view macros.xml @ 3:0a8fe19cebeb draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/flexynesis commit b2463fb68d0ae54864d87718ee72f5e063aa4587

<macros>
    <token name="@TOOL_VERSION@">0.2.20</token>
    <token name="@VERSION_SUFFIX@">0</token>
    <token name="@PROFILE@">24.1</token>
    <xml name="requirements">
        <requirements>
            <requirement type="package" version="@TOOL_VERSION@">flexynesis</requirement>
            <yield/>
        </requirements>
    </xml>
    <xml name="edam">
        <edam_topics>
            <edam_topic>topic_0622</edam_topic>
            <edam_topic>topic_3474</edam_topic>
            <edam_topic>topic_2640</edam_topic>
        </edam_topics>
        <edam_operations>
            <edam_operation>operation_3197</edam_operation>
            <edam_operation>operation_2403</edam_operation>
            <edam_operation>operation_2426</edam_operation>
        </edam_operations>
    </xml>
    <xml name="sanitizer_printable">
        <sanitizer invalid_char="">
            <valid initial="string.printable">
                <remove value="'"/>
                <remove value='"'/>
                <remove value=" "/>
                <yield/>
            </valid>
        </sanitizer>
    </xml>
    <xml name="sanitizer_letters">
        <sanitizer invalid_char=" ">
            <valid initial="string.letters">
                <add value="_"/>
            </valid>
        </sanitizer>
    </xml>
    <token name="@CHECK_NON_COMMERCIAL_USE@"><![CDATA[
        #if not $non_commercial_use
            >&2 echo "this tool is only available for non commercial use";
            exit 1;
        #end if
    ]]></token>
    <xml name="commercial_use_param">
        <param name="non_commercial_use" label="I certify that I am not using this tool for commercial purposes." type="boolean" truevalue="NON_COMMERCIAL_USE" falsevalue="COMMERCIAL_USE" checked="False">
            <validator type="expression" message="This tool is only available for non-commercial use.">value == True</validator>
        </param>
    </xml>
    <xml name="main_inputs">
        <param name="train_clin" type="data" format="csv" label="Training clinical data"/>
        <param name="test_clin" type="data" format="csv" label="Test clinical data"/>
        <param name="train_omics_main" type="data" format="csv" label="Training omics data"/>
        <param name="test_omics_main" type="data" format="csv" label="Test omics data"/>
        <param name="assay_main" type="text" optional="true" label="What type of assay is your input?" help="This would be used as output name.">
            <expand macro="sanitizer_letters"/>
        </param>
    </xml>
    <xml name="extra_inputs">
        <param name="train_omics" type="data" optional="true" format="csv" label="Training omics data"/>
        <param name="test_omics" type="data" optional="true" format="csv" label="Test omics data"/>
        <param name="assay" type="text" optional="true" label="What type of assay is your input?" help="This would be used as output name." >
            <expand macro="sanitizer_letters"/>
        </param>
    </xml>
    <xml name="advanced">
        <section name="advanced" title="Advanced Options">
            <param argument="--fusion_type" type="select" label="Fusion method" help="How to fuse the omics layers?">
                <option value="intermediate">intermediate</option>
                <option value="early">early</option>
            </param>
            <param argument="--finetuning_samples" type="integer" min="0" value="0" label="Number of samples from the test dataset to use for fine-tuning the model." help="Set to 0 to disable fine-tuning."/>
            <param argument="--variance_threshold" type="float" min="0" max="100" value="1" label="Variance threshold (as percentile) to drop low variance features." help="Set to 0 for no variance filtering."/>
