papyrus-scripts 2.0.0__tar.gz → 2.1.0__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.
- {papyrus_scripts-2.0.0/src/papyrus_scripts.egg-info → papyrus_scripts-2.1.0}/PKG-INFO +1 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/__init__.py +1 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/oop.py +4 -4
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/reader.py +3 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/IO.py +1 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/aliases.json +1 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/links.json +47 -47
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0/src/papyrus_scripts.egg-info}/PKG-INFO +1 -1
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/LICENSE +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/README.md +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/setup.cfg +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/setup.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/__main__.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/cli.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/download.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/fingerprint.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/matchRCSB.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/modelling.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/neuralnet.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/preprocess.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/subsim_search.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/UniprotMatch.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/__init__.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts/utils/mol_reader.py +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/SOURCES.txt +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/dependency_links.txt +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/entry_points.txt +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/requires.txt +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/top_level.txt +0 -0
- {papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/tests/test_oop.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.1
|
|
2
2
|
Name: papyrus_scripts
|
|
3
|
-
Version: 2.
|
|
3
|
+
Version: 2.1.0
|
|
4
4
|
Summary: A collection of scripts to handle the Papyrus bioactivity dataset
|
|
5
5
|
Home-page: https://github.com/OlivierBeq/Papyrus-scripts
|
|
6
6
|
Author: Olivier J. M. Béquignon - Brandon J. Bongers - Willem Jespers
|
|
@@ -27,16 +27,16 @@ class PapyrusDataset:
|
|
|
27
27
|
chunksize: Optional[int] = 1_000_000, source_path: Optional[str] = None,
|
|
28
28
|
download_progress: bool = False):
|
|
29
29
|
version = IO.PapyrusVersion(version=version)
|
|
30
|
+
if not IO.is_local_version_available(version=version.version_old_fmt, root_folder=source_path):
|
|
31
|
+
download.download_papyrus(outdir=source_path, version=version.version_old_fmt, nostereo=True, stereo=True,
|
|
32
|
+
only_pp=False, structures=True, descriptors='all', progress=download_progress,
|
|
33
|
+
disk_margin=0.0)
|
|
30
34
|
self.papyrus_params = dict(is3d=is3d, version=version, plusplus=plusplus,
|
|
31
35
|
chunksize=chunksize, source_path=source_path,
|
|
32
36
|
num_rows=IO.get_num_rows_in_file(filetype='bioactivities', is3D=is3d,
|
|
33
37
|
version=version, plusplus=plusplus,
|
|
34
38
|
root_folder=source_path),
|
|
35
39
|
download_progress=download_progress)
|
|
36
|
-
if not IO.is_local_version_available(version=version, root_folder=source_path):
|
|
37
|
-
download.download_papyrus(outdir=source_path, version=version, nostereo=not is3d, stereo=is3d,
|
|
38
|
-
only_pp=plusplus, structures=True, descriptors=None, progress=download_progress,
|
|
39
|
-
disk_margin=0.0)
|
|
40
40
|
self.papyrus_bioactivity_data = reader.read_papyrus(is3d=is3d, version=version, plusplus=plusplus,
|
|
41
41
|
chunksize=chunksize, source_path=source_path)
|
|
42
42
|
self.papyrus_protein_data = reader.read_protein_set(source_path=source_path, version=version)
|
|
@@ -17,7 +17,9 @@ from .utils.mol_reader import MolSupplier
|
|
|
17
17
|
from .utils.IO import locate_file, process_data_version, TypeDecoder, PapyrusVersion
|
|
18
18
|
|
|
19
19
|
|
|
20
|
-
def read_papyrus(is3d: bool = False, version: str = 'latest', plusplus: bool = True,
|
|
20
|
+
def read_papyrus(is3d: bool = False, version: str | PapyrusVersion = 'latest', plusplus: bool = True,
|
|
21
|
+
chunksize: Optional[int] = None, source_path: Optional[str] = None
|
|
22
|
+
) -> Union[Iterator[pd.DataFrame], pd.DataFrame]:
|
|
21
23
|
|
|
22
24
|
"""Read the Papyrus dataset.
