Mercurial > repos > bgruening > keras_model_builder

annotate keras_train_and_eval.py @ 9:e3b420d0b71a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
author bgruening
date Tue, 13 Apr 2021 22:42:14 +0000
parents 449a757be9c9
children 9b6faa256f15
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
1 import argparse
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
2 import json
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
3 import os
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
4 import pickle
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
5 import warnings
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
6 from itertools import chain
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
7
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
8 import joblib
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
9 import numpy as np
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
10 import pandas as pd
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
11 from galaxy_ml.externals.selene_sdk.utils import compute_score
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
12 from galaxy_ml.keras_galaxy_models import _predict_generator
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
13 from galaxy_ml.model_validations import train_test_split
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
14 from galaxy_ml.utils import (
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
15 clean_params,
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
16 get_main_estimator,
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
17 get_module,
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
18 get_scoring,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
19 load_model,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
20 read_columns,
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
21 SafeEval,
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
22 try_get_attr,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
23 )
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
24 from scipy.io import mmread
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
25 from sklearn.metrics.scorer import _check_multimetric_scoring
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
26 from sklearn.model_selection import _search, _validation
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
27 from sklearn.model_selection._validation import _score
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
28 from sklearn.pipeline import Pipeline
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
29 from sklearn.utils import indexable, safe_indexing
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
30
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
31
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
32 _fit_and_score = try_get_attr("galaxy_ml.model_validations", "_fit_and_score")
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
33 setattr(_search, "_fit_and_score", _fit_and_score)
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
34 setattr(_validation, "_fit_and_score", _fit_and_score)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
35
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
36 N_JOBS = int(os.environ.get("GALAXY_SLOTS", 1))
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
37 CACHE_DIR = os.path.join(os.getcwd(), "cached")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
38 del os
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
39 NON_SEARCHABLE = ("n_jobs", "pre_dispatch", "memory", "_path", "nthread", "callbacks")
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
40 ALLOWED_CALLBACKS = (
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
41 "EarlyStopping",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
42 "TerminateOnNaN",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
43 "ReduceLROnPlateau",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
44 "CSVLogger",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
45 "None",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
46 )
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
47
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
48
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
49 def _eval_swap_params(params_builder):
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
50 swap_params = {}
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
51
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
52 for p in params_builder["param_set"]:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
53 swap_value = p["sp_value"].strip()
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
54 if swap_value == "":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
55 continue
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
56
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
57 param_name = p["sp_name"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
58 if param_name.lower().endswith(NON_SEARCHABLE):
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
59 warnings.warn("Warning: `%s` is not eligible for search and was " "omitted!" % param_name)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
60 continue
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
61
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
62 if not swap_value.startswith(":"):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
63 safe_eval = SafeEval(load_scipy=True, load_numpy=True)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
64 ev = safe_eval(swap_value)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
65 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
66 # Have `:` before search list, asks for estimator evaluatio
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
67 safe_eval_es = SafeEval(load_estimators=True)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
68 swap_value = swap_value[1:].strip()
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
69 # TODO maybe add regular express check
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
70 ev = safe_eval_es(swap_value)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
71
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
72 swap_params[param_name] = ev
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
73
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
74 return swap_params
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
75
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
76
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
77 def train_test_split_none(*arrays, **kwargs):
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
78 """extend train_test_split to take None arrays
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
79 and support split by group names.
