Mercurial > repos > goeckslab > bagging_tool

comparison mil_bag.xml @ 0:e6e9ea0703ef draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/goeckslab/gleam.git commit 783551569c645073698fce50f1ed9c4605b3e65a
author goeckslab
date Thu, 19 Jun 2025 23:31:55 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:e6e9ea0703ef
1 <tool id="bagging_tool" name="Bagging Embeddings Processor" version="1.0.0+0">
2 <description>Process CSV files to create bags of embeddings for machine learning</description>
3 <requirements>
4 <container type="docker">quay.io/goeckslab/milbag:1.0.0</container>
5 </requirements>
6 <stdio>
7 <exit_code range="137" level="fatal_oom" description="Out of Memory" />
8 <exit_code range="1:" level="fatal" description="Error occurred. Please check Tool Standard Error" />
9 </stdio>
10 <command>
11 python "$__tool_directory__/mil_bag.py"
12 --embeddings_csv "$embeddings_csv"
13 --metadata_csv "$metadata_csv"
14 --split_proportions "$split_proportions"
15 --bag_size "$bag_size"
16 --pooling_method "$pooling_method"
17 --repeats "$repeats"
18 --output_csv "$output_csv"
19 #if $dataleak
20 --dataleak
21 #end if
22 #if $balance_enforced
23 --balance_enforced
24 #end if
25 #if $ludwig_format
26 --ludwig_format
27 #end if
28 #if $random_seed != ""
29 --random_seed "$random_seed"
30 #end if
31 #if $imbalance_cap != ""
32 --imbalance_cap "$imbalance_cap"
33 #end if
34 #if $truncate_bags
35 --truncate_bags
36 #end if
37 #if $use_gpu
38 --use_gpu
39 #end if
40 #if $by_sample != ""
41 --by_sample "$by_sample"
42 #end if
43 </command>
44 <inputs>
45 <param name="embeddings_csv" type="data" format="csv" label="Embeddings CSV File" help="CSV file containing embeddings with a 'sample_name' column."/>
46 <param name="metadata_csv" type="data" format="csv" label="Metadata CSV File" help="CSV file with metadata containing 'sample_name' and 'label' columns."/>
47 <param name="split_proportions" type="text" value="0.7,0.1,0.2" label="Split Proportions (train,val,test)" help="Comma-separated proportions (e.g., '0.7,0.1,0.2') for train, validation, and test splits."/>
48 <param name="bag_size" type="text" value="3-5" label="Bag Size" help="Single number (e.g., '4') or range (e.g., '3-5') for bag sizes."/>
49 <param name="pooling_method" type="select" label="Pooling Method" help="Method to aggregate embeddings into bags.">
50 <option value="max_pooling">Max Pooling</option>
51 <option value="mean_pooling">Mean Pooling</option>
52 <option value="sum_pooling">Sum Pooling</option>
53 <option value="min_pooling">Min Pooling</option>
54 <option value="median_pooling">Median Pooling</option>
55 <option value="l2_norm_pooling">L2 Norm Pooling</option>
56 <option value="geometric_mean_pooling">Geometric Mean Pooling</option>
57 <option value="first_embedding">First Embedding</option>
58 <option value="last_embedding">Last Embedding</option>
59 <option value="attention_pooling">Attention Pooling</option>
60 </param>
61 <param name="repeats" type="integer" value="1" min="1" label="Number of Repeats" help="Number of times to repeat the bagging process."/>
62 <param name="dataleak" type="boolean" truevalue="--dataleak" falsevalue="" checked="false" label="Prevent Data Leakage?" help="If checked, prevents data leakage by splitting on unique sample names."/>
63 <param name="balance_enforced" type="boolean" truevalue="--balance_enforced" falsevalue="" checked="false" label="Enforce Balanced Bags?" help="If checked, alternates between classes to create balanced bags."/>
64 <param name="ludwig_format" type="boolean" truevalue="--ludwig_format" falsevalue="" checked="false" label="Ludwig Format?" help="If checked, outputs embeddings as a single string column for Ludwig compatibility."/>
65 <param name="by_sample" type="text" value="" optional="true" label="Splits for Within-Sample Bagging" help="Optional comma-separated list of splits (0=train, 1=val, 2=test) to bag within samples (e.g., '0,1'). Defaults to random or balanced bagging if empty."/>
66 <param name="random_seed" type="integer" value="" optional="true" label="Random Seed" help="Optional integer seed for reproducibility (e.g., 42). Leave blank for random behavior."/>
