<tool id="gemini_@BINARY@" name="GEMINI @BINARY@" version="@VERSION@">
    <description>Compute useful variant statistics</description>
    <macros>
        <import>gemini_macros.xml</import>
        <token name="@BINARY@">stats</token>
    </macros>
    <expand macro="requirements" />
    <expand macro="stdio" />
    <expand macro="version_command" />
    <command>
<![CDATA[
        gemini @BINARY@
            $stats_type

            #set $multiline_sql_expr = $gt_filter
            #set $cmdln_param = "--gt-filter"
            @MULTILN_SQL_EXPR_TO_CMDLN@

            #set $multiline_sql_expr = $summarize
            #set $cmdln_param = "--summarize"
            @MULTILN_SQL_EXPR_TO_CMDLN@

            "${ infile }"
            > "${ outfile }"
]]>
    </command>
    <inputs>
        <expand macro="infile" />

        <param name="stats_type" type="select" label="Studying ..." help="">
            <option value="--tstv">Compute the transition and transversion ratios for the snps (--tstv)</option>
            <option value="--tstv-coding">Compute the transition/transversion ratios for the snps in the coding regions (--tstv-coding)</option>
            <option value="--tstv-noncoding">Compute the transition/transversion ratios for the snps in the non-coding regions (--tstv-noncoding)</option>
            <option value="--snp-counts">Compute the type and count of the snps (--snp-counts)</option>
            <option value="--sfs">Calculate the site frequency spectrum of the variants (--sfs)</option>
            <option value="--mds">Compute the pair-wise genetic distance between each sample (--mds)</option>
            <option value="--vars-by-sample">Return the total variants per sample, sum of homozygous and heterozygous variants (--vars-by-sample)</option>
            <option value="--gts-by-sample">Return the count of each genotype class observed per sample (--gts-by-sample)</option>
        </param>

        <param name="gt_filter" type="text" area="True" size="5x50" label="Restrictions to apply to genotype values" help="(--gt-filer)">
            <expand macro="sanitize_query" />
        </param>

        <param name="summarize" type="text" area="True" size="5x50" label="The query to be issued to the database to summarize" help="(--summarize)">
            <expand macro="sanitize_query" />
        </param>
    </inputs>
    <outputs>
        <data name="outfile" format="tabular" />
    </outputs>
    <tests>
        <test>
            <!-- test vars-by-sample report -->
            <param name="infile" value="gemini_de_novo_input.db" ftype="gemini.sqlite" />
            <param name="stats_type" value="--vars-by-sample" />
            <output name="outfile">
                <assert_contents>
                    <has_line_matching expression="sample&#009;total" />
                </assert_contents>
            </output>
        </test>
        <test>
            <!-- test gts-by-sample report -->
            <param name="infile" value="gemini_de_novo_input.db" ftype="gemini.sqlite" />
            <param name="stats_type" value="--gts-by-sample" />
            <output name="outfile">
                <assert_contents>
                    <has_line_matching expression="sample&#009;num_hom_ref&#009;num_het&#009;num_hom_alt&#009;num_unknown&#009;total" />
                </assert_contents>
            </output>
        </test>
    </tests>
    <help><![CDATA[
**What it does**

The stats tool computes some useful variant statistics for a GEMINI database.
Like computing the transition and transversion ratios for the snps.

**Settings and examples**

--tstv-coding:
 Compute the transition/transversion ratios for the snps in the coding regions.

--tstv-noncoding:
 Compute the transition/transversion ratios for the snps in the non-coding regions.

EXAMPLE Compute the type and count of the snps; --snp-counts::

 type    count
 A->G    2
 C->T    1
 G->A    1

EXAMPLE Calculate the site frequency spectrum of the variants; --sfs::

 aaf     count
 0.125   2
 0.375   1

EXAMPLE Compute the pair-wise genetic distance between each sample; --mds::

 sample1  sample2  distance
 M10500   M10500   0.0
 M10475   M10478   1.25
 M10500   M10475   2.0
 M10500   M10478   0.5714

EXAMPLE Return a count of the types of genotypes per sample; --gts-by-sample::

 sample   num_hom_ref   num_het   num_hom_alt   num_unknown   total
 M10475   4             1         3             1             9
 M10478   2             2         4             1             9



EXAMPLE Return the total variants per sample (sum of homozygous and heterozygous variants); --vars-by-sample::

 sample  total
 M10475  4
 M10478  6

**Final solution**

--summarize:
 If none of these tools are exactly what you want, you can summarize the variants per sample of an arbitrary query using the –summarize flag. 

EXAMPLE If you wanted to know, for each sample, how many variants are on chromosome 1 that are also in dbSNP;--summarize "select * from variants where in_dbsnp=1 and chrom='chr1'":: 

 sample   total  num_het  num_hom_alt
 M10475   1      1        0
 M128215  1      1        0
 M10478   2      2        0
 M10500   2      1        1


    ]]></help>
    <expand macro="citations"/>
</tool>