            <param argument="--correlation_threshold" type="float" min="0" max="1" value="0.8" label="Correlation threshold to drop highly redundant features." help="Set to 1 for no redundancy filtering."/>
            <param argument="--subsample" type="integer" min="0" value="0" label="Downsample training set to randomly drawn N samples for training."/>
            <param argument="--features_min" type="integer" min="0" value="500" label="Minimum number of features to retain after feature selection."/>
            <param argument="--features_top_percentile" type="float" min="0" max="100" value="20" label="Top percentile features (among the features remaining after variance filtering and data cleanup) to retain after feature selection."/>
            <param argument="--log_transform" type="boolean" truevalue="--log_transform True" falsevalue="" checked="false" label="Whether to apply log-transformation to input data matrices"/>
            <param argument="--early_stop_patience" type="integer" min="-1" value="10" label="How many epochs to wait when no improvements in validation loss are observed." help="Set to -1 to disable early stopping."/>
            <param argument="--hpo_iter" type="integer" min="1" value="100" label="Number of iterations for hyperparameter optimization."/>
            <param argument="--val_size" type="float" min="0.0" max="1" value="0.2" label="Proportion of training data to be used as validation split"/>
            <param argument="--hpo_patience" type="integer" min="0" value="10" label="How many hyperparameter optimization iterations to wait for when no improvements are observed." help="Set to 0 to disable early stopping."/>
            <param argument="--use_cv" type="boolean" truevalue="--use_cv" falsevalue="" checked="false" label="Cross validation" help="If set, a 5-fold cross-validation training will be done. Otherwise, a single training on 80 percent of the dataset is done. "/>
            <param argument="--use_loss_weighting" type="boolean" truevalue="--use_loss_weighting True" falsevalue="" checked="true" label="Whether to apply loss-balancing using uncertainty weights method."/>
            <param argument="--evaluate_baseline_performance" type="boolean" truevalue="--evaluate_baseline_performance" falsevalue="" checked="false" label="Enable modeling also with Random Forest + SVMs to see the performance of off-the-shelf tools on the same dataset."/>
            <param argument="--feature_importance_method" type="select" label="which method(s) to use to compute feature importance scores.">
                <option value="Both" selected="true">Both</option>
                <option value="IntegratedGradients">IntegratedGradients</option>
                <option value="GradientShap">GradientShap</option>
            </param>
        </section>
    </xml>
    <xml name="plots_common_param">
        <yield/>
        <param name="format" type="select" label="Output format">
            <option value="jpg" selected="true">jpg</option>
            <option value="png">png</option>
            <option value="pdf">pdf</option>
            <option value="svg">svg</option>
        </param>
        <param name="dpi" type="integer" min="0" max="1200" value="300" label="DPI"/>
    </xml>
    <xml name="plots_common_input">
        <yield/>
        <param argument="--labels" type="data" format="tabular,csv" label="Predicted labels" help="Generated by flexynesis"/>
    </xml>
    <token name="@PLOT_COMMON_CONFIG@"><![CDATA[
label_data = load_labels('inputs/$plot_conditional.labels.element_identifier.$plot_conditional.labels.ext')
    ]]></token>
    <token name="@PR_ROC_BOX_CONFIG@"><![CDATA[
@PLOT_COMMON_CONFIG@

