Mercurial > repos > recetox > matchms_formatter

annotate formatter.py @ 0:60f34912b3de draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
author recetox
date Tue, 22 Mar 2022 16:08:45 +0000
parents
children 574c6331e9db
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
1 import click
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
2 from pandas import DataFrame, read_csv
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
3
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
4
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
5 def create_long_table(data: DataFrame, value_id: str) -> DataFrame:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
6 """Convert the table from compact into long format.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
7 See DataFrame.melt(...).
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
8
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
9 Args:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
10 data (DataFrame): The data table to convert.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
11 value_id (str): The name to assign to the added column through conversion to long format.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
12
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
13 Returns:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
14 DataFrame: Table in long format.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
15 """
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
16 return data.transpose().melt(ignore_index=False, var_name='compound', value_name=value_id)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
17
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
18
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
19 def join_df(x: DataFrame, y: DataFrame, on=[], how="inner") -> DataFrame:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
20 """Shortcut functions to join to dataframes on columns and index
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
21
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
22 Args:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
23 x (DataFrame): Table X
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
24 y (DataFrame): Table Y
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
25 on (list, optional): Columns on which to join. Defaults to [].
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
26 how (str, optional): Join method, see DataFrame.join(...). Defaults to "inner".
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
27
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
28 Returns:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
29 DataFrame: Joined dataframe.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
30 """
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
31 df_x = x.set_index([x.index] + on)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
32 df_y = y.set_index([y.index] + on)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
33 combined = df_x.join(df_y, how=how)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
34 return combined
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
35
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
36
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
37 def get_top_k_matches(data: DataFrame, k: int) -> DataFrame:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
38 """Function to get top k matches from dataframe with scores.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
39
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
40 Args:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
41 data (DataFrame): A table with score column.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
42 k (int): Number of top scores to retrieve.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
43
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
44 Returns:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
45 DataFrame: Table containing only the top k best matches for each compound.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
46 """
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
47 return data.groupby(level=0, group_keys=False).apply(DataFrame.nlargest, n=k, columns=['score'])
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
48
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
49
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
50 def filter_thresholds(data: DataFrame, t_score: float, t_matches: float) -> DataFrame:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
51 """Filter a dataframe with scores and matches to only contain values above specified thresholds.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
52
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
53 Args:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
54 data (DataFrame): Table to filter.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
55 t_score (float): Score threshold.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
56 t_matches (float): Matches threshold.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
57
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
58 Returns:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
59 DataFrame: Filtered dataframe.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
60 """
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
61 filtered = data[data['score'] > t_score]
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
62 filtered = filtered[filtered['matches'] > t_matches]
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
63 return filtered
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
64
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
65
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
66 def load_data(scores_filename: str, matches_filename: str) -> DataFrame:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
67 """Load data from filenames and join on compound id.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
68
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
69 Args:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
70 scores_filename (str): Path to scores table.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
71 matches_filename (str): Path to matches table.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
72
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
73 Returns:
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
74 DataFrame: Joined dataframe on compounds containing scores an matches in long format.
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
75 """
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
76 matches = read_csv(matches_filename, sep=None, index_col=0)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
77 scores = read_csv(scores_filename, sep=None, index_col=0)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
78
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
79 scores_long = create_long_table(scores, 'score')
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
80 matches_long = create_long_table(matches, 'matches')
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
81
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
82 combined = join_df(matches_long, scores_long, on=['compound'], how='inner')
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
83 return combined
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
84
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
85
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
86 @click.group()
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
87 @click.option('--sf', 'scores_filename', type=click.Path(exists=True), required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
88 @click.option('--mf', 'matches_filename', type=click.Path(exists=True), required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
89 @click.option('--o', 'output_filename', type=click.Path(writable=True), required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
90 @click.pass_context
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
91 def cli(ctx, scores_filename, matches_filename, output_filename):
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
92 ctx.ensure_object(dict)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
93 ctx.obj['data'] = load_data(scores_filename, matches_filename)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
94 pass
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
95
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
96
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
97 @cli.command()
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
98 @click.option('--st', 'scores_threshold', type=float, required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
99 @click.option('--mt', 'matches_threshold', type=float, required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
100 @click.pass_context
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
101 def get_thresholded_data(ctx, scores_threshold, matches_threshold):
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
102 result = filter_thresholds(ctx.obj['data'], scores_threshold, matches_threshold)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
103 return result
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
104
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
105
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
106 @cli.command()
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
107 @click.option('--k', 'k', type=int, required=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
108 @click.pass_context
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
109 def get_top_k_data(ctx, k):
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
110 result = get_top_k_matches(ctx.obj['data'], k)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
111 return result
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
112
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
113
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
114 @cli.resultcallback()
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
115 def write_output(result: DataFrame, scores_filename, matches_filename, output_filename):
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
116 input_file = read_csv(scores_filename, sep=None, iterator=True)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
117 sep = input_file._engine.data.dialect.delimiter
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
118
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
119 result = result.reset_index().rename(columns={'level_0': 'query', 'compound': 'reference'})
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
120 result.to_csv(output_filename, sep=sep, index=False)
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
121
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
122
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
123 if __name__ == '__main__':
60f34912b3de "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/matchms commit 4d2ac914c951166e386a94d8ebb8cb1becfac122"
recetox
parents:
diff changeset
124 cli(obj={})