<tool id="rgGLM1" name="Linear Models:" version="0.2">
    <description>for genotype data</description>
    <code file="rgGLM_code.py"/>
    <command interpreter="python">
        rgGLM.py '$i.extra_files_path/$i.metadata.base_name' '$phef.extra_files_path/$phef.metadata.base_name'
        "$title" '$predvar' '$covar' '$out_file1' '$logf' '$i.metadata.base_name'
        '$inter' '$cond' '$gender' '$mind' '$geno' '$maf' '$logistic' '$gffout'
    </command>

    <inputs>
      <page>
       <param name='title' label='Title for outputs' type='text' value='GLM' size="80" />
       <param name="i" type="data" format="pbed" label="Genotype file" size="80"  />
       <param name="phef"  type="data" format="pphe" label="Phenotype file" size="80"
       help="Dependent variable and covariates will be chosen from this file on the next page"/>
       <param name="logistic" type="text" value = "0" label="1=Use a logistic model (trait must be 1/2 coded like affection)"
       help="Please read the Plink documentation about this option"  />
       <param name="gender" type="text" value = "0" label="1=Add a gender term to model"  />
       <param name='inter' label='1=Build an interaction model - please read the docs carefully before using this'
         type='text' value='0' size="1" />
       <param name="cond"  type="text"  area='true' size='15x20' value = ""
       label="condition on this whitespace delimited rs (snp id) list"  />
       <param name="mind" type="float" value = "0.1" label="Remove subjects with missing genotypes gt (eg 0.1)"
       help = "Set to 1 to include all subjects in the input file" />
       <param name="geno"  type="float" value = "0.1" label="Remove markers with missing genotypes gt (eg 0.1)"
       help = "Set to 1 to include all markers in the input file"  />
       <param name="maf"  type="float" value = "0.01" label="Remove markers with MAF lt (eg 0.01) "
       help = "Set to 0 to include all markers in the input file"/>
      </page>
      <page>
       <param name="predvar" size="80"  type="select" label="Dependent Trait"
       dynamic_options="get_phecols(phef=phef,selectOne=1)"  display="radio" multiple="false"
       help="Model this characteristic in terms of subject snp genotypes - eg rare allele dosage for additive model" />
       <param name="covar" size="80"  type="select" label="Covariates"
       dynamic_options="get_phecols(phef=phef,selectOne=0)" multiple="true" display="checkboxes"
       help="Use these phenotypes as covariates in models of snp dosage effects on the dependent trait"/>
      </page>
   </inputs>

   <outputs>
       <data format="tabular" name="out_file1" label="${title}_rgGLM.xls"/>
       <data format="txt" name="logf" label="${title}_rgGLMlog.txt" />
       <data format="gff" name="gffout"  label="${title}_rgGLM.gff"/>
   </outputs>
<tests>
 <test>
  <param name='i' value='tinywga' ftype='pbed' >
   <metadata name='base_name' value='tinywga' />
   <composite_data value='tinywga.bim' />
   <composite_data value='tinywga.bed' />
   <composite_data value='tinywga.fam' />
   <edit_attributes type='name' value='tinywga' /> 
 </param>
 <param name='phef' value='tinywga' ftype='pphe' >
   <metadata name='base_name' value='tinywga' />
   <composite_data value='tinywga.pphe' />
   <edit_attributes type='name' value='tinywga' /> 
 </param>
 <param name='title' value='rgGLMtest1' />
 <param name='predvar' value='c1' />
 <param name='covar' value='None' />
 <param name='inter' value='0' />
 <param name='cond' value='' />
 <param name='gender' value='0' />
 <param name='mind' value='1.0' />
 <param name='geno' value='1.0' />
 <param name='maf' value='0.0' />
 <param name='logistic' value='0' />
 <output name='out_file1' file='rgGLMtest1_GLM.xls' ftype='tabular' compare="diff" />
 <output name='logf' file='rgGLMtest1_GLM_log.txt' ftype='txt' compare="diff" lines_diff='36'/>
 <output name='gffout' file='rgGLMtest1_GLM_topTable.gff' compare="diff" ftype='gff' />
 </test>
</tests>
<help>

.. class:: infomark

**Syntax**

Note this is a two form tool - you will choose the dependent trait and covariates
on the second page based on the phenotype file you choose on the first page

- **Genotype file** is the input Plink format compressed genotype (pbed) file
- **Phenotype file** is the input Plink phenotype (pphe) file with FAMID IID followed by phenotypes
- **Dependant variable** is the term on the left of the model and is chosen from the pphe columns on the second page
- **Logistic** if you are (eg) using disease status as the outcome variable (case/control) - otherwise the model is linear.
- **Covariates** are covariate terms on the right of the model, also chosen on the second page
- **Interactions** will add interactions - please be careful how you interpret these - see the Plink documentation.
- **Gender** will add gender as a model term - described in the Plink documentation
- **Condition** will condition the model on one or more specific SNP rs ids as a whitespace delimited sequence
- **Format** determines how your data will be returned to your Galaxy workspace

-----

.. class:: infomark

**Summary**

This tool will test GLM models for SNP predicting a dependent phenotype
variable with adjustment for specified covariates.

If you don't see the genotype or phenotype data set you want here, it can be imported using
one of the methods available from the rg get data tool group.

Output format can be UCSC .bed if you want to see one column of your
results as a fully fledged UCSC genome browser track. A map file containing the chromosome and offset for each marker is
required for writing this kind of output.
Alternatively you can use .gg for the UCSC Genome Graphs tool which has all of the advantages
of the the .bed track, plus a neat, visual front end that displays a lot of useful clues.
Either of these are a very useful way of quickly getting a look
at your data in full genomic context.

Finally, if you can't live without
spreadsheet data, choose the .xls tab delimited format. It's not a stupid binary excel file. Just a plain old tab
delimited
one with a header. Fortunately excel is dumb enough to open these without much protest.

-----

.. class:: infomark

**Attribution**

This Galaxy tool relies on Plink (see Plinksrc_) to test GLM models. 

So, we rely on the author (Shaun Purcell) for the documentation you need specific to those settings - they are very nicely documented - see
DOC_

Tool and Galaxy datatypes originally designed and written for the Rgenetics
series of whole genome scale statistical genetics tools by ross lazarus (ross.lazarus@gmail.com)

Copyright Ross Lazarus March 2007
This Galaxy wrapper is released licensed under the LGPL_ but is about as useful as a chocolate teapot without Plink which is GPL.

I'm no lawyer, but it looks like you got GPL if you use this software. Good luck.

.. _Plinksrc: http://pngu.mgh.harvard.edu/~purcell/plink/ 

.. _LGPL: http://www.gnu.org/copyleft/lesser.html

.. _DOC: http://pngu.mgh.harvard.edu/~purcell/plink/anal.shtml#glm

</help>
</tool>