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
80 """
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
81 nones = []
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
82 new_arrays = []
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
83 for idx, arr in enumerate(arrays):
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
84 if arr is None:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
85 nones.append(idx)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
86 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
87 new_arrays.append(arr)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
88
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
89 if kwargs["shuffle"] == "None":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
90 kwargs["shuffle"] = None
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
91
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
92 group_names = kwargs.pop("group_names", None)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
93
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
94 if group_names is not None and group_names.strip():
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
95 group_names = [name.strip() for name in group_names.split(",")]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
96 new_arrays = indexable(*new_arrays)
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
97 groups = kwargs["labels"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
98 n_samples = new_arrays[0].shape[0]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
99 index_arr = np.arange(n_samples)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
100 test = index_arr[np.isin(groups, group_names)]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
101 train = index_arr[~np.isin(groups, group_names)]
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
102 rval = list(chain.from_iterable((safe_indexing(a, train), safe_indexing(a, test)) for a in new_arrays))
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
103 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
104 rval = train_test_split(*new_arrays, **kwargs)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
105
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
106 for pos in nones:
9
e3b420d0b71a "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 208a8d348e7c7a182cfbe1b6f17868146428a7e2"
bgruening
parents: 8
diff changeset
107 rval[pos * 2: 2] = [None, None]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
108
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
109 return rval
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
110
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
111
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
112 def _evaluate(y_true, pred_probas, scorer, is_multimetric=True):
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
113 """output scores based on input scorer
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
114
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
115 Parameters
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
116 ----------
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
117 y_true : array
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
118 True label or target values
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
119 pred_probas : array
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
120 Prediction values, probability for classification problem
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
121 scorer : dict
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
122 dict of `sklearn.metrics.scorer.SCORER`
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
123 is_multimetric : bool, default is True
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
124 """
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
125 if y_true.ndim == 1 or y_true.shape[-1] == 1:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
126 pred_probas = pred_probas.ravel()
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
127 pred_labels = (pred_probas > 0.5).astype("int32")
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
128 targets = y_true.ravel().astype("int32")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
129 if not is_multimetric:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
130 preds = pred_labels if scorer.__class__.__name__ == "_PredictScorer" else pred_probas
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
131 score = scorer._score_func(targets, preds, **scorer._kwargs)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
132
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
133 return score
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
134 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
135 scores = {}
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
136 for name, one_scorer in scorer.items():
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
137 preds = pred_labels if one_scorer.__class__.__name__ == "_PredictScorer" else pred_probas
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
138 score = one_scorer._score_func(targets, preds, **one_scorer._kwargs)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
139 scores[name] = score
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
140
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
141 # TODO: multi-class metrics
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
142 # multi-label
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
143 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
144 pred_labels = (pred_probas > 0.5).astype("int32")
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
145 targets = y_true.astype("int32")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
146 if not is_multimetric:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
147 preds = pred_labels if scorer.__class__.__name__ == "_PredictScorer" else pred_probas
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
148 score, _ = compute_score(preds, targets, scorer._score_func)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
149 return score
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
150 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
151 scores = {}
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
152 for name, one_scorer in scorer.items():
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
153 preds = pred_labels if one_scorer.__class__.__name__ == "_PredictScorer" else pred_probas
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
154 score, _ = compute_score(preds, targets, one_scorer._score_func)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
155 scores[name] = score
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
156
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
157 return scores
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
158
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
159
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
160 def main(
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
161 inputs,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
162 infile_estimator,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
163 infile1,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
164 infile2,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
165 outfile_result,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
166 outfile_object=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
167 outfile_weights=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
168 outfile_y_true=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
169 outfile_y_preds=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
170 groups=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
171 ref_seq=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
172 intervals=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
173 targets=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
174 fasta_path=None,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
175 ):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
176 """
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
177 Parameter
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
178 ---------
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
179 inputs : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
180 File path to galaxy tool parameter
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
181
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
182 infile_estimator : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
183 File path to estimator
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
184
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
185 infile1 : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
186 File path to dataset containing features
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
187
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
188 infile2 : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
189 File path to dataset containing target values
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
190
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
191 outfile_result : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