|
|
23
25
|
|
|
@@ -205,7 +205,7 @@ def is_local_version_available(version: str, root_folder: str = None):
|
|
|
205
205
|
try:
|
|
206
206
|
_ = process_data_version(version=version, root_folder=root_folder)
|
|
207
207
|
return True
|
|
208
|
-
except IOError:
|
|
208
|
+
except (IOError, ValueError):
|
|
209
209
|
return False
|
|
210
210
|
|
|
211
211
|
def locate_file(dirpath: str, regex_pattern: str):
|
|
@@ -310,99 +310,99 @@
|
|
|
310
310
|
"05.7": {
|
|
311
311
|
"readme": {
|
|
312
312
|
"name": "README.txt",
|
|
313
|
-
"url": "https://zenodo.org/records/
|
|
313
|
+
"url": "https://zenodo.org/records/13987985/files/README.txt?download=1",
|
|
314
314
|
"size": 12813,
|
|
315
|
-
"sha256": "
|
|
315
|
+
"sha256": "fab159a88e302cad3e5b84ccea72a7c9fb2c212f656324d6191865460511f50d"
|
|
316
316
|
},
|
|
317
317
|
"requirements": {
|
|
318
318
|
"name": "05.7_additional_files.zip",
|
|
319
|
-
"url": "https://zenodo.org/records/
|
|
320
|
-
"size":
|
|
321
|
-
"sha256": "
|
|
319
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_additional_files.zip?download=1",
|
|
320
|
+
"size": 113945,
|
|
321
|
+
"sha256": "0621cd63111286777769e6ea1c59e7adc1d05833bb1f61e50ba9e5be189d60da"
|
|
322
322
|
},
|
|
323
323
|
"papyrus++": {
|
|
324
324
|
"name": "05.7++_combined_set_without_stereochemistry.tsv.xz",
|
|
325
|
-
"url": "https://zenodo.org/records/
|
|
326
|
-
"size":
|
|
327
|
-
"sha256": "
|
|
325
|
+
"url": "https://zenodo.org/records/13987985/files/05.7++_combined_set_without_stereochemistry.tsv.xz?download=1",
|
|
326
|
+
"size": 56759540 ,
|
|
327
|
+
"sha256": "8004e0d1027a760f205b45264386f792e7d49658da39f77f52e660a6f19760dd"
|
|
328
328
|
},
|
|
329
329
|
"2D_papyrus": {
|
|
330
330
|
"name": "05.7_combined_set_without_stereochemistry.tsv.xz",
|
|
331
|
-
"url": "https://zenodo.org/records/
|
|
332
|
-
"size":
|
|
333
|
-
"sha256": "
|
|
331
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_set_without_stereochemistry.tsv.xz?download=1",
|
|
332
|
+
"size": 751521856,
|
|
333
|
+
"sha256": "9a7657f2636473ea1f4b033c3abbc9709608517f262f97e8adcc8f59d4f1189b"
|
|
334
334
|
},
|
|
335
335
|
"2D_structures": {
|
|
336
336
|
"name": "05.7_combined_2D_set_without_stereochemistry.sd.xz",
|
|
337
|
-
"url": "https://zenodo.org/records/
|
|
338
|
-
"size":
|
|
339
|
-
"sha256": "
|
|
337
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_2D_set_without_stereochemistry.sd.xz?download=1",
|
|
338
|
+
"size": 453721500 ,
|
|
339
|
+
"sha256": "56be7c058130e9e861d884dc6094cf0ac4c3f37a75c7d2c4302685c4720f69ae"
|
|
340
340
|
},
|
|
341
341
|
"3D_papyrus": {
|
|
342
342
|
"name": "05.7_combined_set_with_stereochemistry.tsv.xz",
|
|
343
|
-
"url": "https://zenodo.org/records/
|
|
344
|
-
"size":
|
|
345
|
-
"sha256": "
|
|
343
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_set_with_stereochemistry.tsv.xz?download=1",
|
|
344
|
+
"size": 715716852,
|
|
345
|
+
"sha256": "88a965ef8827692b1489bc947249e9fc00287ab6b63cbd2767862080a98e9a4c"
|
|
346
346
|
},
|
|
347
347
|
"3D_structures": {
|
|
348
348
|
"name": "05.7_combined_3D_set_with_stereochemistry.sd.xz",
|
|
349
|
-
"url": "https://zenodo.org/records/
|
|
350
|
-
"size":
|
|
351
|
-
"sha256": "
|
|
349
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_3D_set_with_stereochemistry.sd.xz?download=1",
|
|
350
|
+
"size": 507933052,
|
|
351
|
+
"sha256": "8f1490a701e918e013770ea589651825ca2a459b214f50d6ff9ce892af398def"
|
|
352
352
|
},
|
|
353
353
|
"2D_fingerprint": {
|
|
354
354
|
"name": "05.7_combined_2D_moldescs_ECFP6.tsv.xz",
|
|
355
|
-
"url": "https://zenodo.org/records/
|
|
356
|
-
"size":
|
|
357
|
-
"sha256": "
|
|
355
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_2D_moldescs_ECFP6.tsv.xz?download=1",
|
|
356
|
+
"size": 99889764,
|
|
357
|
+
"sha256": "6689cd5d3841abc350cb2dba719a2af02e119af2a595f15790ad14e5c4ace108"
|
|
358
358
|
},
|
|
359
359
|
"3D_fingerprint": {
|
|
360
360
|
"name": "05.7_combined_3D_moldescs_E3FP.tsv.xz",
|
|
361
|
-
"url": "https://zenodo.org/records/
