<tool id="gd_find_intervals" name="Remarkable Intervals" version="1.1.0">
  <description>: Find high-scoring runs of SNPs</description>

  <command interpreter="python">
    find_intervals.py "$input" "$input.metadata.dbkey" "$output" "$output.files_path"

    #if $override_metadata.choice == "0"
      "$input.metadata.ref" "$input.metadata.rPos"
    #else
      "$override_metadata.ref_col" "$override_metadata.rpos_col"
    #end if

    "$score_col" "$shuffles"

    #if $cutoff.type == 'percentage'
      "$cutoff.cutoff_pct"
    #else
      "=$cutoff.cutoff_val"
    #end if

    "$out_format"
  </command>

  <inputs>
    <param name="input" type="data" format="tabular" label="Dataset">
      <validator type="unspecified_build" message="This dataset does not have a reference species and cannot be used with this tool" />
    </param>

    <param name="score_col" type="data_column" data_ref="input" numerical="true" label="Column with score"/>

    <conditional name="cutoff">
      <param name="type" type="select" label="Score-shift type">
        <option value="percentage">percentage</option>
        <option value="value">value</option>
      </param>
      <when value="percentage">
        <param name="cutoff_pct" type="float" value="95" min="0" max="100" label="Percentage score-shift"/>
      </when>
      <when value="value">
        <param name="cutoff_val" type="float" value="0.0" label="Value score-shift"/>
      </when>
    </conditional>

    <param name="shuffles" type="integer" min="0" value="0" label="Number of randomizations"/>

    <param name="out_format" type="select" format="integer" label="Report individual positions">
      <option value="0" selected="true">no</option>
      <option value="1">yes</option>
    </param>

    <conditional name="override_metadata">
      <param name="choice" type="select" format="integer" label="Choose columns" help="Note: you must choose the columns if the input dataset is not gd_snp.">
        <option value="0" selected="true">no, get columns from metadata</option>
        <option value="1" >yes, choose columns here</option>
      </param>
      <when value="0" />
      <when value="1">
        <param name="ref_col" type="data_column" data_ref="input" numerical="false" label="Column with reference chromosome" help="Note: be sure this corresponds to the build recorded in the metadata."/>
        <param name="rpos_col" type="data_column" data_ref="input" numerical="true" label="Column with reference position" help="Note: either zero-based or one-based positions will work."/>
      </when>
    </conditional>
  </inputs>

  <outputs>
    <data name="output" format="interval">
        <change_format>
            <when input="out_format" value="1" format="bigwigpos" />
        </change_format>
    </data>
  </outputs>

  <tests>
    <test>
      <param name="input" value="test_in/sample.gd_snp" ftype="gd_snp" />
      <param name="score_col" value="5" />
      <param name="type" value="value" />
      <param name="cutoff_val" value="700.0" />
      <param name="shuffles" value="10" />
      <param name="out_format" value="0" />
      <param name="choice" value="0" />

      <output name="output" file="test_out/find_intervals/find_intervals.interval" />
    </test>
  </tests>

  <help>

**Dataset formats**

The input dataset is tabular_, with required columns of chromosome, position,
and score (in any column).
The output dataset is interval_.  (`Dataset missing?`_)

.. _interval: ./static/formatHelp.html#interval
.. _tabular: ./static/formatHelp.html#tab
.. _Dataset missing?: ./static/formatHelp.html

-----

**What it does**

The user selects a tabular dataset (such as a gd_snp dataset) and 
if the dataset is not also gd_snp format, specifies 
the columns containing chromosome, position, and scores (such as an Fst-value for the SNP). 
For gd_snp format the metadata can be used to specify the chromosome and 
position.
Other inputs include
a percentage or raw score for the "score-shift" which should be greater than the 
average value for the scores column.  A higher value will give smaller intervals
in the output.
If a percentage (e.g. 95%) is specified
then that percentile of the scores is used as the shift; 
percentile may not work well if many rows or SNPs have the same score
(in that case use a raw score).  The program subtracts the
shift from every score, then finds genomic intervals (i.e., consecutive runs
of SNPs) whose total score cannot be increased by adding or subtracting one
or more adjusted scores at the ends of the interval.
Another input is the number of times the
data should be randomized (only intervals with score exceeding the maximum for
the randomized data are reported).  
If 100 shuffles are requested, then any interval reported by the tool has a 
score with probability less than 0.01 of being equaled or exceeded by chance.

-----

**Example**

- input (gd_snp)::

    Contig222_chr2_9817738_9818143   220     C       T       888.0   chr2    9817960         C       17      0       2       78      12      0       2       63      20      0       2       87      8       0       2       51      11      0       2       60      12      0       2       63      Y       76      0.093   1
    Contig47_chr2_25470778_25471576  126     G       A       888.0   chr2    25470896        G       12      0       2       63      14      0       2       69      14      0       2       69      10      0       2       57      18      0       2       81      13      0       2       66      N       11      0.289   1
    ...
    Contig115_chr2_61631913_61632510 310     G       T       999.3   chr2    61632216        G       7       0       2       48      9       0       2       54      7       0       2       48      11      0       2       60      10      0       2       57      10      0       2       57      N       13      0.184   0
    Contig31_chr2_67331584_67331785  39      C       T       999.0   chr2    67331623        C       11      0       2       60      10      0       2       57      7       0       2       48      9       0       2       54      2       0       2       33      4       0       2       39      N       110     0.647   1
    etc.

- output not reporting individual positions::

    chr2    9817960 67331624        1272.2000

  </help>
</tool>