<tool id="ember" name="EMBER" version="1.3.1">
  <description>Step 3 of analysis: searches for over-represented patterns among potential targets</description>
  <command interpreter="bash">Ember_run.sh $matched $expression_data $target_output $model_output $thresh $allgenes</command>
  <inputs>
    <param format="txt" name="matched" type="data" label="Integrated data"/>
    <param format="txt" name="expression_data" type="data" label="Discretized expression data"/>
    <param name="thresh" type="float" min="0" label="Threshold scalar" value="1.0" optional="true"/>
    <param name="allgenes" type="select" label="Print all genes?">
        <option value="n" selected="true">No</option>
        <option value="y">Yes</option>
    </param>
  </inputs>
  <outputs>
    <data format="txt" name="target_output"/>
    <data format="txt" name="model_output"/>
  </outputs>

  <tests>
    <test>
      <param name="matched" value="EMBER/integrated.txt"/>
      <param name="expression_data" value="EMBER/expression_profiles.txt"/>
      <param name="thresh" value="1.0"/>
      <param name="allgenes" value="n"/>
      <output name="target_output" file="EMBER/patterns-1.targets"/>
      <output name="model_output" file="EMBER/patterns-1.model"/>
    </test>
  </tests>

  <help>

This tool uses expectation maximization to determine a subset of potential targets that behave self-consistently, and are thus inferred to be regulatory targets of the transcription factor.

-----

Description of inputs:

*Integrated Data*: output of Integrate Data.

*Discretized Expression Data*: output of PreProcess_Expression_Data, which was also used in Integrate Data (defines the background).

*Threshold scalar*: you can make it harder (or easier) to define a gene as a target by raising (or lowering) the naturally arising threshold by this multiplicative scalar. This setting does not change the model output, it just sets restrictions on which genes are included in the targets output.

*Print all genes*: if set to "Yes", this will print all the potential target genes (and their scores) in the targets list, rather than only those genes with scores above the threshold (so setting this to "Yes" will override any threshold scalar settings). However, genes that do not have a discretized expression profile (because they were always expressed at a low level) will not be included even if this is set to "Yes".

  </help>

</tool>