<tool id="sucos_clustering" name="Cluster ligands using SuCOS" version="0.1.1">
    <description>based on the overlap of 3D features</description>
    <macros>
        <import>sucos_macros.xml</import>
    </macros>
    <expand macro="requirements">
        <requirement type="package" version="1.3.0">scipy</requirement>
    </expand>
    <command detect_errors="exit_code"><![CDATA[
        python '$__tool_directory__/sucos_cluster.py'
            -i '$input'
            -t $threshold
    ]]></command>
    <inputs>
        <param type="data" name="input" format="sdf" label="Ligands to be clustered" help="Input in SDF format." />
        <param name="threshold" type="float" value="0.8" min="0" max="1" label="Clustering threshold"/>
    </inputs>
    <outputs>
        <collection name="clusters" type="list" label="Clustered ligands">
            <discover_datasets pattern="(?P&lt;designation&gt;^cluster\d+)\.sdf$" ext="sdf"/>
        </collection>
    </outputs>
    <tests>
        <test>
            <param name="input" ftype="sdf" value="sucos_cluster.sdf"/>
            <output_collection name="clusters" type="list" count="6">
                <element name="cluster1" ftype="sdf" file="cluster1.sdf" compare="sim_size" />
                <element name="cluster2" ftype="sdf" file="cluster2.sdf" compare="sim_size" />
                <element name="cluster3" ftype="sdf" file="cluster3.sdf" compare="sim_size" />
                <element name="cluster4" ftype="sdf" file="cluster4.sdf" compare="sim_size" />
                <element name="cluster5" ftype="sdf" file="cluster5.sdf" compare="sim_size" />
                <element name="cluster6" ftype="sdf" file="cluster6.sdf" compare="sim_size" />
            </output_collection>
        </test>
    </tests>
    <help><![CDATA[
.. class:: infomark

**What it does?**

This tool clusters molecules based on the overlap of 3D features as determined by SuCOS.

Clustering uses a clustering threshold that can be set by the user. The default threshold is 0.8.

The original SuCOS code is on GitHub_ under a MIT license. The SuCOS work is described here_.

.. _GitHub: https://github.com/susanhleung/SuCOS
.. _here: https://chemrxiv.org/articles/SuCOS_is_Better_than_RMSD_for_Evaluating_Fragment_Elaboration_and_Docking_Poses/8100203

.. class:: infomark

**Input**

Molecules such as an SD file dataset in the history.

.. class:: infomark

**Output**

A series of SD files, one for each cluster containing the molecules in that cluster.

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