# Check if this is a regression problem (no class probabilities)
non_na_probs = label_data['probability'].notna().sum()

print(f"  Non-NaN probabilities: {non_na_probs}/{len(label_data)}")

# If most probabilities are NaN, this is likely a regression problem
if non_na_probs < len(label_data) * 0.1:  # Less than 10% valid probabilities
    raise ValueError("  Detected regression problem - precision-recall curves not applicable")

# Debug: Check data quality
total_rows = len(label_data)
missing_labels = label_data['known_label'].isna().sum()
missing_probs = label_data['probability'].isna().sum()
unique_samples = label_data['sample_id'].nunique()
unique_classes = label_data['class_label'].nunique()

print(f"  Data summary: {total_rows} total rows, {unique_samples} unique samples, {unique_classes} unique classes")
print(f"  Missing data: {missing_labels} missing known_label, {missing_probs} missing probability")

if missing_labels > 0:
    print(f"  Warning: Found {missing_labels} missing known_label values")
    missing_samples = label_data[label_data['known_label'].isna()]['sample_id'].unique()[:5]
    print(f"  Sample IDs with missing known_label: {list(missing_samples)}")

    # Remove rows with missing known_label
    label_data = label_data.dropna(subset=['known_label'])
    if label_data.empty:
        raise ValueError("Error: No valid known_label data remaining")

    ]]></token>
    <token name="@PR_ROC_CONFIG@"><![CDATA[
@PR_ROC_BOX_CONFIG@

# 1. Pivot to wide format
prob_df = label_data.pivot(index='sample_id', columns='class_label', values='probability')

print(f"  After pivot: {prob_df.shape[0]} samples x {prob_df.shape[1]} classes")
print(f"  Class columns: {list(prob_df.columns)}")

# Check for NaN values in probability data
nan_counts = prob_df.isna().sum()
if nan_counts.any():
    print(f"  NaN counts per class: {dict(nan_counts)}")
    print(f"  Samples with any NaN: {prob_df.isna().any(axis=1).sum()}/{len(prob_df)}")

    # Drop only rows where ALL probabilities are NaN
    all_nan_rows = prob_df.isna().all(axis=1)
    if all_nan_rows.any():
        print(f"  Dropping {all_nan_rows.sum()} samples with all NaN probabilities")
        prob_df = prob_df[~all_nan_rows]

    remaining_nans = prob_df.isna().sum().sum()
    if remaining_nans > 0:
        print(f"  Warning: {remaining_nans} individual NaN values remain - filling with 0")
        prob_df = prob_df.fillna(0)

    if prob_df.empty:
        raise ValueError(f"  Error: No valid probability data remaining for")

# 2. Get true labels
true_labels_df = label_data.drop_duplicates('sample_id')[['sample_id', 'known_label']].set_index('sample_id')

# 3. Align indices - only keep samples that exist in both datasets
common_indices = prob_df.index.intersection(true_labels_df.index)
if len(common_indices) == 0:
    raise ValueError(f"  Error: No common sample_ids between probability and true label data")

print(f"  Found {len(common_indices)} samples with both probability and true label data")

# Filter both datasets to common indices
prob_df_aligned = prob_df.loc[common_indices]
y_true = true_labels_df.loc[common_indices]['known_label']

# 4. Final check for NaN values
if y_true.isna().any():
    raise ValueError(f"  Error: True labels still contain NaN after alignment")

if prob_df_aligned.isna().any().any():
    raise ValueError(f"  Error: Probability data still contains NaN after alignment")

# 5. Convert categorical labels to integer labels
# Create a mapping from class names to integers
class_names = list(prob_df_aligned.columns)
class_to_int = {class_name: i for i, class_name in enumerate(class_names)}

print(f"  Class mapping: {class_to_int}")

# Convert true labels to integers
y_true_np = y_true.map(class_to_int).to_numpy()
y_probs_np = prob_df_aligned.to_numpy()

print(f"  Data shape: y_true={y_true_np.shape}, y_probs={y_probs_np.shape}")
print(f"  Unique true labels (integers): {set(y_true_np)}")
print(f"  Class labels (columns): {class_names}")
print(f"  Label distribution: {dict(zip(*np.unique(y_true_np, return_counts=True)))}")

# Check for any unmapped labels (will be NaN)
if pd.isna(y_true_np).any():
    raise ValueError("  Error: Some true labels could not be mapped to class columns")

