<tool id="abundance_estimates_to_matrix" name="Build expression matrix" version="2.1.1">
    <description>for a de novo assembly of RNA-Seq data by Trinity</description>
    <requirements>
        <requirement type="package" version="2.1.1">trinity</requirement>
        <requirement type="package" version="1.1.2">bowtie</requirement>
        <requirement type="package" version="1.2">samtools</requirement>
        <requirement type="package" version="3.12.0">edger</requirement>
        <requirement type="package" version="3.12.0">bioconductor-edger</requirement>
        <requirement type="package" version="1.2.28">rsem</requirement>
    </requirements>
    <stdio>
        <exit_code range="1:"/>
    </stdio>
    <command><![CDATA[
        #for $entry in $samples:
            ln -s "${entry.file}" "${entry.sample_name}.tsv" &&
        #end for

        abundance_estimates_to_matrix.pl

        --est_method ${est_method}
        --cross_sample_norm ${additional_params.cross_sample_norm}

        #for $entry in $samples:
            "${entry.sample_name}.tsv"
        #end for
    ]]></command>
    <inputs>
        <repeat name="samples" title="RSEM abundance estimates for samples">
            <param name="file" label="Add file" type="data" format="tabular"/>
            <param name="sample_name" label="Sample name" type="text">
                <validator type="regex" message="Value must be a not empty string composed by alphanumeric characters and underscores">^\w+$</validator>
            </param>
        </repeat>
        
        <param type="select" name="est_method" label="Abundance estimation method">
            <option value="RSEM">RSEM</option>
            <option value="eXpress">eXpress</option>
        </param>

        <section name="additional_params" title="Additional Options" expanded="False">
            <param type="select" name="cross_sample_norm" label="Cross sample normalization">
                <option value="TMM">TMM</option>
                <option value="UpperQuartile">UpperQuartile</option>
                <option value="none">None</option>
            </param>
        </section>
    </inputs>
    <outputs>
        <data format="tabular" name="trans_counts" label="${tool.name} on ${on_string}: estimated RNA-Seq fragment counts (raw counts)" from_work_dir="matrix.counts.matrix"/>
        <data format="tabular" name="TPM_no_norm" label="${tool.name} on ${on_string}: matrix of TPM expression values (not cross-sample normalized)" from_work_dir="matrix.TPM.not_cross_norm"/>
        <data format="tabular" name="norm" label="${tool.name} on ${on_string}: matrix of TMM-normalized expression values" from_work_dir="matrix.TMM.EXPR.matrix">
            <filter>additional_params['cross_sample_norm'] == "TMM"</filter>
        </data>
        <data format="tabular" name="norm" label="${tool.name} on ${on_string}: matrix of UpperQuartile-normalized expression values" from_work_dir="matrix.UpperQuartile.EXPR.matrix">
            <filter>additional_params['cross_sample_norm'] == "UpperQuartile"</filter>
        </data>
    </outputs>
    <tests>
        <test>
            <param name="samples_0|file" value="count/rsem_bowtie/RSEM.genes.results"/>
            <param name="samples_0|sample_name" value="sample_A"/>
            <param name="samples_1|file" value="count/rsem_bowtie/RSEM.genes_B.results"/>
            <param name="samples_1|sample_name" value="sample_B"/>
            <param name="est_method" value="RSEM"/>
            <param name="cross_sample_norm" value="TMM"/>
            <output name="trans_counts">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="TPM_no_norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="samples_0|file" value="count/rsem_bowtie2/RSEM.genes.results"/>
            <param name="samples_0|sample_name" value="sample_A"/>
            <param name="samples_1|file" value="count/rsem_bowtie2/RSEM.genes_B.results"/>
            <param name="samples_1|sample_name" value="sample_B"/>
            <param name="est_method" value="RSEM"/>
            <param name="cross_sample_norm" value="TMM"/>
            <output name="trans_counts">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="TPM_no_norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="samples_0|file" value="count/express_bowtie/results.xprs.genes"/>
            <param name="samples_0|sample_name" value="sample_A"/>
            <param name="samples_1|file" value="count/express_bowtie/results_B.xprs.genes"/>
            <param name="samples_1|sample_name" value="sample_B"/>
            <param name="est_method" value="eXpress"/>
            <param name="cross_sample_norm" value="TMM"/>
            <output name="trans_counts">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="TPM_no_norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
        </test>
        <test>
            <param name="samples_0|file" value="count/rsem_bowtie/RSEM.genes.results"/>
            <param name="samples_0|sample_name" value="sample_A"/>
            <param name="samples_1|file" value="count/rsem_bowtie/RSEM.genes_B.results"/>
            <param name="samples_1|sample_name" value="sample_B"/>
            <param name="est_method" value="RSEM"/>
            <param name="cross_sample_norm" value="UpperQuartile"/>
            <output name="trans_counts">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="TPM_no_norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="norm" file="count/matrix_rsem_bowtie_UpperQuartile/matrix.UpperQuartile.EXPR.matrix" />
        </test>
        <test>
            <param name="samples_0|file" value="count/rsem_bowtie/RSEM.genes.results"/>
            <param name="samples_0|sample_name" value="sample_A"/>
            <param name="samples_1|file" value="count/rsem_bowtie/RSEM.genes_B.results"/>
            <param name="samples_1|sample_name" value="sample_B"/>
            <param name="est_method" value="RSEM"/>
            <param name="cross_sample_norm" value="none"/>
            <output name="trans_counts">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="TPM_no_norm">
                <assert_contents>
                    <has_line_matching expression="TRINITY_DN3_c0_g1&#009;.*" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
        </test>
    </tests>
    <help>
<![CDATA[
Trinity_ assembles transcript sequences from Illumina RNA-Seq data.
This tool will combine abundance estimations (produced by 'Align reads and estimate abundance on a de novo assembly of RNA-Seq data' tool) from multiple samples into a single tabular file.
This matrix can then be used by 'RNASeq samples quality check for transcript quantification' and 'Differential Expression Analysis using a Trinity assembly' tools.

**Inputs**

It takes as input multiple results from 'Align reads and estimate abundance on a de novo assembly of RNA-Seq data' tool/
Each sample must have a name, that should be used in subsequent tools.

**Output**

This tool will produce a single matrix file. More details on this page:

.. _Trinity manual: https://github.com/trinityrnaseq/trinityrnaseq/wiki/Trinity-Transcript-Quantification


.. _Trinity: http://trinityrnaseq.github.io
]]>
    </help>
    
     <citations>
        <citation type="doi">doi:10.1038/nbt.1883</citation>
    </citations>
</tool>