<tool id="ecology_link_between_var" name="Variables exploration" version="@VERSION@" profile="20.01">
    <description>Shows interaction, correlation, colinearity, produces a PCA and computes VIF for biodiversity abundance data</description>
    <macros>
        <import>macro.xml</import>
    </macros>
    <expand macro="Explo_requirements">
        <requirement type="package" version="4.1">r-base</requirement>
        <requirement type="package" version="1.1.1">r-cowplot</requirement>
        <requirement type="package" version="2.1.2">r-ggally</requirement>
        <requirement type="package" version="3.0_11">r-car</requirement>
        <requirement type="package" version="1.0.7">r-dplyr</requirement>
        <requirement type="package" version="0.1.3">r-ggcorrplot</requirement>
        <requirement type="package" version="2.4">r-factominer</requirement>
        <requirement type="package" version="1.0.7">r-factoextra</requirement>
    </expand>
    <command detect_errors="exit_code"><![CDATA[
        Rscript
            '$__tool_directory__/graph_link_var.r'
            '$input'
            '$colnames'
            #if $method.type == 'collinearity':
              'TRUE'
              'FALSE'
              'FALSE'
              'FALSE'
              'FALSE'
              '$method.species'
              '$method.columns'
              ''
              ''
              ''
            #elif $method.type == 'vif':
              'FALSE'
              'TRUE'
              'FALSE'
              'FALSE'
              'FALSE'
              ''
              '$method.columns'
              ''
              ''
              ''
            #elif $method.type == 'pca':
              'FALSE'
              'FALSE'
              'TRUE'
              'FALSE'
              'FALSE'
              ''
              '$method.columns'
              ''
              ''
              ''
            #elif $method.type == 'interr':
              'FALSE'
              'FALSE'
              'FALSE'
              'TRUE'
              'FALSE'
              '$method.species'
              ''
              '$method.variable'
              '$method.variable2'
              '$method.variable3'
            #else:
              'FALSE'
              'FALSE'
              'FALSE'
              'FALSE'
              'TRUE'
              ''
              ''
              '$method.variable'
              ''
              ''
            #end if
        ]]>
        </command>
    <inputs>
        <expand macro="explo_input"/>
        <conditional name="method">
            <param name="type" type="select" label="Variables links exploration">
                <option value="collinearity">Collinearity between selected numerical variables for each species</option>
                <option value="vif">Variance inflation factor (vif) on selected numerical variables</option>
                <option value="pca">Principal component analysis (pca) on selected numerical variables</option>
                <option value="interr">Interactions between 2 selected numerical variables</option>
                <option value="autocorr">Autocorrelation of one selected numerical variable</option>
            </param>
            <when value="collinearity">
                <param name="species" type="data_column" data_ref="input" numerical="false" label="Select column containing species" use_header_names="true"/>
                 <param name="columns" type="data_column" data_ref="input" numerical="true" multiple="true" label="Select columns containing numerical values" help="Select at least two columns" use_header_names="true"/>
            </when>
            <when value="vif">
                <param name="columns" type="data_column" data_ref="input" numerical="true" multiple="true" label="Select columns containing numerical values" use_header_names="true"/>
            </when>
            <when value="pca">
                <param name="columns" type="data_column" data_ref="input" numerical="true" multiple="true" label="Select columns containing numerical values" use_header_names="true"/>
            </when>
            <when value="interr">
                <param name="variable" type="data_column" data_ref="input" numerical="true" label="Select column containing numerical values for x-axis" use_header_names="true"/>
                <param name="variable2" type="data_column" data_ref="input" numerical="true" label="Select column containing numerical values for y-axis" use_header_names="true"/>
                <param name="species" type="data_column" data_ref="input" numerical="false" label="Select column containing species" help="This parameter allows you to divide your scatterplot according to species" use_header_names="true"/>
                <param name="variable3" type="data_column" data_ref="input" label="Select column" help="This parameter allows you to divide your scatterplot once more" use_header_names="true"/>
            </when>
            <when value="autocorr">
                <param name="variable" type="data_column" data_ref="input" numerical="true" label="Select column containing numerical values" use_header_names="true"/>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_coli" from_work_dir="Data.txt" format="txt" label="Collinearity analysis - Missing species">
            <expand macro="explo_filter_colli"/>
        </data>
        <data name="output_acp" from_work_dir="valeurs.txt" format="txt" label="PCA (Principal Component Analysis) - Eigen values">
            <expand macro="explo_filter_pca"/>
        </data>
        <data name="output_vif" from_work_dir="vif.tabular" format="tabular" label="Your VIF tabular">
            <expand macro="explo_filter_vif"/>
        </data>
        <data name="output_corr" from_work_dir="corr.tabular" format="tabular" label="Correlation matrix">
            <expand macro="explo_filter_vif"/>
        </data>
        <data name="output_interr" from_work_dir="Species.txt" format="txt" label="Interactions analysis - Species in data">
            <expand macro="explo_filter_interr"/>
        </data>
        <data name="output_autocorr" from_work_dir="acf.txt" format="txt" label="Autocorrelation analysis - ACF table">
            <expand macro="explo_filter_autocorr"/>
        </data>
        <collection type="list" name="plots">
            <discover_datasets pattern="(?P&lt;designation&gt;.+)\.png" visible="false" format="png"/>
            <filter>method['type'] != 'vif'</filter>
        </collection>
    </outputs>
    <tests>
        <test>
            <param name="input" value="Reel_life_survey_fish_modif2.tabular"/>
            <param name="colnames" value="true"/>
            <conditional name="method">
                <param name="type" value="collinearity"/>
                <param name="species" value="15"/>
                <param name="columns" value="12,17,18"/>
            </conditional>
            <output name="output_coli" value="Missing_species.txt"/>
            <output_collection name="plots" type="list" count="3">
                <element name="collinarity_of_Blenniidae" ftype="png">
                    <assert_contents>
                        <has_text text="PNG"/>
                    </assert_contents>
                </element>
                <element name="collinarity_of_Gobiidae" ftype="png">
                    <assert_contents>
                        <has_text text="PNG"/>
                    </assert_contents>
                </element>
                <element name="collinarity_of_Tripterygiidae" ftype="png">
                    <assert_contents>
                        <has_text text="PNG"/>
                    </assert_contents>
                </element>
            </output_collection>
        </test>
    </tests>
    <expand macro="topic"/>
    <help><![CDATA[
=================================
Determine links between variables
=================================

