annotate simple_model_fit.py @ 8:03e0da430d44 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
author bgruening
date Wed, 09 Aug 2023 13:37:05 +0000
parents c16818ce0424
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
1 import argparse
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
2 import json
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
3
2
e36ab18cbaca "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 1
diff changeset
4 import pandas as pd
7
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
5 from galaxy_ml.model_persist import dump_model_to_h5, load_model_from_h5
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
6 from galaxy_ml.utils import read_columns
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
7 from scipy.io import mmread
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
8 from sklearn.pipeline import Pipeline
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
9
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
10 N_JOBS = int(__import__("os").environ.get("GALAXY_SLOTS", 1))
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
11
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
12
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
13 # TODO import from galaxy_ml.utils in future versions
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
14 def clean_params(estimator, n_jobs=None):
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
15 """clean unwanted hyperparameter settings
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
16
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
17 If n_jobs is not None, set it into the estimator, if applicable
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
18
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
19 Return
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
20 ------
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
21 Cleaned estimator object
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
22 """
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
23 ALLOWED_CALLBACKS = (
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
24 "EarlyStopping",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
25 "TerminateOnNaN",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
26 "ReduceLROnPlateau",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
27 "CSVLogger",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
28 "None",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
29 )
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
30
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
31 estimator_params = estimator.get_params()
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
32
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
33 for name, p in estimator_params.items():
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
34 # all potential unauthorized file write
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
35 if name == "memory" or name.endswith("__memory") or name.endswith("_path"):
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
36 new_p = {name: None}
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
37 estimator.set_params(**new_p)
3
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
38 elif n_jobs is not None and (name == "n_jobs" or name.endswith("__n_jobs")):
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
39 new_p = {name: n_jobs}
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
40 estimator.set_params(**new_p)
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
41 elif name.endswith("callbacks"):
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
42 for cb in p:
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
43 cb_type = cb["callback_selection"]["callback_type"]
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
44 if cb_type not in ALLOWED_CALLBACKS:
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
45 raise ValueError("Prohibited callback type: %s!" % cb_type)
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
46
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
47 return estimator
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
48
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
49
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
50 def _get_X_y(params, infile1, infile2):
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
51 """read from inputs and output X and y
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
52
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
53 Parameters
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
54 ----------
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
55 params : dict
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
56 Tool inputs parameter
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
57 infile1 : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
58 File path to dataset containing features
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
59 infile2 : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
60 File path to dataset containing target values
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
61
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
62 """
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
63 # store read dataframe object
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
64 loaded_df = {}
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
65
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
66 input_type = params["input_options"]["selected_input"]
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
67 # tabular input
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
68 if input_type == "tabular":
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
69 header = "infer" if params["input_options"]["header1"] else None
3
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
70 column_option = params["input_options"]["column_selector_options_1"][
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
71 "selected_column_selector_option"
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
72 ]
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
73 if column_option in [
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
74 "by_index_number",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
75 "all_but_by_index_number",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
76 "by_header_name",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
77 "all_but_by_header_name",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
78 ]:
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
79 c = params["input_options"]["column_selector_options_1"]["col1"]
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
80 else:
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
81 c = None
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
82
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
83 df_key = infile1 + repr(header)
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
84 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True)
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
85 loaded_df[df_key] = df
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
86
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
87 X = read_columns(df, c=c, c_option=column_option).astype(float)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
88 # sparse input
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
89 elif input_type == "sparse":
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
90 X = mmread(open(infile1, "r"))
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
91
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
92 # Get target y
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
93 header = "infer" if params["input_options"]["header2"] else None
3
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
94 column_option = params["input_options"]["column_selector_options_2"][
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
95 "selected_column_selector_option2"
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
96 ]
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
97 if column_option in [
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
98 "by_index_number",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
99 "all_but_by_index_number",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
100 "by_header_name",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
101 "all_but_by_header_name",
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
102 ]:
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
103 c = params["input_options"]["column_selector_options_2"]["col2"]
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
104 else:
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
105 c = None
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
106
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
107 df_key = infile2 + repr(header)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
108 if df_key in loaded_df:
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
109 infile2 = loaded_df[df_key]
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
110 else:
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
111 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True)
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
112 loaded_df[df_key] = infile2
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
113
3
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
114 y = read_columns(
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
115 infile2, c=c, c_option=column_option, sep="\t", header=header, parse_dates=True
24c1cc2dd4a4 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
bgruening
parents: 2
diff changeset
116 )
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
117 if len(y.shape) == 2 and y.shape[1] == 1:
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
118 y = y.ravel()
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
119
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
120 return X, y
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
121
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
122
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
123 def main(inputs, infile_estimator, infile1, infile2, out_object, out_weights=None):
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
124 """main
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
125
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
126 Parameters
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
127 ----------
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
128 inputs : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
129 File path to galaxy tool parameter
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
130
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
131 infile_estimator : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
132 File paths of input estimator
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
133
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
134 infile1 : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
135 File path to dataset containing features
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
136
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
137 infile2 : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
138 File path to dataset containing target labels
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
139
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
140 out_object : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
141 File path for output of fitted model or skeleton
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
142
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
143 out_weights : str
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
144 File path for output of weights
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
145
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
146 """
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
147 with open(inputs, "r") as param_handler:
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
148 params = json.load(param_handler)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
149
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
150 # load model
7
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
151 estimator = load_model_from_h5(infile_estimator)
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
152
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
153 estimator = clean_params(estimator)
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
154
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
155 X_train, y_train = _get_X_y(params, infile1, infile2)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
156
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
157 estimator.fit(X_train, y_train)
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
158
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
159 main_est = estimator
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
160 if isinstance(main_est, Pipeline):
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
161 main_est = main_est.steps[-1][-1]
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
162 if hasattr(main_est, "model_") and hasattr(main_est, "save_weights"):
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
163 if out_weights:
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
164 main_est.save_weights(out_weights)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
165 del main_est.model_
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
166 del main_est.fit_params
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
167 del main_est.model_class_
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
168 if getattr(main_est, "validation_data", None):
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
169 del main_est.validation_data
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
170 if getattr(main_est, "data_generator_", None):
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
171 del main_est.data_generator_
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
172
7
c16818ce0424 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents: 3
diff changeset
173 dump_model_to_h5(estimator, out_object)
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
174
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
175
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
176 if __name__ == "__main__":
0
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
177 aparser = argparse.ArgumentParser()
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
178 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
179 aparser.add_argument("-X", "--infile_estimator", dest="infile_estimator")
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
180 aparser.add_argument("-y", "--infile1", dest="infile1")
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
181 aparser.add_argument("-g", "--infile2", dest="infile2")
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
182 aparser.add_argument("-o", "--out_object", dest="out_object")
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
183 aparser.add_argument("-t", "--out_weights", dest="out_weights")
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
184 args = aparser.parse_args()
2d7016b3ae92 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 2afb24f3c81d625312186750a714d702363012b5"
bgruening
parents:
diff changeset
185
1
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
186 main(
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
187 args.inputs,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
188 args.infile_estimator,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
189 args.infile1,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
190 args.infile2,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
191 args.out_object,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
192 args.out_weights,
132805688fa3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 0
diff changeset
193 )