
+ Use `--instrument_type qtof` for Q-TOF data and `--instrument_type timstof` for timsTOF data. Each spectrum needs `TITLE`, `PRECURSOR_MZ`, `PRECURSOR_TYPE` and `COLLISION_ENERGY`. The results list ranked candidates per spectrum in `Refined Formula (0)`, `Refined Formula (1)`, … with their scores in `Rescore (k)`.
34
34
 
35
- The [documentation](https://msfiddle.readthedocs.io) covers the Python API, BUDDY and SIRIUS integration, custom checkpoints, and all output columns.
35
+ When the true formulas are known (a `SMILES` or `FORMULA` field in each spectrum), add `--eval` to also report top-k accuracy, saved as `results_eval.csv`. `msfiddle validate --test_data your.mgf` checks a file without the models and says why any spectrum would be skipped; if accuracy is lower than expected, see [troubleshooting](https://msfiddle.readthedocs.io/en/latest/troubleshooting.html).
36
+
37
+ The [documentation](https://msfiddle.readthedocs.io) covers the Python API, BUDDY and SIRIUS integration, custom checkpoints, all output columns, and [benchmark results](https://msfiddle.readthedocs.io/en/latest/benchmarks.html).
36
38
 
37
39
  ## Reproduce the paper
38
40
 
@@ -44,7 +46,12 @@ This repository also holds the research code for training and evaluation. Run ev
44
46
  conda env create -f environment.yml
45
47
  ```
46
48
 
47
- 2. (optional) Install [BUDDY](https://github.com/Philipbear/msbuddy) and [SIRIUS](https://v6.docs.sirius-ms.io/) for the comparison experiments.
49
+ 2. (optional) Install [BUDDY](https://github.com/Philipbear/msbuddy) and [SIRIUS](https://v6.docs.sirius-ms.io/) for the comparison experiments. SIRIUS needs a free account; log in once as `install_sirius.sh` prints at the end.
50
+
51
+ ```bash
52
+ bash running_scripts/install_buddy.sh
53
+ bash running_scripts/install_sirius.sh
54
+ ```
48
55
 
49
56
  3. Download the models into `./check_point/`, where the scripts expect them:
50
57
 
@@ -54,15 +61,16 @@ msfiddle download -d ./check_point
54
61
 
55
62
  | Script | Description |
56
63
  |---|---|
57
- | `running_scripts/train_release_v2.2.sh` | Build the data and train the release models (TCN and rescore, Q-TOF and Orbitrap) |
64
+ | `running_scripts/train_release_v2.4.sh` | As `train_release_v2.3.sh`, but as in the paper the theoretical precursor m/z of NIST20 and NIST23 is replaced by the theoretical m/z plus a seeded Gaussian error within the instrument's tolerance; trains one instrument after another on two GPUs |
65
+ | `running_scripts/train_release_v2.3.sh` | Build the data, train the release models (TCN and rescore) for Q-TOF, Orbitrap and timsTOF in parallel and evaluate them; `INSTRUMENTS=timstof` runs one of them. timsTOF trains on the Enveda CASMI 2026 library at `data/origin/casmi2026/train.parquet` |
58
66
  | `running_scripts/experiments_test_benchmark.sh` | Evaluate on external benchmarks (CASMI 2016, CASMI 2017, EMBL-MCF 2.0) |
59
67
  | `running_scripts/experiments_test_nist23.sh` | Train and evaluate with compounds new in NIST23 held out |
60
68
  | `running_scripts/experiments_test_chimeric.sh` | Evaluate on chimeric spectra |
61
69
  | `running_scripts/experiments_test_noised.sh` | Evaluate on spectra with added noise peaks |
62
70
 
63
- The benchmark, chimeric and noise experiments use the released models; the chimeric and noise ones also need the data built by `train_release_v2.2.sh`.
71
+ The benchmark, chimeric and noise experiments use the released models; the chimeric and noise ones also need the data those models were trained on.
64
72
 
65
- To train from scratch, see `scripts/train_tcn_gpus.py`, `scripts/train_tcn_gpus_cl.py` and `scripts/train_rescore.py` with the configs in [`msfiddle/config/`](https://github.com/JosieHong/FIDDLE/tree/main/msfiddle/config).
73
+ To train from scratch, see `scripts/train_tcn_gpus_cl.py` and `scripts/train_rescore.py` with the configs in [`msfiddle/config/`](https://github.com/JosieHong/FIDDLE/tree/main/msfiddle/config).
66
74
 
67
75
  ## Citation
68
76
 
@@ -14,6 +14,7 @@ from .api import (
14
14
  predict_batch_from_spectra,
15
15
  predict_from_mgf,
16
16
  predict_from_spectrum,
17
+ validate_mgf,
17
18
  )
18
19
  from .utils.mol_utils import vector_to_formula, formula_to_vector
19
20
  from .utils.msms_utils import mass_calculator
@@ -39,6 +40,7 @@ __all__ = [
39
40
  "predict_from_spectrum",
40
41
  "predict_batch_from_spectra",
41
42
  "predict_from_mgf",
43
+ "validate_mgf",
42
44
  "MS2FNet_tcn",
43
45
  "FormulaEncoder",
44
46
  "RescoreHead",