view keras_train_and_eval.xml @ 14:818f9b69d8a0 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit 80417bf0158a9b596e485dd66408f738f405145a

<tool id="keras_train_and_eval" name="Deep learning training and evaluation" version="@VERSION@" profile="@PROFILE@">
    <description>conduct deep training and evaluation either implicitly or explicitly</description>
    <macros>
        <import>main_macros.xml</import>
        <import>keras_macros.xml</import>
    </macros>
    <expand macro="python_requirements" />
    <expand macro="macro_stdio" />
    <version_command>echo "@VERSION@"</version_command>
    <command>
        <![CDATA[        
        export HDF5_USE_FILE_LOCKING='FALSE';
        #if $input_options.selected_input == 'refseq_and_interval'
        bgzip -c '$input_options.target_file' > '${target_file.element_identifier}.gz' &&
        tabix -p bed '${target_file.element_identifier}.gz' &&
        cp '$input_options.ref_genome_file' '${ref_genome_file.element_identifier}' &&
        #end if
        python '$__tool_directory__/keras_train_and_eval.py'
            --inputs '$inputs'
            --estimator '$experiment_schemes.infile_estimator'
            #if $input_options.selected_input == 'seq_fasta'
            --fasta_path '$input_options.fasta_path'
            #elif $input_options.selected_input == 'refseq_and_interval'
            --ref_seq "`pwd`/${ref_genome_file.element_identifier}"
            --interval '$input_options.interval_file'
            --targets "`pwd`/${target_file.element_identifier}.gz"
            #else
            --infile1 '$input_options.infile1'
            #end if
            --infile2 '$input_options.infile2'
            --outfile_result '$outfile_result'
            #if $save and 'save_csvlogger' in str($save)
            --outfile_history '$outfile_history'
            #end if
            #if $save and 'save_estimator' in str($save)
            --outfile_object '$outfile_object'
            #end if
            #if $save and 'save_prediction' in str($save)
            --outfile_y_true '$outfile_y_true'
            --outfile_y_preds '$outfile_y_preds'
            #end if
            #if $experiment_schemes.test_split.split_algos.shuffle == 'group'
            --groups '$experiment_schemes.test_split.split_algos.groups_selector.infile_g'
            #end if
        ]]>
    </command>
    <configfiles>
        <inputs name="inputs" />
    </configfiles>
    <inputs>
        <conditional name="experiment_schemes">
            <param name="selected_exp_scheme" type="select" label="Select a scheme">
                <option value="train_val" selected="true">Train and Validate</option>
                <option value="train_val_test">Train, Validate and and Evaluate</option>
            </param>
            <when value="train_val">
                <expand macro="estimator_and_hyperparameter" />
                <section name="test_split" title="Validation holdout" expanded="false">
                    <expand macro="train_test_split_params">
                        <expand macro="cv_groups" />
                    </expand>
                </section>
                <section name="metrics" title="Metrics for evaluation" expanded="false">
                    <expand macro="scoring_selection" help="" />
                </section>
            </when>
            <when value="train_val_test">
                <expand macro="estimator_and_hyperparameter" />
                <section name="test_split" title="Test holdout" expanded="false">
                    <expand macro="train_test_split_params">
                        <expand macro="cv_groups" />
                    </expand>
                </section>
                <section name="val_split" title="Validation holdout (recommend using the same splitting method as for test holdout)" expanded="false">
                    <expand macro="train_test_split_params" />
                </section>
                <section name="metrics" title="Metrics from scikit-learn" expanded="false">
                    <expand macro="scoring_selection" help="" />
                </section>
            </when>
        </conditional>
        <expand macro="sl_mixed_input_plus_sequence" />
        <param name="save" type="select" multiple='true' display="checkboxes" label="Save the fitted model" optional="true" help="Evaluation scores will be output by default.">
            <option value="save_estimator" selected="true">Fitted estimator</option>
            <option value="save_prediction">True labels and prediction results from evaluation for downstream analysis</option>
            <option value="save_csvlogger">Display CSVLogger if selected as a callback in the Keras model builder tool</option>
        </param>
    </inputs>
    <outputs>
        <data format="tabular" name="outfile_result" />
         <data format="tabular" name="outfile_history" label="Deep learning training history log on ${on_string}">
            <filter>str(save) and 'save_csvlogger' in str(save)</filter>
        </data>
        <data format="h5mlm" name="outfile_object" label="Fitted estimator or estimator skeleton on ${on_string}">
            <filter>str(save) and 'save_estimator' in str(save)</filter>
        </data>
        <data format="tabular" name="outfile_y_true" label="True labels/target values on ${on_string}">
            <filter>str(save) and 'save_prediction' in str(save)</filter>
        </data>
        <data format="tabular" name="outfile_y_preds" label="All predictions on ${on_string}">
            <filter>str(save) and 'save_prediction' in str(save)</filter>
        </data>
    </outputs>
    <tests>
        <test>
            <conditional name="experiment_schemes">
                <param name="selected_exp_scheme" value="train_val_test" />
                <param name="infile_estimator" value="keras_model04" ftype="h5mlm" />
                <section name="hyperparams_swapping">
                    <repeat name="param_set">
                        <param name="sp_value" value="999" />
                        <param name="sp_name" value="layers_1_Dense__config__kernel_initializer__config__seed" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="999" />
                        <param name="sp_name" value="layers_3_Dense__config__kernel_initializer__config__seed" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="0.1" />
                        <param name="sp_name" value="learning_rate" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="'adamax'" />
                        <param name="sp_name" value="optimizer" />
                    </repeat>
                </section>
                <section name="test_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="simple" />
                        <param name="test_size" value="0.2" />
                        <param name="random_state" value="123" />
                    </conditional>