- Show the collinearity among the covariates
- Plot a Pincipal Component Analysis (PCA)
- Compute the Variance Inflation Factor (VIF)
- Show if there is auto-correlation
- Show the interactions between variables

**Collinearity between selected numerical variables for each species**

This tool shows if multiple numerical variables shows colinearity or not between one another.

Input description :

A tabular file with observation data. Must at least contain three columns, species and multiple numerical variable.

+-------------+------------+---------------+
|   number1   |   number2  | species.code  |
+=============+============+===============+
|      2      |      4     |   speciesID   |
+-------------+------------+---------------+
|     ...     |     ...    |      ...      |
+-------------+------------+---------------+

Output description :

A png file with one plot containing multiple correlation plots and the correlation values between each variables.

Warning : When there are more than 3 species in the data this tool shows one plot for each species.


**Variance Inflation Factor (VIF) on selected numerical variables**

This tool calculates the correlation matrix and the Variance Inflation Factor between each pair of the selected numerical variables.

Input description:

A tabular file with observation data. Must at least contain two columns of numerical variables.


Output description :

Two tabulars :

- One with VIF values for each pair, it measures how much the behavior (variance) of an independent variable is influenced, or inflated, by its interaction/correlation with the other independent variables. A large VIF on an independent variable indicates a highly collinear relationship to the other variable that should be considered or adjusted for in the structure of the model and selection of independent variable.

- One containing the correlation matrix.


**Principal Component Analysis (PCA) on selected numerical variables**

This tool computes a Principal Component Analysis.

Input description:

A tabular file with observation data with numerical variables.

Output description:

Two png files with plots. The first one is showing the PCA plot :

- The positively correlated variables are grouped together.

- The negatively ones are on opposite sides of the plot's origin.

- the distance between the variables and the origin calculates the quality of the representation of the variables. The variables far from the origin are well represented by the PCA. 

The quality of the representation is also calculated and represented with the cos2 determined with colors : 

- A high cos2 indicates a good representation of the variable. In this case, the variable is near the circumference of the correlation circle.

- A low cos2 indicates that the variable is not perfectly represented. In this case, the variable is near the center of the circle.

The second plot is about the quality of the PCA, it represents the correlation between dimensions of the PCA and the selected variables.

A text file containing eigen values of the PCA.


**Interactions between two selected numerical variables**

This tool represents the interactions between variables through multiple scatterplots.

Input description:

A tabular file with observation data. Must at least contain four columns two numerical variables, any other variables and species.

+----------+-----------+--------------+------------+
|  number1 | variable  | species.code |   number2  |
+==========+===========+==============+============+
|    2     |    var    |   speciesID  |      4     |
+----------+-----------+--------------+------------+
|  ...     |    ...    |      ...     |     ...    |
+----------+-----------+--------------+------------+

Output description:

PNG files (one per species) with plots showing the interactions between the two numerical variables for each separation factor.

A text file with a recap of the species column used for the analysis.


**Autocorrelation of one selected numerical variable**

This tool computes the ACF (Auto-Correlation Function) and represents the autocorrelation of a numerical variable.

Input description:

A tabular file with observation data. Must at least contain one column with a numerical variable.


Output description:

A png file with one plot showing the autocorrelation for a variable. If the bars of the histogram are strictly confined between the dashed lines (representing 95% confidence interval without white noise), there is auto-correlation.

A text file containing the ACF values.

    ]]></help>
    <expand macro="explo_bibref"/>
</tool>