    ]]></token>
    <xml name="common_test">
        <param name="non_commercial_use" value="True"/>
        <conditional name="training_type">
            <param name="model" value="s_train"/>
            <param name="train_clin" value="train/clin" ftype="csv"/>
            <param name="test_clin" value="test/clin" ftype="csv"/>
            <param name="train_omics_main" value="train/gex" ftype="csv"/>
            <param name="test_omics_main" value="test/gex" ftype="csv"/>
            <param name="assay_main" value="bar"/>
            <repeat name="omics">
                <param name="train_omics" value="train/cnv" ftype="csv"/>
                <param name="test_omics" value="test/cnv" ftype="csv"/>
                <param name="assay" value="foo"/>
            </repeat>
            <conditional name="model_class">
                <param name="model_class" value="DirectPred"/>
            </conditional>
            <param name="target_variables" value="Erlotinib"/>
            <param name="surv_event_var" value="OS_STATUS"/>
            <param name="surv_time_var" value="OS_MONTHS"/>
            <section name="advanced">
                <param name="hpo_iter" value="1"/>
            </section>
        </conditional>
        <yield/>
        <output_collection name="results" type="list">
            <element name="job.embeddings_test">
                <assert_contents>
                    <has_n_lines n="50"/>
                </assert_contents>
            </element>
            <element name="job.embeddings_train">
                <assert_contents>
                    <has_n_lines n="50"/>
                </assert_contents>
            </element>
            <element name="job.feature_importance.GradientShap">
                <assert_contents>
                    <has_text_matching expression="Erlotinib,0,,bar,A2M,"/>
                    <has_text_matching expression="Erlotinib,0,,bar,ABCC4,"/>
                    <has_text_matching expression="GradientShap"/>
                </assert_contents>
            </element>
            <element name="job.feature_importance.IntegratedGradients">
                <assert_contents>
                    <has_text_matching expression="Erlotinib,0,,bar,A2M,"/>
                    <has_text_matching expression="Erlotinib,0,,bar,ABCC4,"/>
                    <has_text_matching expression="IntegratedGradients"/>
                </assert_contents>
            </element>
            <element name="job.feature_logs.bar">
                <assert_contents>
                    <has_n_lines n="25"/>
                </assert_contents>
            </element>
            <element name="job.feature_logs.omics_foo">
                <assert_contents>
                    <has_n_lines n="25"/>
                </assert_contents>
            </element>
            <element name="job.predicted_labels">
                <assert_contents>
                    <has_text_matching expression="source_dataset:A-704,Erlotinib,"/>
                    <has_text_matching expression="target_dataset:KMRC-20,Erlotinib,"/>
                </assert_contents>
            </element>
            <element name="job.stats">
                <assert_contents>
                    <has_text_matching expression="DirectPred,Erlotinib,numerical,mse,"/>
                    <has_text_matching expression="DirectPred,Erlotinib,numerical,r2,"/>
                    <has_text_matching expression="DirectPred,Erlotinib,numerical,pearson_corr,"/>
                </assert_contents>
            </element>
        </output_collection>
    </xml>
    <token name="@COMMON_HELP@">

.. class:: warningmark

**WARNING: This tool is only available for NON-COMMERCIAL use. Permission is only granted for academic, research, and educational purposes. Before using, be sure to review, agree, and comply with the license.**

Flexynesis is a deep-learning based multi-omics bulk sequencing data integration suite with a focus on (pre-)clinical endpoint prediction.
The package includes multiple types of deep learning architectures such as simple fully connected networks, supervised variational autoencoders, graph convolutional networks, multi-triplet networks different options of data layer fusion, and automates feature selection and hyperparameter optimization.

For more information, please check the Documentation_ :

For commercial use, please review the flexynesis license on GitHub and contact the `copyright holders`_ .

-----

    </token>
    <xml name="creator">
        <creator>
            <organization name="European Galaxy Team" url="https://galaxyproject.org/eu/"/>
            <person givenName="Amirhossein" familyName="Naghsh Nilchi" email="[e-mail]"/>
            <yield/>
            <person givenName="Björn" familyName="Grüning" email="[e-mail]"/>
        </creator>
    </xml>
    <xml name="citations">
        <citations>
            <citation type="doi">10.1101/2024.07.16.603606</citation>
        </citations>
    </xml>
</macros>