Mercurial > repos > bgruening > sklearn_generalized_linear
annotate label_encoder.py @ 37:98d1d251f772 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit d970dae980fbe349414fc0889f719d875d999c5b"
| author | bgruening | 
|---|---|
| date | Fri, 27 Aug 2021 09:18:18 +0000 | 
| parents | 34f295eb5782 | 
| children | a8771df897b2 | 
| rev | line source | 
|---|---|
| 
36
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
1 import argparse | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
2 import json | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
3 import warnings | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
4 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
5 import numpy as np | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
6 import pandas as pd | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
7 from sklearn.preprocessing import LabelEncoder | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
8 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
9 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
10 def main(inputs, infile, outfile): | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
11 """ | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
12 Parameter | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
13 --------- | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
14 input : str | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
15 File path to galaxy tool parameter | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
16 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
17 infile : str | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
18 File paths of input vector | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
19 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
20 outfile : str | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
21 File path to output vector | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
22 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
23 """ | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
24 warnings.simplefilter('ignore') | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
25 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
26 with open(inputs, 'r') as param_handler: | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
27 params = json.load(param_handler) | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
28 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
29 input_header = params['header0'] | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
30 header = 'infer' if input_header else None | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
31 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
32 input_vector = pd.read_csv(infile, sep='\t', header=header) | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
33 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
34 le = LabelEncoder() | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
35 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
36 output_vector = le.fit_transform(input_vector) | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
37 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
38 np.savetxt(outfile, output_vector, fmt="%d", delimiter='\t') | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
39 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
40 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
41 if __name__ == '__main__': | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
42 aparser = argparse.ArgumentParser() | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
43 aparser.add_argument("-i", "--inputs", dest="inputs", required=True) | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
44 aparser.add_argument("-y", "--infile", dest="infile") | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
45 aparser.add_argument("-o", "--outfile", dest="outfile") | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
46 args = aparser.parse_args() | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
47 | 
| 
 
34f295eb5782
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 
bgruening 
parents:  
diff
changeset
 | 
48 main(args.inputs, args.infile, args.outfile) | 
