annotate retropath2_wrapper.xml @ 17:60a0330a2ba9 draft

planemo upload for repository https://github.com/brsynth/retropath2_wrapper commit 1dfdc8e67615cd0174729991e30181410cf0cd11

<tool id="retropath2" name="RetroPath2.0" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@"
    profile="21.09" license="MIT">
    <description>Build a reaction network from a set of source compounds to a set of sink compounds</description>
    <macros>
        <token name="@VERSION_SUFFIX@">0</token>
        <token name="@TOOL_VERSION@">3.6.0</token>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">retropath2_wrapper</requirement>
    </requirements>
    <stdio>
        <!-- Fatal -->
        <regex match="It seems that the target product is already in the chassis" source="both"
            level="fatal"
            description="It seems that the target product is already in the chassis (Exit code: 1)" />
        <regex match="The following error occured" source="both" level="fatal"
            description="The following error occured (Exit code: 2)" />
        <regex match="is not a valid InChI notation" source="both" level="fatal"
            description="is not a valid InChI notation (Exit code: 3)" />
        <regex match="The sink file is malformed" source="both" level="fatal"
            description="The sink file is malformed (Exit code: 4)" />
        <!-- Warning -->
        <regex match="Source has been found in the sink" source="both" level="warning"
            description="Source has been found in the sink (Exit code: 10)" />
        <regex match="No solution has been found" source="both" level="warning"
            description="No solution has been found (Exit code: 11)" />
    </stdio>
    <command detect_errors="exit_code"><![CDATA[
        #if $sink.emptysink:
            echo '"Name","InChI"' > empty_sink.csv &&
            #set $sinkfile = "empty_sink.csv"
        #end if
        python -m retropath2_wrapper
            '$sinkfile'
            '$rulesfile'
            out
            --source_inchi '$source_inchi'
            --source_name '$source_name'
            --rp2_version '$adv.version'
            --max_steps '$max_steps'
            --topx '$adv.topx'
            --dmin '$adv.dmin'
            --dmax '$adv.dmax'
            --mwmax_source '$adv.mwmax_source'
            --kinstall "\${TMPDIR:-.}"
            --std_hydrogen '$adv.std_hydrogen'
            #if not $sink.error_codes:
                --quiet
            #end if
            --msc_timeout '$adv.timeout' &&
        cp out/*_scope.csv "$Reaction_Network" 2>/dev/null || cp out/results.csv "$Reaction_Network"
    ]]></command>
    <inputs>
        <param name="rulesfile" type="data" format="csv,tar" label="Rules File"
            help="File containing reaction rules which reflect the enzymatic potential of the chassis organism." />
        <conditional name="sink">
            <param name="emptysink" type="boolean" checked="false" label="Empty sink" />
            <when value="false">
                <param name="sinkfile" type="data" format="csv" label="Sink File"
                    help="Sink file which comprises all compounds that are considered as granted in your system." />
                <param name="error_codes" type="boolean" checked="true" label="Return error codes" />
            </when>
            <when value="true">
                <param name="error_codes" type="boolean" checked="false" label="Return error codes" />
            </when>
        </conditional>
        <param name="source_inchi" type="text" label="Source InChI" optional="false"
            help="InChI of compound to produce.">
            <validator type="empty_field" message="You must provide the InChI string" />
        </param>
        <param name="max_steps" type="integer" value="3" min="1" max="10"
            label="Maximal Pathway length"
            help="The maximal number of metabolic steps (i.e. reactions) for exploring around the target." />
        <param name="source_name" type="text" value="target" optional="true" label="Source name"
            help="Name of compound to produce.">
            <validator type="empty_field" message="Source name is required." />
        </param>
        <section name="adv" title="Advanced Options" expanded="false">
            <param name="version" type="select" label="Workflow version"
                help="Retropath2.0 Workflow versions, formatted as 'r{year}{month}{day}' (the latest version 'r20220104' is used by default).">
                <option value="v9">v9</option>
                <option value="r20210127">r20210127</option>
                <option value="r20220104">r20220104</option>
                <option value="r20220224">r20220224</option>
                <option value="r20250728" selected="true">r20250728</option>
            </param>
            <param name="topx" type="integer" value="100" min="1" max="1000" label="TopX"
