<tool id="Annovar" name="ANNOVAR" version="2013aug">
	<description> Annotate a file using ANNOVAR </description>
	
	<requirements>		
		<requirement type="package" version="1.7">cgatools</requirement>
	</requirements>
	
	<command interpreter="bash">
		annovar.sh		
		--impactscores ${impactscores}
		--esp ${esp}
		--gerp ${gerp}
		--cosmic61 ${cosmic61}
		--cosmic63 ${cosmic63}	
		--cosmic64 ${cosmic64}		
		--cosmic65 ${cosmic65}		
		--outall ${annotated}		
		--outinvalid ${invalid}
		--dorunannovar ${dorun}
		--inputfile ${infile}
		--buildver ${reference.fields.dbkey}
		--humandb ${reference.fields.ANNOVAR_humandb}
		--scriptsdir ${reference.fields.ANNOVAR_scripts}	
		--verdbsnp ${verdbsnp}
		--geneanno ${geneanno}
		--tfbs ${tfbs}
		--mce ${mce}
		--cytoband ${cytoband}
		--segdup ${segdup}
        --dgv ${dgv}
		--gwas ${gwas}				
		#if $filetype.type == "other"
			--varfile N
			--VCF N
			--chrcol ${filetype.col_chr}
			--startcol ${filetype.col_start}
			--endcol ${filetype.col_end}
			--obscol ${filetype.col_obs}
			--refcol ${filetype.col_ref}
		
			#if $filetype.convertcoords.convert == "Y"
				--vartypecol ${filetype.convertcoords.col_vartype}
				--convertcoords Y
			#else
				--convertcoords N
			#end if
		#end if
		#if $filetype.type == "vcf"
			--varfile N
			--VCF Y
			--convertcoords N
		#end if
		#if $filetype.type == "varfile"
			--varfile Y
			--VCF N			
		#end if			
		--cg46 ${cgfortysix}
		--cg69 ${cgsixtynine}
		--ver1000g ${ver1000g}
		
	</command>
		
	<inputs>
		<param name="dorun" type="hidden" value="Y"/> <!-- will add tool in future to filter on annovar columns, then will call annovar.sh with dorun==N -->
		<param name="reference" type="select" label="Reference">
			<options from_data_table="annovar_loc" />				
		</param>
				
		<param name="infile" type="data" label="Select file to annotate" help="Must be CG varfile or a tab-separated file with a 1 line header"/>
		<conditional name="filetype">
			<param name="type" type="select" label="Select filetype" >
				<option value="vcf" selected="false"> VCF4 file </option>
				<option value="varfile" selected="false"> CG varfile </option>
				<option value="other" selected="false"> Other </option>
			</param>
			<when value="other">
				<param name="col_chr"     type="data_column"   data_ref="infile" multiple="False" label="Chromosome Column"  /> 
				<param name="col_start"   type="data_column"   data_ref="infile" multiple="False" label="Start Column"  /> 
				<param name="col_end"     type="data_column"   data_ref="infile" multiple="False" label="End Column"  /> 
				<param name="col_ref"     type="data_column"   data_ref="infile" multiple="False" label="Reference Allele Column"  /> 
				<param name="col_obs"     type="data_column"   data_ref="infile" multiple="False" label="Observed Allele Column"  /> 	
				<conditional name="convertcoords">
					<param name="convert" type="select" label="Is this file using Complete Genomics (0-based half-open) cooridinates?" >
						<option value="Y"> Yes </option>
						<option value="N" selected="True"> No </option>
					</param>
					<when value="Y">
						<param name="col_vartype" type="data_column"   data_ref="infile" multiple="False" label="varType Column"  /> 
					</when>
				</conditional>
			</when>
		</conditional>



		<param name="geneanno" type="select" label="Select Gene Annotation(s)" multiple="true" optional="true" display="checkboxes">			
			<option value="refSeq" selected="true"  > RefSeq </option>
			<option value="knowngene"> UCSC KnownGene </option>
			<option value="ensgene"  > Ensembl </option>			
		</param>	


		<!-- region-based annotation -->
		<param name="cytoband" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Cytogenic band Annotation?" help="This option identifies Giemsa-stained chromosomes bands, (e.g. 1q21.1-q23.3)."/>
		<param name="tfbs" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Transcription Factor Binding Site Annotation?"/>
		<param name="mce" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Most Conserved Elements Annotation?" help="This option phastCons 44-way alignments to annotate variants that fall within conserved genomic regions."/>
		<param name="segdup" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Segmental Duplication Annotation?" help="Genetic variants that are mapped to segmental duplications are most likely sequence alignment errors and should be treated with extreme caution."/>
		<param name="dgv" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="DGV (Database of Genomic Variants) Annotation?" help="Identify previously reported structural variants in DGV (Database of Genomic Variants) "/>
		<param name="gwas" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="GWAS studies Annotation?" help="Identify variants reported in previously published GWAS (Genome-wide association studies) "/>



