annotate prepare_data.py @ 0:9bf25dbe00ad draft

"planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
author bgruening
date Wed, 28 Aug 2019 07:19:38 -0400
parents
children 5b3c08710e47
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
1 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
2 Prepare the workflow paths to be used by downstream
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
3 machine learning algorithm. The paths are divided
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
4 into the test and training sets
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
5 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
6
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
7 import os
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
8 import collections
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
9 import numpy as np
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
10 import random
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
11
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
12 import predict_tool_usage
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
13
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
14 main_path = os.getcwd()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
15
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
16
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
17 class PrepareData:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
18
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
19 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
20 def __init__(self, max_seq_length, test_data_share):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
21 """ Init method. """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
22 self.max_tool_sequence_len = max_seq_length
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
23 self.test_share = test_data_share
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
24
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
25 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
26 def process_workflow_paths(self, workflow_paths):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
27 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
28 Get all the tools and complete set of individual paths for each workflow
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
29 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
30 tokens = list()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
31 raw_paths = workflow_paths
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
32 raw_paths = [x.replace("\n", '') for x in raw_paths]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
33 for item in raw_paths:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
34 split_items = item.split(",")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
35 for token in split_items:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
36 if token is not "":
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
37 tokens.append(token)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
38 tokens = list(set(tokens))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
39 tokens = np.array(tokens)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
40 tokens = np.reshape(tokens, [-1, ])
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
41 return tokens, raw_paths
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
42
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
43 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
44 def create_new_dict(self, new_data_dict):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
45 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
46 Create new data dictionary
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
47 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
48 reverse_dict = dict((v, k) for k, v in new_data_dict.items())
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
49 return new_data_dict, reverse_dict
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
50
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
51 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
52 def assemble_dictionary(self, new_data_dict, old_data_dictionary={}):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
53 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
54 Create/update tools indices in the forward and backward dictionary
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
55 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
56 new_data_dict, reverse_dict = self.create_new_dict(new_data_dict)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
57 return new_data_dict, reverse_dict
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
58
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
59 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
60 def create_data_dictionary(self, words, old_data_dictionary={}):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
61 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
62 Create two dictionaries having tools names and their indexes
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
63 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
64 count = collections.Counter(words).most_common()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
65 dictionary = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
66 for word, _ in count:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
67 dictionary[word] = len(dictionary) + 1
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
68 dictionary, reverse_dictionary = self.assemble_dictionary(dictionary, old_data_dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
69 return dictionary, reverse_dictionary
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
70
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
71 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
72 def decompose_paths(self, paths, dictionary):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
73 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
74 Decompose the paths to variable length sub-paths keeping the first tool fixed
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
75 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
76 sub_paths_pos = list()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
77 for index, item in enumerate(paths):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
78 tools = item.split(",")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
79 len_tools = len(tools)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
80 if len_tools <= self.max_tool_sequence_len:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
81 for window in range(1, len_tools):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
82 sequence = tools[0: window + 1]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
83 tools_pos = [str(dictionary[str(tool_item)]) for tool_item in sequence]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
84 if len(tools_pos) > 1:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
85 sub_paths_pos.append(",".join(tools_pos))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
86 sub_paths_pos = list(set(sub_paths_pos))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
87 return sub_paths_pos
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
88
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
89 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
90 def prepare_paths_labels_dictionary(self, dictionary, reverse_dictionary, paths, compatible_next_tools):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
91 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
92 Create a dictionary of sequences with their labels for training and test paths
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
93 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
94 paths_labels = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
95 random.shuffle(paths)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
96 for item in paths:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
97 if item and item not in "":
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
98 tools = item.split(",")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
99 label = tools[-1]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
100 train_tools = tools[:len(tools) - 1]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
101 last_but_one_name = reverse_dictionary[int(train_tools[-1])]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
102 try:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
103 compatible_tools = compatible_next_tools[last_but_one_name].split(",")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
104 except Exception:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
105 continue
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
106 if len(compatible_tools) > 0:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
107 compatible_tools_ids = [str(dictionary[x]) for x in compatible_tools]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
108 compatible_tools_ids.append(label)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
109 composite_labels = ",".join(compatible_tools_ids)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
110 train_tools = ",".join(train_tools)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
111 if train_tools in paths_labels:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
112 paths_labels[train_tools] += "," + composite_labels
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
113 else:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
114 paths_labels[train_tools] = composite_labels
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
115 for item in paths_labels:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
116 paths_labels[item] = ",".join(list(set(paths_labels[item].split(","))))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
117 return paths_labels
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
118
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
119 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
120 def pad_paths(self, paths_dictionary, num_classes):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
121 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
122 Add padding to the tools sequences and create multi-hot encoded labels
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
123 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
124 size_data = len(paths_dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
125 data_mat = np.zeros([size_data, self.max_tool_sequence_len])
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
126 label_mat = np.zeros([size_data, num_classes + 1])
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
127 train_counter = 0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
128 for train_seq, train_label in list(paths_dictionary.items()):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
