<tool id="dpocket" name="dpocket" version="@TOOL_VERSION@+galaxy@GALAXY_VERSION@">
    <description>to calculate descriptors for protein pockets</description>
    <macros>
        <import>macros.xml</import>
        <token name="@GALAXY_VERSION@">0</token>
    </macros>
    <expand macro="requirements" />
    <command detect_errors="exit_code"><![CDATA[
        ## create config file
        #for i in $input:
            ln -s '$i' $i.name.replace(' ', '_') &&
            echo $i.name.replace(' ', '_') >> pdbs.txt &&
        #end for
        paste pdbs.txt '$ligs' > dp_conf.txt &&

        ## run dpocket
        dpocket
            -f dp_conf.txt

            ## pocket definition params common with fpocket
            #if $inp.pocket_type == 'channel':
                --min_alpha_size 2.8 --max_alpha_size 5.5 --min_spheres_per_pocket 30
            #elif $inp.pocket_type == 'external':
                --min_alpha_size 3.5 --max_alpha_size 10 --min_spheres_per_pocket 30
            #elif $inp.pocket_type == 'custom':
                --min_alpha_size '$inp.min'
                --max_alpha_size '$inp.max'
                --min_spheres_per_pocket '$inp.i'
                --clustering_distance '$inp.D'
                --clustering_method '$inp.C'
                ##--clustering_measure $inp.e
            #end if

            ## specific params for explicit pocket definition
            #if $explicit_output.output == "true":
                $explicit_output.interface_definition
                -d $explicit_output.dist
            #end if

            &&
        ## reformat output files to match Galaxy tabular
        sed -e "s/\s\+/\t/g" dpout_fpocketp.txt > $fpocketp &&
        sed -e "s/\s\+/\t/g" dpout_fpocketnp.txt > $fpocketnp
        #if $explicit_output.output:
            &&
            sed -e "s/\s\+/\t/g" dpout_explicitp.txt > $explicitp
        #end if

    ]]></command>
    <inputs>
        <param name="input" type="data" format="pdb" multiple="true" label="Input file" help="Protein structure file(s) (PDB) to search."/>
        <param name="ligs" type="data" format="txt" label="Ligand codes" help="List of ligand codes (as provided in the PDB files), one per line. The number of codes must match the number of input PDB files."/>
        <expand macro="inputs" />
        <!-- <param name="outputs" type="select" display="checkboxes" multiple="true" label="Output files">
            <option value="fpocketp" selected="true">Pockets which match the ligand position</option>
            <option value="fpocketnp">Pockets which do NOT match the ligand position</option>
            <option value="explicitp">Define pocket explicitly as all vertices/atoms within a given distance from the ligand, regardless of what fpocket found. </option>
        </param> -->
        <conditional name="explicit_output">
            <param name="output" type="select" label="Calculate explicit pocket?" help="Define pocket explicitly as all vertices/atoms within a given distance from the ligand, regardless of what fpocket found.">
                <option value="true" selected="true">Yes</option>
                <option value="false">No</option>
            </param>
            <when value="false" />
            <when value="true">
                <param name="interface_definition" type="select" label="Definition of the explicit pocket" help="Protein-ligand explicit interface definition">
                    <option value="-e" selected="true">All atoms in contact with alpha spheres within a distance threshold from any ligand atom</option>
                    <option value="-E">All atoms within a distance threshold from any ligand atom</option>
                </param>
                <param name="dist" type="float" value="4.0" min="0" label="Distance threshold for explicit pocket definition (angstroms)" />
            </when>

        </conditional>
    </inputs>

    <outputs>
        <data name="fpocketp" format="tabular" label="Pockets matching the submitted ligands"/>
        <data name="fpocketnp" format="tabular" label="Pockets NOT matching the submitted ligands"/>
        <data name="explicitp" format="tabular" label="Pockets defined explicitly based on the submitted ligands">
            <filter>explicit_output['output'] == "true"</filter>
        </data>
    </outputs>

