<?xml version="1.0" encoding="utf-8"?>
<tool id="scanpy_find_variable_genes" name="Scanpy FindVariableGenes" version="@TOOL_VERSION@+galaxy0">
  <description>based on normalised dispersion of expression</description>
  <macros>
    <import>scanpy_macros2.xml</import>
  </macros>
  <expand macro="requirements"/>
  <command detect_errors="exit_code"><![CDATA[
ln -s '${input_obj_file}' input.h5 &&
PYTHONIOENCODING=utf-8 scanpy-find-variable-genes
    --flavor '${method.flavor}'
#if $method.flavor == 'seurat'
    --mean-limits ${method.min_mean} ${method.max_mean}
    --disp-limits ${method.min_disp} ${method.max_disp}
#else
    --n-top-genes ${method.n_top_gene}
#end if
    --n-bins '${n_bin}'
    ${filter}
    @INPUT_OPTS@
    @OUTPUT_OPTS@
]]></command>

  <inputs>
    <expand macro="input_object_params"/>
    <expand macro="output_object_params"/>
    <conditional name="method">
      <param name="flavor" argument="--flavor" type="select" label="Flavor of computing normalised dispersion">
        <option value="seurat" selected="true">Seurat</option>
        <option value="cell_ranger">Cell-ranger</option>
      </param>
      <when value="seurat">
        <param name="min_mean" argument="--min-mean" type="float" min="0" value="0.0125"
               label="Min value for normalised mean expression (in log1p scale)"/>
        <param name="max_mean" argument="--max-mean" type="float" min="0" value="3"
               label="Max value for normalised mean expresssion (in log1p scale)"/>
        <param name="min_disp" argument="--min-disp" type="float" min="0" value="0.5"
               label="Min value for dispersion of expression"/>
        <param name="max_disp" argument="--max-disp" type="float" min="0" value="50"
               label="Max value for dispersion of expresssion"/>
      </when>
      <when value="cell_ranger">
        <param name="n_top_gene" argument="--n-top-genes" type="integer" value="2000"
               label="Number of top variable genes to keep"/>
      </when>
    </conditional>
    <param name="n_bin" argument="--n-bins" type="integer" value="20" label="Number of bins for binning the mean expression"/>
    <param name="filter" argument="--subset" type="boolean" truevalue="--subset" falsevalue="" checked="false"
           label="Remove genes not marked as highly variable"/>
  </inputs>

  <outputs>
    <data name="output_h5" format="h5" from_work_dir="output.h5" label="${tool.name} on ${on_string}: Variable genes"/>
  </outputs>

  <tests>
    <test>
      <param name="input_obj_file" value="normalise_data.h5"/>
      <param name="input_format" value="anndata"/>
      <param name="output_format" value="anndata"/>
      <param name="flavor" value="seurat"/>
      <param name="n_bin" value="20"/>
      <param name="min_mean" value="0.0125"/>
      <param name="max_mean" value="3"/>
      <param name="min_disp" value="0.5"/>
      <param name="max_disp" value="1e9"/>
      <output name="output_h5" file="find_variable_genes.h5" ftype="h5" compare="sim_size"/>
    </test>
  </tests>

  <help><![CDATA[
==============================================================
Mark highly variable genes (`scanpy.pp.highly_variable_genes`)
==============================================================

Depending on `flavor`, this reproduces the R-implementations of Seurat or Cell Ranger.

The normalized dispersion is obtained by scaling with the mean and standard
deviation of the dispersions for genes falling into a given bin for mean
expression of genes. This means that for each bin of mean expression, highly
variable genes are selected.

@HELP@

@VERSION_HISTORY@
]]></help>
  <expand macro="citations">
    <citation type="doi">10.1038/nbt.3192</citation>
    <citation type="doi">10.1038/ncomms14049</citation>
  </expand>
</tool>