                </section>
                <section name="val_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="simple" />
                        <param name="test_size" value="0.2" />
                        <param name="random_state" value="456" />
                    </conditional>
                </section>
                <section name="metrics">
                    <conditional name="scoring">
                        <param name="primary_scoring" value="r2" />
                        <param name="secondary_scoring" value="neg_mean_absolute_error" />
                    </conditional>
                </section>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular" />
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns" />
            <param name="infile2" value="regression_y.tabular" ftype="tabular" />
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns" />
            <param name="save" value="save_estimator" />
            <output name="outfile_result">
                <assert_contents>
                    <has_n_columns n="4" />
                    <has_text text="0.794" />
                    <has_text text="-4.62" />
                </assert_contents>
            </output>
            <output name="outfile_object" file="train_test_eval_model01" compare="sim_size" delta="5" />
        </test>
        <test>
            <conditional name="experiment_schemes">
                <param name="selected_exp_scheme" value="train_val_test" />
                <param name="infile_estimator" value="keras_model04" ftype="h5mlm" />
                <section name="hyperparams_swapping">
                    <repeat name="param_set">
                        <param name="sp_value" value="999" />
                        <param name="sp_name" value="layers_1_Dense__config__kernel_initializer__config__seed" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="999" />
                        <param name="sp_name" value="layers_3_Dense__config__kernel_initializer__config__seed" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="0.1" />
                        <param name="sp_name" value="learning_rate" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value="'adamax'" />
                        <param name="sp_name" value="optimizer" />
                    </repeat>
                </section>
                <section name="test_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="group" />
                        <param name="group_names" value="test" />
                        <section name="groups_selector">
                            <param name="infile_g" value="regression_groups.tabular" ftype="tabular" />
                            <param name="header_g" value="true" />
                            <conditional name="column_selector_options_g">
                                <param name="selected_column_selector_option_g" value="by_index_number" />
                                <param name="col_g" value="1" />
                            </conditional>
                        </section>
                    </conditional>
                </section>
                <section name="val_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="group" />
                        <param name="group_names" value="validation" />
                    </conditional>
                </section>
                <section name="metrics">
                    <conditional name="scoring">
                        <param name="primary_scoring" value="r2" />
                        <param name="secondary_scoring" value="neg_mean_absolute_error" />
                    </conditional>
                </section>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular" />
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns" />
            <param name="infile2" value="regression_y.tabular" ftype="tabular" />
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns" />
            <param name="save" value="save_estimator,save_prediction" />
            <output name="outfile_result" >
                <assert_contents>
                    <has_n_columns n="4" />
                    <has_text text="0.779" />
                    <has_text text="-4.5" />
                </assert_contents>
            </output>
            <output name="outfile_object" file="train_test_eval_model02" compare="sim_size" delta="5" />
            <output name="outfile_y_true" file="keras_train_eval_y_true02.tabular" ftype="tabular" />
        </test>
        <test>
            <conditional name="experiment_schemes">
                <param name="selected_exp_scheme" value="train_val" />
                <param name="infile_estimator" value="pipeline10" ftype="h5mlm" />
                <section name="hyperparams_swapping">
                    <repeat name="param_set">
                        <param name="sp_value" value="10" />
                        <param name="sp_name" value="adaboostregressor__random_state" />
                    </repeat>
                    <repeat name="param_set">
                        <param name="sp_value" value=": sklearn_tree.ExtraTreeRegressor(random_state=0)" />
                        <param name="sp_name" value="adaboostregressor__base_estimator" />
                    </repeat>
                </section>
                <section name="test_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="simple" />
                        <param name="test_size" value="0.2" />
                        <param name="random_state" value="123" />
                    </conditional>
                </section>
                <section name="val_split">
                    <conditional name="split_algos">
                        <param name="shuffle" value="simple" />
                        <param name="test_size" value="0.2" />
                        <param name="random_state" value="456" />
                    </conditional>
                </section>
                <section name="metrics">
                    <conditional name="scoring">
                        <param name="primary_scoring" value="r2" />
                        <param name="secondary_scoring" value="neg_mean_absolute_error" />
                    </conditional>
                </section>
            </conditional>
            <param name="infile1" value="regression_X.tabular" ftype="tabular" />
            <param name="header1" value="true" />
            <param name="selected_column_selector_option" value="all_columns" />
            <param name="infile2" value="regression_y.tabular" ftype="tabular" />
            <param name="header2" value="true" />
            <param name="selected_column_selector_option2" value="all_columns" />
            <param name="save" value="" />
            <output name="outfile_result" file="train_test_eval03.tabular" />
        </test>
    </tests>
    <help>
        <![CDATA[
**What it does**