    <tests>
        <test expect_num_outputs="3">
            <param name="input" ftype="pdb" value="2brc.pdb"/>
            <param name="ligs" ftype="txt" value="ligs.txt"/>
            <param name='pocket_type' value='custom' />
            <param name="min" value="4.0"/>
            <param name="max" value="7.0"/>
            <param name="i" value="20" />
            <param name="D" value="2.0"/>
            <param name="C" value="s" />
            <param name="interface_definition" value="-E" />
            <param name="dist" value="5" />
            <param name='explicit_output' value='true' />
            <output name="fpocketp" ftype="tabular">
                <assert_contents>
                    <has_n_columns n="57" />
                    <has_n_lines n="2" />
                    <has_text_matching expression="2brc.pdb\s+CT5\s+31.15\s+1\s+1.07\s+0.77\s+1.00\s+1\s+1.54"/>
                    <has_text_matching expression="6.50\s+0.96\s+0.44\s+14.39\s+23.64\s+12.20\s+42.27\s+15.29\s+2\s+1\s+0\s+2"/>
                </assert_contents>
            </output>
            <output name="fpocketnp" ftype="tabular">
                <assert_contents>
                    <has_n_columns n="57" />
                    <has_n_lines n="2" />
                    <has_text_matching expression="2brc.pdb\s+CT5\s+0.00\s+0\s+17.62\s+0.00\s+0.00\s+0\s+-1.25"/>
                    <has_text_matching expression="4.36\s+0.69\s+0.00\s+4.37\s+31.43\s+42.10\s+33.81\s+17.20\s+5\s+0\s+1\s+1"/>
                </assert_contents>
            </output>
            <output name="explicitp" ftype="tabular">
                <assert_contents>
                    <has_n_columns n="57" />
                    <has_n_lines n="2" />
                    <has_text_matching expression="2brc.pdb\s+CT5\s+100.00\s+1\s+0.00\s+1.00\s+1.00\s+1\s+2.00"/>
                    <has_text_matching expression="11.97\s+0.00\s+0.00\s+107.35\s+56.94\s+61.01\s+57.97\s+34.40\s+3\s+2\s+0\s+2"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="3">
            <param name="input" ftype="pdb" value="2brc.pdb,1L83.pdb"/>
            <param name="ligs" ftype="txt" value="ligs2.txt"/>
            <param name='pocket_type' value='small_mol' />
            <param name='explicit_output' value='false' />
            <output name="fpocketp" ftype="tabular">
                <assert_contents>
                    <has_n_columns n="57" />
                    <has_n_lines n="3" />
                    <has_text_matching expression="2brc.pdb\s+CT5\s+38.60\s+1\s+0.54\s+0.46\s+0.80\s+0\s+0.67"/>
                    <has_text_matching expression="1L83.pdb\s+BNZ\s+100.00\s+1\s+0.83\s+1.00\s+0.85\s+1\s+1.81"/>
                </assert_contents>
            </output>
            <output name="fpocketnp" ftype="tabular">
                <assert_contents>
                    <has_n_columns n="57" />
                    <has_n_lines n="19" />
                    <has_text_matching expression="2brc.pdb\s+CT5\s+0.00\s+0\s+17.15\s+0.00\s+0.00\s+0\s+-1.25"/>
                    <has_text_matching expression="29.17\s+4.03\s+0.65\s+11.05\s+0.65\s+0.14\s+13.17\s+29.13\s+19.24"/>
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[

Calculate descriptors (i.e. featurize) 'pockets' in a protein structure, using the
dpocket module of the fpocket software.

To use, upload one or more protein structures in PDB format and select the type of pocket to 
detect. 'Custom options' can also be selected - this exposes all internal fpocket 
parameters. Using this option requires some knowledge of the fpocket prediction 
algorithm. Please consult the cited publications for more details.

In addition, a list of ligand identifiers should be provided as a text file, one per line and
one for each of the protein structures.


-----

.. class:: infomark

**Input**

- One or more protein structures in PDB format. If using multiple structures, submitting them as a dataset collection is recommended to ensure ordering is preserved.
- Text file with ligand identifiers, matching the ligand code in the PDB files. One identifier should be listed per line and one per PDB file. 

-----

.. class:: infomark

**Output**

- Tabular file with descriptors of pockets matching the positions of the input ligands.
- Tabular file with descriptors of pockets which do NOT match the positions of the input ligands.
- (Optional) Tabular file with descriptors of pockets which are explicitly defined by the positions of the input ligands.

Each tabular file contains 57 columns (55 descriptors, plus protein and ligand names), a header line, and
one row per pocket found.

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