129 positions = train_seq.split(",")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
130 start_pos = self.max_tool_sequence_len - len(positions)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
131 for id_pos, pos in enumerate(positions):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
132 data_mat[train_counter][start_pos + id_pos] = int(pos)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
133 for label_item in train_label.split(","):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
134 label_mat[train_counter][int(label_item)] = 1.0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
135 train_counter += 1
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
136 return data_mat, label_mat
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
137
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
138 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
139 def split_test_train_data(self, multilabels_paths):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
140 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
141 Split into test and train data randomly for each run
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
142 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
143 train_dict = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
144 test_dict = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
145 all_paths = multilabels_paths.keys()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
146 random.shuffle(list(all_paths))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
147 split_number = int(self.test_share * len(all_paths))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
148 for index, path in enumerate(list(all_paths)):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
149 if index < split_number:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
150 test_dict[path] = multilabels_paths[path]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
151 else:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
152 train_dict[path] = multilabels_paths[path]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
153 return train_dict, test_dict
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
154
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
155 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
156 def verify_overlap(self, train_paths, test_paths):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
157 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
158 Verify the overlapping of samples in train and test data
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
159 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
160 intersection = list(set(train_paths).intersection(set(test_paths)))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
161 print("Overlap in train and test: %d" % len(intersection))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
162
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
163 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
164 def get_predicted_usage(self, data_dictionary, predicted_usage):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
165 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
166 Get predicted usage for tools
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
167 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
168 usage = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
169 epsilon = 0.0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
170 # index 0 does not belong to any tool
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
171 usage[0] = epsilon
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
172 for k, v in data_dictionary.items():
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
173 try:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
174 usg = predicted_usage[k]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
175 if usg < epsilon:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
176 usg = epsilon
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
177 usage[v] = usg
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
178 except Exception:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
179 usage[v] = epsilon
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
180 continue
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
181 return usage
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
182
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
183 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
184 def assign_class_weights(self, n_classes, predicted_usage):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
185 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
186 Compute class weights using usage
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
187 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
188 class_weights = dict()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
189 class_weights[str(0)] = 0.0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
190 for key in range(1, n_classes):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
191 u_score = predicted_usage[key]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
192 if u_score < 1.0:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
193 u_score += 1.0
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
194 class_weights[key] = np.log(u_score)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
195 return class_weights
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
196
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
197 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
198 def get_sample_weights(self, train_data, reverse_dictionary, paths_frequency):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
199 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
200 Compute the frequency of paths in training data
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
201 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
202 path_weights = np.zeros(len(train_data))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
203 for path_index, path in enumerate(train_data):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
204 sample_pos = np.where(path > 0)[0]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
205 sample_tool_pos = path[sample_pos[0]:]
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
206 path_name = ",".join([reverse_dictionary[int(tool_pos)] for tool_pos in sample_tool_pos])
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
207 try:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
208 path_weights[path_index] = int(paths_frequency[path_name])
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
209 except Exception:
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
210 path_weights[path_index] = 1
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
211 return path_weights
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
212
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
213 @classmethod
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
214 def get_data_labels_matrices(self, workflow_paths, tool_usage_path, cutoff_date, compatible_next_tools, old_data_dictionary={}):
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
215 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
216 Convert the training and test paths into corresponding numpy matrices
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
217 """
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
218 processed_data, raw_paths = self.process_workflow_paths(workflow_paths)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
219 dictionary, reverse_dictionary = self.create_data_dictionary(processed_data, old_data_dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
220 num_classes = len(dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
221
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
222 print("Raw paths: %d" % len(raw_paths))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
223 random.shuffle(raw_paths)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
224
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
225 print("Decomposing paths...")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
226 all_unique_paths = self.decompose_paths(raw_paths, dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
227 random.shuffle(all_unique_paths)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
228
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
229 print("Creating dictionaries...")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
230 multilabels_paths = self.prepare_paths_labels_dictionary(dictionary, reverse_dictionary, all_unique_paths, compatible_next_tools)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
231
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
232 print("Complete data: %d" % len(multilabels_paths))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
233 train_paths_dict, test_paths_dict = self.split_test_train_data(multilabels_paths)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
234
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
235 print("Train data: %d" % len(train_paths_dict))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
236 print("Test data: %d" % len(test_paths_dict))
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
237
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
238 test_data, test_labels = self.pad_paths(test_paths_dict, num_classes)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
239 train_data, train_labels = self.pad_paths(train_paths_dict, num_classes)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
240
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
241 # Predict tools usage
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
242 print("Predicting tools' usage...")
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
243 usage_pred = predict_tool_usage.ToolPopularity()
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
244 usage = usage_pred.extract_tool_usage(tool_usage_path, cutoff_date, dictionary)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
245 tool_usage_prediction = usage_pred.get_pupularity_prediction(usage)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
246 tool_predicted_usage = self.get_predicted_usage(dictionary, tool_usage_prediction)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
247
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
248 # get class weights using the predicted usage for each tool
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
249 class_weights = self.assign_class_weights(train_labels.shape[1], tool_predicted_usage)
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
250
9bf25dbe00ad "planemo upload for repository https://github.com/bgruening/galaxytools/tree/recommendation_training/tools/tool_recommendation_model commit 7fac577189d01cedd01118a77fc2baaefe7d5cad"
bgruening
parents:
diff changeset
251 return train_data, train_labels, test_data, test_labels, dictionary, reverse_dictionary, class_weights, tool_predicted_usage