Given a pre-built keras deep learning model and labeled training dataset, this tool works in two modes.

- Train and Validate: the intput dataset is split into training and validation portions. The model is fitted on the training portion, in the meantime performances are evaluated on the validation portion multiple times while the training is progressing. Finally, a fitted model and its validation performance scores are outputted.


- Train, Validate and and Evaluate: the input dataset is split into three portions, training, validation and testing. The same `Train and Validate` described above is performed on the training and validation portions. The testing portion is used exclusively for testing (evaluation). As a result, a fitted model and test performance scores are outputted.

In both modes, besides the performance scores, the true labels and predicted values are outputted, which could be used in generating plots in other tools, machine learning visualization extensions, for example.

Note that since all training and model parameters are accessible and changeable in the `Hyperparameter Swapping` section, the training and evaluation processes are flexible and transparent.

For metrics, there are two sets of metrics for deep learning training and evaluation, one from the keras model builder and the other from scikit-learn. Keras metrics, if selected, are always evaluated, while the sklearn metrics could be ignored when `default` is the selection. Please be aware that not every sklearn metric works with deep learning model at current moment. Feel free to file a ticket if an issue is found and contibuting with PRs is always welcomed.

**Input**

- tabular
- sparse
- `sequences in a fasta file` to work with DNA, RNA and proteins with corresponding fasta data generator
- `reference genome and intervals` exclusively work with `GenomicIntervalBatchGenerator`.

**Output**

- performance scores from evaluation
- fitted estimator
- true labels or values and predicted values from the evaluation

        ]]>
    </help>
    <expand macro="sklearn_citation">
        <expand macro="keras_citation" />
    </expand>
</tool>