MCPcopy Create free account
hub / github.com/THUNLP-MT/MEAN / parse

Function parse

data/split.py:13–33  ·  view source on GitHub ↗
()

Source from the content-addressed store, hash-verified

11
12
13def parse():
14 parser = ArgumentParser(description='split train / valid / test')
15 parser.add_argument('--data', type=str, required=True, help='Path to the data file')
16 parser.add_argument('--out_dir', type=str, default=None, help='Directory to save results. Default the same as input data.')
17 parser.add_argument('--valid_ratio', type=float, default=0.1,
18 help='Ratio of validation set')
19 parser.add_argument('--test_ratio', type=float, default=0.1,
20 help='Ratio of test set')
21 parser.add_argument('--cdr', type=str, choices=[f'cdrh{i}' for i in range(1, 4)] + [f'cdrl{i}' for i in range(1, 4)],
22 default='cdrh3', help='Cluster according to which cdr')
23 parser.add_argument('--filter', type=str, default='1*1', help='Filter out complex with heavy / light / antigen.' + \
24 'The code refers to heavy / light / antigen sequentially. 1 for has, 0 for not has, * for either.' + \
25 'e.g default 1*1 means has heavy chain and antigen, either has light chain or not.')
26 parser.add_argument('--k_fold', type=int, default=-1, help='K fold dataset. -1 for not do k-fold.' + \
27 'Note that if this is enabled, the test/valid ratio will be automatically calculated.')
28 parser.add_argument('--seed', type=int, default=2022, help='seed')
29 parser.add_argument('--rabd', type=str, default=None, help='Path to rabd json file. If this is enabled, '+ \
30 'RAbD complexes will be used as test set and complexes from data will be used as train/valid.' + \
31 'Note that complexes sharing clusters with RAbD will be dropped.' + \
32 'K fold will also be turned off.')
33 return parser.parse_args()
34
35
36def load_file(fpath):

Callers 1

split.pyFile · 0.70

Calls

no outgoing calls

Tested by

no test coverage detected