192 File path to save the results, either cv_results or test result
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
193
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
194 outfile_object : str, optional
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
195 File path to save searchCV object
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
196
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
197 outfile_weights : str, optional
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
198 File path to save deep learning model weights
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
199
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
200 outfile_y_true : str, optional
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
201 File path to target values for prediction
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
202
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
203 outfile_y_preds : str, optional
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
204 File path to save deep learning model weights
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
205
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
206 groups : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
207 File path to dataset containing groups labels
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
208
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
209 ref_seq : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
210 File path to dataset containing genome sequence file
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
211
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
212 intervals : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
213 File path to dataset containing interval file
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
214
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
215 targets : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
216 File path to dataset compressed target bed file
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
217
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
218 fasta_path : str
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
219 File path to dataset containing fasta file
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
220 """
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
221 warnings.simplefilter("ignore")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
222
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
223 with open(inputs, "r") as param_handler:
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
224 params = json.load(param_handler)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
225
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
226 # load estimator
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
227 with open(infile_estimator, "rb") as estimator_handler:
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
228 estimator = load_model(estimator_handler)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
229
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
230 estimator = clean_params(estimator)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
231
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
232 # swap hyperparameter
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
233 swapping = params["experiment_schemes"]["hyperparams_swapping"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
234 swap_params = _eval_swap_params(swapping)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
235 estimator.set_params(**swap_params)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
236
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
237 estimator_params = estimator.get_params()
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
238
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
239 # store read dataframe object
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
240 loaded_df = {}
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
241
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
242 input_type = params["input_options"]["selected_input"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
243 # tabular input
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
244 if input_type == "tabular":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
245 header = "infer" if params["input_options"]["header1"] else None
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
246 column_option = params["input_options"]["column_selector_options_1"]["selected_column_selector_option"]
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
247 if column_option in [
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
248 "by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
249 "all_but_by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
250 "by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
251 "all_but_by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
252 ]:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
253 c = params["input_options"]["column_selector_options_1"]["col1"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
254 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
255 c = None
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
256
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
257 df_key = infile1 + repr(header)
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
258 df = pd.read_csv(infile1, sep="\t", header=header, parse_dates=True)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
259 loaded_df[df_key] = df
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
260
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
261 X = read_columns(df, c=c, c_option=column_option).astype(float)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
262 # sparse input
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
263 elif input_type == "sparse":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
264 X = mmread(open(infile1, "r"))
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
265
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
266 # fasta_file input
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
267 elif input_type == "seq_fasta":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
268 pyfaidx = get_module("pyfaidx")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
269 sequences = pyfaidx.Fasta(fasta_path)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
270 n_seqs = len(sequences.keys())
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
271 X = np.arange(n_seqs)[:, np.newaxis]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
272 for param in estimator_params.keys():
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
273 if param.endswith("fasta_path"):
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
274 estimator.set_params(**{param: fasta_path})
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
275 break
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
276 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
277 raise ValueError(
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
278 "The selected estimator doesn't support "
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
279 "fasta file input! Please consider using "
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
280 "KerasGBatchClassifier with "
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
281 "FastaDNABatchGenerator/FastaProteinBatchGenerator "
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
282 "or having GenomeOneHotEncoder/ProteinOneHotEncoder "
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
283 "in pipeline!"
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
284 )
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
285
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
286 elif input_type == "refseq_and_interval":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
287 path_params = {
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
288 "data_batch_generator__ref_genome_path": ref_seq,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
289 "data_batch_generator__intervals_path": intervals,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
290 "data_batch_generator__target_path": targets,
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
291 }
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
292 estimator.set_params(**path_params)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
293 n_intervals = sum(1 for line in open(intervals))
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
294 X = np.arange(n_intervals)[:, np.newaxis]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
295
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
296 # Get target y
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
297 header = "infer" if params["input_options"]["header2"] else None
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
298 column_option = params["input_options"]["column_selector_options_2"]["selected_column_selector_option2"]
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
299 if column_option in [
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
300 "by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
301 "all_but_by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
302 "by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
