Mercurial > repos > bgruening > sklearn_build_pipeline
annotate label_encoder.py @ 21:0c74d3164ac8 draft
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
| author | bgruening | 
|---|---|
| date | Sat, 01 May 2021 01:32:53 +0000 | 
| parents | |
| children | 118e230e85ce | 
| rev | line source | 
|---|---|
| 21 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 1 import argparse | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 2 import json | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 3 import warnings | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 4 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 5 import numpy as np | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 6 import pandas as pd | 
| 
0c74d3164ac8
"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 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 8 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 9 | 
| 
0c74d3164ac8
"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): | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 11 """ | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 12 Parameter | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 13 --------- | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 14 input : str | 
| 
0c74d3164ac8
"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 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 16 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 17 infile : str | 
| 
0c74d3164ac8
"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 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 19 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 20 outfile : str | 
| 
0c74d3164ac8
"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 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 22 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 23 """ | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 24 warnings.simplefilter('ignore') | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 25 | 
| 
0c74d3164ac8
"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: | 
| 
0c74d3164ac8
"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) | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 28 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 29 input_header = params['header0'] | 
| 
0c74d3164ac8
"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 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 31 | 
| 
0c74d3164ac8
"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) | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 33 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 34 le = LabelEncoder() | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 35 | 
| 
0c74d3164ac8
"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) | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 37 | 
| 
0c74d3164ac8
"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') | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 39 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 40 | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 41 if __name__ == '__main__': | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 42 aparser = argparse.ArgumentParser() | 
| 
0c74d3164ac8
"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) | 
| 
0c74d3164ac8
"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") | 
| 
0c74d3164ac8
"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") | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 46 args = aparser.parse_args() | 
| 
0c74d3164ac8
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
 bgruening parents: diff
changeset | 47 | 
| 
0c74d3164ac8
"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) | 