67 <param name="imbalance_cap" type="integer" value="" optional="true" label="Maximum Imbalance Percentage" help="Optional maximum allowable imbalance percentage between classes (e.g., 50). If set, balances bags to this threshold."/>
68 <param name="truncate_bags" type="boolean" truevalue="--truncate_bags" falsevalue="" checked="false" label="Truncate Bags for Balance?" help="If checked, truncates bags to ensure equal counts of positive and negative bags."/>
69 <param name="use_gpu" type="boolean" truevalue="--use_gpu" falsevalue="" checked="false" label="Use GPU?" help="If checked, uses GPU for pooling operations (requires compatible hardware and libraries)."/>
70 </inputs>
71 <outputs>
72 <data name="output_csv" format="csv" label="processed_bags.csv"/>
73 </outputs>
74 <tests>
75 <test>
76 <param name="embeddings_csv" value="100_digits_embeddings.csv" />
77 <param name="metadata_csv" value="100_digits_metadata.csv" />
78 <param name="split_proportions" value="0.7,0.2,0.1" />
79 <param name="bag_size" value="2" />
80 <param name="pooling_method" value="mean_pooling" />
81 <param name="repeats" value="1" />
82 <param name="dataleak" value="true" />
83 <param name="balance_enforced" value="false" />
84 <param name="ludwig_format" value="true" />
85 <param name="by_sample" value="" />
86 <param name="random_seed" value="42" />
87 <param name="imbalance_cap" value="" />
88 <param name="truncate_bags" value="false" />
89 <param name="use_gpu" value="false" />
90 <output name="output_csv">
91 <assert_contents>
92 <has_text text="bag_size" />
93 <has_n_columns min="1" />
94 </assert_contents>
95 </output>
96 </test>
97 <test>
98 <param name="embeddings_csv" value="100_digits_embeddings.csv" />
99 <param name="metadata_csv" value="100_digits_metadata.csv" />
100 <param name="split_proportions" value="0.7,0.2,0.1" />
101 <param name="bag_size" value="2" />
102 <param name="pooling_method" value="mean_pooling" />
103 <param name="repeats" value="1" />
104 <param name="dataleak" value="true" />
105 <param name="balance_enforced" value="false" />
106 <param name="ludwig_format" value="true" />
107 <param name="by_sample" value="2" />
108 <param name="random_seed" value="123" />
109 <param name="imbalance_cap" value="50" />
110 <param name="truncate_bags" value="true" />
111 <param name="use_gpu" value="true" />
112 <output name="output_csv">
113 <assert_contents>
114 <has_text text="bag_size" />
115 <has_n_columns min="1" />
116 </assert_contents>
117 </output>
118 </test>
119 </tests>
120 <help>
121 <![CDATA[
122 **What it does**
123 This tool processes embedding and metadata CSV files to create bags of samples with specified sizes and pooling methods, suitable for machine learning tasks.
124
125 **Inputs**
126 - **Embeddings CSV File**: A CSV file containing embeddings with a `sample_name` column.
127 - **Metadata CSV File**: A CSV file with metadata containing `sample_name` and `label` columns.
128 - **Split Proportions**: Define train, validation, and test split ratios (e.g., '0.7,0.1,0.2').
129 - **Bag Size**: Set a fixed number (e.g., '4') or range (e.g., '3-5') for bag sizes.
130 - **Pooling Method**: Choose how embeddings are aggregated into bags (e.g., mean, max).
131 - **Number of Repeats**: Specify how many times to repeat bagging (useful for augmentation).
132 - **Prevent Data Leakage**: Avoid leakage by splitting on unique sample names.
133 - **Enforce Balanced Bags**: Alternate classes for balanced bagging.
134 - **Ludwig Format**: Convert embeddings to a single string column for Ludwig compatibility.
135 - **Splits for Within-Sample Bagging**: Optional splits (0, 1, 2) to bag within samples (e.g., '0,1').
136 - **Random Seed**: Optional seed for reproducible results.
137 - **Maximum Imbalance Percentage**: Optional cap (e.g., 50) to balance class distribution.
138 - **Truncate Bags for Balance**: Truncate bags to equalize positive and negative counts.
139 - **Use GPU**: Enable GPU acceleration for pooling operations (if available).
140
141 **Outputs**
142 - A CSV file with bags of embeddings, including labels, split information, and processed embedding vectors.
143 ]]>
144 </help>
145 </tool>