<tool id="enumerate_charges" name="Enumerate changes" version="0.1">
    <description>calculated with Dimorphite DL and RDKit</description>
      <requirements>
        <requirement type="package" version="2019.09.2">rdkit</requirement>
    </requirements>
    <command detect_errors="exit_code">
<![CDATA[
        python '$__tool_directory__/dimorphite_dl.py'
            --smiles_file '$input'
            --output_file '$output'
            --min_ph $min_ph
            --max_ph $max_ph
]]>
    </command>
    <inputs>
        <param name="input" format="smi" type="data" label="Input molecule data"
            help="In SMILES format"/>
        <param name="min_ph" type="float" label="Minimum pH" value="6.4" min="0" max="14"/>
        <param name="max_ph" type="float" label="Maximum pH" value="8.4" min="0" max="14"/>
    </inputs>
    <outputs>
        <data format="smi" name="output" />
    </outputs>
    <tests>
        <test>
            <param name="input" ftype='smi' value="mols.smi" />
            <param name="min_ph"  value="6.4" />
            <param name="max_ph"  value="8.4" />
            <output name="output" ftype='smi'>
                <!-- there are problems comparing the output to the expected due to whitespace differences
                so we just check that the number of lines is as expected -->
                <assert_contents>
                    <has_n_lines n="7" />
                    <has_text text="NC(CCCC[NH3+])C(=O)[O-]" />
                    <has_text text="NCCCCC(N)C(=O)[O-]" />
                    <has_text text="NCCCCC([NH3+])C(=O)[O-]" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="input" ftype='smi' value="mols.smi" />
            <param name="min_ph"  value="4.4" />
            <param name="max_ph"  value="10.4" />
            <output name="output" ftype='smi'>
                <assert_contents>
                    <has_n_lines n="14" />
                    <has_text text="NC(CCCC[NH3+])C(=O)O" />
                    <has_text text="NCCCCC([NH3+])C(=O)[O-]" />
                    <has_text text="NC(CCCC[NH3+])C(=O)[O-]" />
                    <has_text text="NCCCCC([NH3+])C(=O)O" />
                    <has_text text="NC(CCCC[NH3+])C(=O)O" />
                    <has_text text="NCCCCC(N)C(=O)O" />
                </assert_contents>
            </output>
        </test>
    </tests>
    <help>
<![CDATA[

.. class:: infomark

**What this tool does**

Dimorphite DL provides rules on how to enumerate the different charge forms of molecules using RDKit.
Source code can be found here: https://git.durrantlab.pitt.edu/jdurrant/dimorphite_dl


-----

.. class:: infomark

**Input**

- `SMILES Format`_

.. _SMILES Format: http://en.wikipedia.org/wiki/Simplified_molecular_input_line_entry_specification

-----

.. class:: infomark

 **Output**

SMILES format with the enumerated molecules. If an ID was provided in the input this is retained in the output.


]]>
    </help>
    <citations>
        <citation type="doi">10.1186/s13321-019-0336-9</citation>
        <citation type="bibtex">
            @article{rdkit,
            author = {Greg Landrum and others},
            title = {RDKit: Open-source cheminformatics},
            url ={http://www.rdkit.org}
            }</citation>
    </citations>
</tool>