annotate main.py @ 1:79c9b4b34b63 draft default tip

planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ef62aa3d86abd4b911e35447646712a4628e7fe
author rnateam
date Fri, 29 Jul 2016 03:27:44 -0400
parents d04fa5201f51
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
1 #!/usr/bin/env python
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
2 """Recommendation."""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
3
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
4 import argparse
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
5 import sys
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
6 from rbpfeatures import RBPVectorizer
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
7 from data import PredictDataset
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
8 from recommend import Predictor
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
9
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
10 __author__ = "Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
11 __copyright__ = "Copyright 2016, Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
12 __license__ = "MIT"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
13 __maintainer__ = "Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
14 __email__ = "gianluca.corrado@unitn.it"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
15 __status__ = "Production"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
16
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
17
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
18 if __name__ == '__main__':
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
19 parser = argparse.ArgumentParser(
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
20 description=__doc__,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
21 formatter_class=argparse.ArgumentDefaultsHelpFormatter)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
22 parser.add_argument('fasta', metavar='fasta', type=str,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
23 help="""Fasta file containing the RBP \
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
24 sequences.""")
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
25
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
26 args = parser.parse_args()
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
27
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
28 v = RBPVectorizer(fasta=args.fasta)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
29 rbp_fea = v.vectorize()
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
30
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
31 if rbp_fea is not None:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
32 # Define and load dataset
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
33 D = PredictDataset(
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
34 fp=rbp_fea, fr="AURA_Human_data/RNA_features/HT_utrs.h5")
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
35 dataset = D.load()
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
36
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
37 model = "AURA_Human_data/model/trained_model.pkl"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
38
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
39 # Define the Trainer and train the model
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
40 P = Predictor(predict_dataset=dataset,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
41 trained_model=model,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
42 serendipity_dic=model + '_',
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
43 output="output.txt")
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
44 P.predict()
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
45 else:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
46 sys.stdout.write("""
1
79c9b4b34b63 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ef62aa3d86abd4b911e35447646712a4628e7fe
rnateam
parents: 0
diff changeset
47 ########################################
79c9b4b34b63 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ef62aa3d86abd4b911e35447646712a4628e7fe
rnateam
parents: 0
diff changeset
48 WARNING: The queried protein has no domain similarity with the proteins in the training dataset. It cannot be predicted.
79c9b4b34b63 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ef62aa3d86abd4b911e35447646712a4628e7fe
rnateam
parents: 0
diff changeset
49 ########################################
0
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
50 """)