303 "all_but_by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
304 ]:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
305 c = params["input_options"]["column_selector_options_2"]["col2"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
306 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
307 c = None
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
308
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
309 df_key = infile2 + repr(header)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
310 if df_key in loaded_df:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
311 infile2 = loaded_df[df_key]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
312 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
313 infile2 = pd.read_csv(infile2, sep="\t", header=header, parse_dates=True)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
314 loaded_df[df_key] = infile2
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
315
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
316 y = read_columns(infile2,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
317 c=c,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
318 c_option=column_option,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
319 sep='\t',
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
320 header=header,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
321 parse_dates=True)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
322 if len(y.shape) == 2 and y.shape[1] == 1:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
323 y = y.ravel()
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
324 if input_type == "refseq_and_interval":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
325 estimator.set_params(data_batch_generator__features=y.ravel().tolist())
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
326 y = None
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
327 # end y
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
328
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
329 # load groups
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
330 if groups:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
331 groups_selector = (params["experiment_schemes"]["test_split"]["split_algos"]).pop("groups_selector")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
332
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
333 header = "infer" if groups_selector["header_g"] else None
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
334 column_option = groups_selector["column_selector_options_g"]["selected_column_selector_option_g"]
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
335 if column_option in [
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
336 "by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
337 "all_but_by_index_number",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
338 "by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
339 "all_but_by_header_name",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
340 ]:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
341 c = groups_selector["column_selector_options_g"]["col_g"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
342 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
343 c = None
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
344
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
345 df_key = groups + repr(header)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
346 if df_key in loaded_df:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
347 groups = loaded_df[df_key]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
348
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
349 groups = read_columns(groups,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
350 c=c,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
351 c_option=column_option,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
352 sep='\t',
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
353 header=header,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
354 parse_dates=True)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
355 groups = groups.ravel()
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
356
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
357 # del loaded_df
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
358 del loaded_df
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
359
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
360 # cache iraps_core fits could increase search speed significantly
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
361 memory = joblib.Memory(location=CACHE_DIR, verbose=0)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
362 main_est = get_main_estimator(estimator)
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
363 if main_est.__class__.__name__ == "IRAPSClassifier":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
364 main_est.set_params(memory=memory)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
365
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
366 # handle scorer, convert to scorer dict
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
367 scoring = params['experiment_schemes']['metrics']['scoring']
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
368 if scoring is not None:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
369 # get_scoring() expects secondary_scoring to be a comma separated string (not a list)
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
370 # Check if secondary_scoring is specified
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
371 secondary_scoring = scoring.get("secondary_scoring", None)
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
372 if secondary_scoring is not None:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
373 # If secondary_scoring is specified, convert the list into comman separated string
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
374 scoring["secondary_scoring"] = ",".join(scoring["secondary_scoring"])
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
375
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
376 scorer = get_scoring(scoring)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
377 scorer, _ = _check_multimetric_scoring(estimator, scoring=scorer)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
378
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
379 # handle test (first) split
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
380 test_split_options = params["experiment_schemes"]["test_split"]["split_algos"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
381
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
382 if test_split_options["shuffle"] == "group":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
383 test_split_options["labels"] = groups
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
384 if test_split_options["shuffle"] == "stratified":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
385 if y is not None:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
386 test_split_options["labels"] = y
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
387 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
388 raise ValueError("Stratified shuffle split is not " "applicable on empty target values!")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
389
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
390 (
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
391 X_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
392 X_test,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
393 y_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
394 y_test,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
395 groups_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
396 _groups_test,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
397 ) = train_test_split_none(X, y, groups, **test_split_options)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
398
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
399 exp_scheme = params["experiment_schemes"]["selected_exp_scheme"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
400
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
401 # handle validation (second) split
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
402 if exp_scheme == "train_val_test":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
403 val_split_options = params["experiment_schemes"]["val_split"]["split_algos"]
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
404
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
405 if val_split_options["shuffle"] == "group":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
406 val_split_options["labels"] = groups_train
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
407 if val_split_options["shuffle"] == "stratified":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
408 if y_train is not None:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
409 val_split_options["labels"] = y_train
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
410 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
411 raise ValueError("Stratified shuffle split is not " "applicable on empty target values!")
