Mercurial > repos > bgruening > sklearn_estimator_attributes
annotate label_encoder.py @ 17:2a16adf9a353 draft
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
author | bgruening |
---|---|
date | Wed, 09 Aug 2023 12:59:27 +0000 |
parents | dd157beef3e4 |
children |
rev | line source |
---|---|
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
1 import argparse |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
2 import json |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
3 import warnings |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
4 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
5 import numpy as np |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
6 import pandas as pd |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
7 from sklearn.preprocessing import LabelEncoder |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
8 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
9 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
10 def main(inputs, infile, outfile): |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
11 """ |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
12 Parameter |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
13 --------- |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
14 input : str |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
15 File path to galaxy tool parameter |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
16 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
17 infile : str |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
18 File paths of input vector |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
19 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
20 outfile : str |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
21 File path to output vector |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
22 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
23 """ |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
24 warnings.simplefilter("ignore") |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
25 |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
26 with open(inputs, "r") as param_handler: |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
27 params = json.load(param_handler) |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
28 |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
29 input_header = params["header0"] |
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
30 header = "infer" if input_header else None |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
31 |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
32 input_vector = pd.read_csv(infile, sep="\t", header=header) |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
33 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
34 le = LabelEncoder() |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
35 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
36 output_vector = le.fit_transform(input_vector) |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
37 |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
38 np.savetxt(outfile, output_vector, fmt="%d", delimiter="\t") |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
39 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
40 |
16
dd157beef3e4
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 9981e25b00de29ed881b2229a173a8c812ded9bb
bgruening
parents:
11
diff
changeset
|
41 if __name__ == "__main__": |
11
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
42 aparser = argparse.ArgumentParser() |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
43 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
44 aparser.add_argument("-y", "--infile", dest="infile") |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
45 aparser.add_argument("-o", "--outfile", dest="outfile") |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
46 args = aparser.parse_args() |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
47 |
16d3fe9f96d7
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff
changeset
|
48 main(args.inputs, args.infile, args.outfile) |