                help="The maximal number of intermediate compounds to use as new starting points for a next iteration of the retrosynthesis algorithm." />
            <param name="dmin" type="integer" value="0" min="0" max="1000"
                label="Minimum rule diameter"
                help="Minimum rule diameter of the sphere including the atoms around the reacting center. The higher is the diameter, the more specific are the rules." />
            <param name="dmax" type="integer" value="1000" min="0" max="1000"
                label="Maximum rule diameter"
                help="Maximum rule diameter of the sphere including the atoms around the reacting center. The higher is the diameter, the more specific are the rules." />
            <param name="mwmax_source" type="integer" value="1000" min="0" max="2000"
                label="Molecular weight of source (Da)"
                help="The molecular weight cutoff (in Da), above which initial source (ie target) and intermediate compounds will be filtered out." />
            <param name="timeout" type="integer" value="60" min="30" max="600" label="Timeout (min)"
                help="Maximal time of RetroPath2.0 execution (60 minutes by default)." />
            <param name="std_hydrogen" type="select" label="Hygrogens in chemical rules"
                help="How hydrogens are represented in chemical rules, auto mode will try to guess from the chemical rules">
                <option value="auto" selected="true">auto</option>
                <option value="implicit">Aromatized (no Hs added)</option>
                <option value="explicit">H added + Aromatized</option>
            </param>
        </section>
    </inputs>
    <outputs>
        <data name="Reaction_Network" format="csv" label="${tool.name}">
            <actions>
                <action name="column_names" type="metadata"
                    default="Initial source,Transformation ID,Reaction SMILES,Substrate SMILES,Substrate InChI,Product SMILES,Product InChI,In Sink,Sink name,Diameter,Rule ID,EC number,Score,Starting Source SMILES,Iteration" />
            </actions>
        </data>
    </outputs>
    <tests>
        <test>
            <!-- test 1: check if identical outputs are produced with default parameters  -->
            <param name="rulesfile" value="rules.csv" />
            <conditional name="sink">
                <param name="emptysink" value="false" />
                <param name="sinkfile" value="sink.csv" />
            </conditional>
            <param name="source_inchi"
                value="InChI=1S/C40H56/c1-33(2)19-13-23-37(7)27-17-31-39(9)29-15-25-35(5)21-11-12-22-36(6)26-16-30-40(10)32-18-28-38(8)24-14-20-34(3)4/h11-12,15-22,25-32H,13-14,23-24H2,1-10H3/b12-11+,25-15+,26-16+,31-17+,32-18+,35-21+,36-22+,37-27+,38-28+,39-29+,40-30+" />
            <output name="Reaction_Network" file="results_retropath2.csv" ftype="csv" compare="diff" />
        </test>
        <test>
            <!-- test 2: check empty sink file -->
            <param name="rulesfile" value="rules.csv" />
            <conditional name="sink">
                <param name="emptysink" value="true" />
            </conditional>
            <param name="source_inchi"
                value="InChI=1S/C40H56/c1-33(2)19-13-23-37(7)27-17-31-39(9)29-15-25-35(5)21-11-12-22-36(6)26-16-30-40(10)32-18-28-38(8)24-14-20-34(3)4/h11-12,15-22,25-32H,13-14,23-24H2,1-10H3/b12-11+,25-15+,26-16+,31-17+,32-18+,35-21+,36-22+,37-27+,38-28+,39-29+,40-30+" />
            <output name="Reaction_Network" file="results_emptysink.csv" ftype="csv" compare="diff" />
        </test>
    </tests>
    <help><![CDATA[
Retropath2.0 wrapper
====================

Perform retrosynthesis search of possible metabolic routes between a source molecule and a collection of sink molecules. This tool is an implementation of the `KNIME retropath2.0 workflow <https://www.myexperiment.org/workflows/4987.html>`_. It takes for input the minimal (dmin) and maximal (dmax) diameter for the reaction rules and the maximal path length (maxSteps). The tool expects the following files: rules.csv, sink.csv and source.csv and produce a retrosynthesis network as a CSV file providing reactions in the reaction SMILES format and chemicals in both SMILES and InChI formats along with other information like the score for each reaction. Only a single source molecule is processed at this time.

Output
------

* **Reaction Network**\ : CSV file containing retrosynthesis network which provides reactions in the reaction SMILES format and chemicals in both SMILES and InChI formats along with other information like the score for each reaction.
    ]]></help>
    <creator>
        <organization name="BioRetroSynth" url="https://github.com/brsynth" />
    </creator>
    <citations>
        <citation type="doi">10.1016/j.ymben.2017.12.002 </citation>
    </citations>
</tool>