annotate label_encoder.py @ 9:022e85ead64f draft

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
author bgruening
date Tue, 20 Apr 2021 16:45:52 +0000
parents
children 0edcdeaad6f4
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
9
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
1 import argparse
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
2 import json
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
3 import warnings
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
4
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
5 import numpy as np
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
6 import pandas as pd
022e85ead64f "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
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
8
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
9
022e85ead64f "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):
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
11 """
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
12 Parameter
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
13 ---------
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
14 input : str
022e85ead64f "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
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
16
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
17 infile : str
022e85ead64f "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
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
19
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
20 outfile : str
022e85ead64f "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
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
22
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
23 """
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
24 warnings.simplefilter('ignore')
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
25
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
26 with open(inputs, 'r') as param_handler:
022e85ead64f "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)
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
28
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
29 input_header = params['header0']
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
30 header = 'infer' if input_header else None
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
31
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
32 input_vector = pd.read_csv(infile, sep='\t', header=header)
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
33
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
34 le = LabelEncoder()
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
35
022e85ead64f "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)
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
37
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
38 np.savetxt(outfile, output_vector, fmt="%d", delimiter='\t')
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
39
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
40
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
41 if __name__ == '__main__':
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
42 aparser = argparse.ArgumentParser()
022e85ead64f "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)
022e85ead64f "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")
022e85ead64f "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")
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
46 args = aparser.parse_args()
022e85ead64f "planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ca87db9c038a6fcf96aa39da50f384865fd932ff"
bgruening
parents:
diff changeset
47
022e85ead64f "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)