<tool id="gd_evaluate_population_numbers" name="Population Complexity" version="1.0.0">
  <description>: Evaluate possible numbers of ancestral populations</description>

  <command interpreter="bash">
    evaluate_population_numbers.bash "${input.extra_files_path}/admix.ped" "$output" "$max_populations"
  </command>

  <inputs>
    <param name="input" type="data" format="gd_ped" label="Dataset" />
    <param name="max_populations" type="integer" min="1" value="5" label="Maximum number of populations" />
  </inputs>

  <outputs>
    <data name="output" format="txt" />
  </outputs>

  <!--
  <tests>
    <test>
      <param name="input" value="fake" ftype="gd_ped" >
        <metadata name="base_name" value="admix" />
        <composite_data value="test_out/prepare_population_structure/prepare_population_structure.html" />
        <composite_data value="test_out/prepare_population_structure/admix.ped" />
        <composite_data value="test_out/prepare_population_structure/admix.map" />
        <edit_attributes type="name" value="fake" />
      </param>
      <param name="max_populations" value="2" />

      <output name="output" file="test_out/evaluate_population_numbers/evaluate_population_numbers.txt" />
    </test>
  </tests>
  -->

  <help>

**Dataset formats**

The input dataset is in gd_ped_ format.
The output dataset is text.  (`Dataset missing?`_)

.. _gd_ped: ./static/formatHelp.html#gd_ped
.. _Dataset missing?: ./static/formatHelp.html

-----

**What it does**

The user selects a gd_ped dataset generated by the Prepare Input tool.
For all possible numbers K of ancestral
populations, from 1 up to a user-specified maximum, this tool produces values
that indicate how well the data can be explained as genotypes from individuals
derived from K ancestral populations.  These values are computed by a 5-fold
cross-validation procedure, so that a good choice for K will exhibit a low
cross-validation error (CVE) compared with other potential settings for K.

-----

**Acknowledgments**

We use the program "Admixture", downloaded from

http://www.genetics.ucla.edu/software/admixture/

and described in the paper "Fast model-based estimation of ancestry in
unrelated individuals" by David H. Alexander, John Novembre and Kenneth Lange,
Genome Research 19 (2009), pp. 1655-1664. Admixture is called with the "--cv"
flag to produce these values.

----- 

**Example**

- output with max populations of 6::

    CVE (K=1): 1.10120
    CVE (K=2): 1.34683
    CVE (K=3): 1.80611
    CVE (K=4): 1.96339
    CVE (K=5): 1.21522
    CVE (K=6): 0.51501

  </help>
</tool>