|
|
362
|
-
"size":
|
|
363
|
-
"sha256": "
|
|
361
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_3D_moldescs_E3FP.tsv.xz?download=1",
|
|
362
|
+
"size": 119952372,
|
|
363
|
+
"sha256": "6c247e785e5885e08ecebc2b452e05dcbb24395adabdef71b903d6491e9ae096"
|
|
364
364
|
},
|
|
365
365
|
"2D_mordred": {
|
|
366
366
|
"name": "05.7_combined_2D_moldescs_mordred2D.tsv.xz",
|
|
367
|
-
"url": "https://zenodo.org/records/
|
|
368
|
-
"size":
|
|
369
|
-
"sha256": "
|
|
367
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_2D_moldescs_mordred2D.tsv.xz?download=1",
|
|
368
|
+
"size": 3149391660,
|
|
369
|
+
"sha256": "26781e0879af798a6b7df4e6c515efd79599335a04706d5335fdc8e5c5565fc3"
|
|
370
370
|
},
|
|
371
371
|
"3D_mordred": {
|
|
372
372
|
"name": "05.7_combined_3D_moldescs_mordred3D.tsv.xz",
|
|
373
|
-
"url": "https://zenodo.org/records/
|
|
374
|
-
"size":
|
|
375
|
-
"sha256": "
|
|
373
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_3D_moldescs_mordred3D.tsv.xz?download=1",
|
|
374
|
+
"size": 3421107392,
|
|
375
|
+
"sha256": "c03f3d4c702815d5bfa2ddf80e20717d0bd13a542b0ca30e62534126eef03b0d"
|
|
376
376
|
},
|
|
377
377
|
"2D_cddd": {
|
|
378
378
|
"name": "05.7_combined_2D_moldescs_CDDDs.tsv.xz",
|
|
379
|
-
"url": "https://zenodo.org/records/
|
|
380
|
-
"size":
|
|
381
|
-
"sha256": "
|
|
379
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_2D_moldescs_CDDDs.tsv.xz?download=1",
|
|
380
|
+
"size": 2167302956,
|
|
381
|
+
"sha256": "d86539cc76a537878725c4ef8a6703d316add737cb51915ad203e346fe92f6c9"
|
|
382
382
|
},
|
|
383
383
|
"2D_mold2": {
|
|
384
384
|
"name": "05.7_combined_2D_moldescs_mold2.tsv.xz",
|
|
385
|
-
"url": "https://zenodo.org/records/
|
|
386
|
-
"size":
|
|
387
|
-
"sha256": "
|
|
385
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_2D_moldescs_mold2.tsv.xz?download=1",
|
|
386
|
+
"size": 1539905628,
|
|
387
|
+
"sha256": "493436f96d30680568c2a70ed08d76a159b916c57e9df03f639ae7b414fb87cb"
|
|
388
388
|
},
|
|
389
389
|
"proteins": {
|
|
390
390
|
"name": "05.7_combined_set_protein_targets.tsv.xz",
|
|
391
|
-
"url": "https://zenodo.org/records/
|
|
392
|
-
"size":
|
|
393
|
-
"sha256": "
|
|
391
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_set_protein_targets.tsv.xz?download=1",
|
|
392
|
+
"size": 1780032,
|
|
393
|
+
"sha256": "832e564fb82daea0e4da79abcb44834d10104229382874e79915a1288d80783c"
|
|
394
394
|
},
|
|
395
395
|
"proteins_unirep": {
|
|
396
396
|
"name": "05.7_combined_protdescs_unirep.tsv.xz",
|
|
397
|
-
"url": "https://zenodo.org/records/
|
|
398
|
-
"size":
|
|
399
|
-
"sha256": "
|
|
397
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_protdescs_unirep.tsv.xz?download=1",
|
|
398
|
+
"size": 209957000,
|
|
399
|
+
"sha256": "a39f21db7ed5ba72ef881a9f05f5362f7aaaa0f2709c023a0060417678f30dec"
|
|
400
400
|
},
|
|
401
401
|
"proteins_prodec": {
|
|
402
402
|
"name": "05.7_combined_protdescs_ProDEC.tsv.xz",
|
|
403
|
-
"url": "https://zenodo.org/records/
|
|
404
|
-
"size":
|
|
405
|
-
"sha256": "
|
|
403
|
+
"url": "https://zenodo.org/records/13987985/files/05.7_combined_protdescs_ProDEC.tsv.xz?download=1",
|
|
404
|
+
"size": 435015384,
|
|
405
|
+
"sha256": "1ec2d7b0cd95c93aaabacf4153e58e464e4327f0ebb3bad0077fd740b7334cb1"
|
|
406
406
|
}
|
|
407
407
|
}
|
|
408
408
|
}
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.1
|
|
2
2
|
Name: papyrus_scripts
|
|
3
|
-
Version: 2.
|
|
3
|
+
Version: 2.1.0
|
|
4
4
|
Summary: A collection of scripts to handle the Papyrus bioactivity dataset
|
|
5
5
|
Home-page: https://github.com/OlivierBeq/Papyrus-scripts
|
|
6
6
|
Author: Olivier J. M. Béquignon - Brandon J. Bongers - Willem Jespers
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
{papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/dependency_links.txt
RENAMED
|
File without changes
|
{papyrus_scripts-2.0.0 → papyrus_scripts-2.1.0}/src/papyrus_scripts.egg-info/entry_points.txt
RENAMED
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|