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
412
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
413 (
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
414 X_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
415 X_val,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
416 y_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
417 y_val,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
418 groups_train,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
419 _groups_val,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
420 ) = train_test_split_none(X_train, y_train, groups_train, **val_split_options)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
421
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
422 # train and eval
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
423 if hasattr(estimator, "validation_data"):
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
424 if exp_scheme == "train_val_test":
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
425 estimator.fit(X_train, y_train, validation_data=(X_val, y_val))
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
426 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
427 estimator.fit(X_train, y_train, validation_data=(X_test, y_test))
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
428 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
429 estimator.fit(X_train, y_train)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
430
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
431 if hasattr(estimator, "evaluate"):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
432 steps = estimator.prediction_steps
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
433 batch_size = estimator.batch_size
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
434 generator = estimator.data_generator_.flow(X_test, y=y_test, batch_size=batch_size)
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
435 predictions, y_true = _predict_generator(estimator.model_, generator, steps=steps)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
436 scores = _evaluate(y_true, predictions, scorer, is_multimetric=True)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
437
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
438 else:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
439 if hasattr(estimator, "predict_proba"):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
440 predictions = estimator.predict_proba(X_test)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
441 else:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
442 predictions = estimator.predict(X_test)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
443
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
444 y_true = y_test
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
445 scores = _score(estimator, X_test, y_test, scorer, is_multimetric=True)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
446 if outfile_y_true:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
447 try:
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
448 pd.DataFrame(y_true).to_csv(outfile_y_true, sep="\t", index=False)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
449 pd.DataFrame(predictions).astype(np.float32).to_csv(
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
450 outfile_y_preds,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
451 sep="\t",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
452 index=False,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
453 float_format="%g",
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
454 chunksize=10000,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
455 )
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
456 except Exception as e:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
457 print("Error in saving predictions: %s" % e)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
458
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
459 # handle output
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
460 for name, score in scores.items():
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
461 scores[name] = [score]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
462 df = pd.DataFrame(scores)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
463 df = df[sorted(df.columns)]
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
464 df.to_csv(path_or_buf=outfile_result, sep="\t", header=True, index=False)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
465
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
466 memory.clear(warn=False)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
467
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
468 if outfile_object:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
469 main_est = estimator
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
470 if isinstance(estimator, Pipeline):
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
471 main_est = estimator.steps[-1][-1]
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
472
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
473 if hasattr(main_est, "model_") and hasattr(main_est, "save_weights"):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
474 if outfile_weights:
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
475 main_est.save_weights(outfile_weights)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
476 del main_est.model_
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
477 del main_est.fit_params
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
478 del main_est.model_class_
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
479 if getattr(main_est, "validation_data", None):
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
480 del main_est.validation_data
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
481 if getattr(main_est, "data_generator_", None):
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
482 del main_est.data_generator_
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
483
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
484 with open(outfile_object, "wb") as output_handler:
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
485 pickle.dump(estimator, output_handler, pickle.HIGHEST_PROTOCOL)
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
486
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
487
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
488 if __name__ == "__main__":
5
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
489 aparser = argparse.ArgumentParser()
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
490 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
491 aparser.add_argument("-e", "--estimator", dest="infile_estimator")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
492 aparser.add_argument("-X", "--infile1", dest="infile1")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
493 aparser.add_argument("-y", "--infile2", dest="infile2")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
494 aparser.add_argument("-O", "--outfile_result", dest="outfile_result")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
495 aparser.add_argument("-o", "--outfile_object", dest="outfile_object")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
496 aparser.add_argument("-w", "--outfile_weights", dest="outfile_weights")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
497 aparser.add_argument("-l", "--outfile_y_true", dest="outfile_y_true")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
498 aparser.add_argument("-p", "--outfile_y_preds", dest="outfile_y_preds")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
499 aparser.add_argument("-g", "--groups", dest="groups")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
500 aparser.add_argument("-r", "--ref_seq", dest="ref_seq")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
501 aparser.add_argument("-b", "--intervals", dest="intervals")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
502 aparser.add_argument("-t", "--targets", dest="targets")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
503 aparser.add_argument("-f", "--fasta_path", dest="fasta_path")
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
504 args = aparser.parse_args()
ed7c222e47e3 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 5b2ac730ec6d3b762faa9034eddd19ad1b347476"
bgruening
parents:
diff changeset
505
8
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
506 main(
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
507 args.inputs,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
508 args.infile_estimator,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
509 args.infile1,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
510 args.infile2,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
511 args.outfile_result,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
512 outfile_object=args.outfile_object,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
513 outfile_weights=args.outfile_weights,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
514 outfile_y_true=args.outfile_y_true,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
515 outfile_y_preds=args.outfile_y_preds,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
516 groups=args.groups,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
517 ref_seq=args.ref_seq,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
518 intervals=args.intervals,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
519 targets=args.targets,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
520 fasta_path=args.fasta_path,
449a757be9c9 "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit e2a5eade6d0e5ddf3a47630381a0ad90d80e8a04"
bgruening
parents: 5
diff changeset
521 )