		<!-- filter-based annotation -->
		<param name="verdbsnp" type="select" label="Select dbSNP version(s) to annotate with" multiple="true" display="checkboxes"  optional="true" help="SNPs in dbSNP may be flagged as Clinically Associated, Select the NonFlagged version if you do not wish to annotate with these SNPs ">			
			<option value="snp128"          > 128            (hg18/hg19) </option>
			<option value="snp128NonFlagged"> 128 NonFlagged  </option>
			<option value="snp129"          > 129            (hg18/hg19) </option>
			<option value="snp129NonFlagged"> 129 NonFlagged  </option>			
			<option value="snp130"          > 130            (hg18/hg19) </option>
			<option value="snp130NonFlagged"> 130 NonFlagged  </option>
			<option value="snp131"          > 131            (hg18/hg19) </option>	
			<option value="snp131NonFlagged"> 131 NonFlagged  </option>
			<option value="snp132"          > 132            (hg18/hg19) </option>
			<option value="snp132NonFlagged"> 132 NonFlagged  </option>
			<option value="snp135"          > 135            (hg19 only) </option>	
			<option value="snp135NonFlagged"> 135 NonFlagged  </option>
			<option value="snp137"          > 137            (hg19 only) </option>				
			<option value="snp137NonFlagged"> 137 NonFlagged  </option>			
		</param>	

		<param name="ver1000g" type="select" label="Select 1000Genomes Annotation(s)" multiple="true" display="checkboxes"  optional="true" help="2012april database for ALL populations was converted to hg18 using the UCSC liftover program">			
			<option value="1000g2012apr"> 2012apr (hg18/hg19) (5 populations: AMR,AFR,ASN,CEU,ALL) </option>
			<option value="1000g2012feb"> 2012feb (hg19) (1 population: ALL) </option>
			<option value="1000g2010nov"> 2010nov (hg19) (1 population: ALL) </option>
			<option value="1000g2010jul"> 2010jul (hg18) (4 populations: YRI,JPT,CHB,CEU)</option>			
		</param>	
		<!-- 
		<param name="g1000" type="boolean" checked="True" truevalue="Y" falsevalue="N" label="Annotate with 1000genomes project? (version 2012april)"/>
		-->


	<param name="esp" type="select" label="Select Exome Variant Server  version(s) to annotate with" multiple="true" display="checkboxes"  optional="true" help="si versions of databases contain indels and chrY calls">			
			<option value="esp6500si_all"       > ESP6500si ALL  </option>
			<option value="esp6500si_ea"        > ESP6500si European Americans  </option>
			<option value="esp6500si_aa"        > ESP6500si African Americans  </option>
			<option value="esp6500_all"         > ESP6500   ALL </option>
			<option value="esp6500_ea"          > ESP6500   European Americans  </option>
			<option value="esp6500_aa"          > ESP6500   African Americans   </option>			
			<option value="esp5400_all"         > ESP5400   ALL  </option>
			<option value="esp5400_ea"          > ESP5400   European Americans  </option>
			<option value="esp5400_aa"          > ESP5400   African Americans  </option>			
		</param>	


		<param name="gerp" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="GERP++ Annotation?" help="GERP identifies constrained elements in multiple alignments by quantifying substitution deficits (see http://mendel.stanford.edu/SidowLab/downloads/gerp/ for details) This option annotates those variants having GERP++>2 in human genome, as this threshold is typically regarded as evolutionarily conserved and potentially functional"/>
	
		<param name="cgfortysix" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Complete Genomics 46 Genomes?" help="Diversity Panel; 46 unrelated individuals"/>
		<param name="cgsixtynine" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Complete Genomics 69 Genomes?" help="Diversity Panel, Pedigree, YRI trio and PUR trio"/>
		<param name="cosmic61" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Annotate with COSMIC61? (hg19 only)"/>
		<param name="cosmic63" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Annotate with COSMIC63? (hg19 only)"/>
		<param name="cosmic64" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Annotate with COSMIC64? (hg19 only)"/>
		<param name="cosmic65" type="boolean" checked="False" truevalue="Y" falsevalue="N" label="Annotate with COSMIC65? (hg19 only)"/>
		
<param name="impactscores" type="select" label="Select functional impact scores annotate with" multiple="true" display="checkboxes" optional="true" help="LJB refers to Liu, Jian, Boerwinkle paper in Human Mutation, pubmed ID 21520341.">			
			<option value="avsift"> AV SIFT </option>
			<option value="ljbsift"> LJB SIFT (corresponds to 1-SIFT)</option>
			<option value="pp2"> PolyPhen2 </option>
			<option value="mutationtaster" > MutationTaster </option>
			<option value="lrt"> LRT (Likelihood Ratio Test) </option>			
			<option value="phylop"> PhyloP </option>
		</param>	
			

		<!-- prefix for output file so you dont have to manually rename history items -->
		<param name="fname" type="text" value="" label="Prefix for your output file" help="Optional"/>		
				
	</inputs>

	<outputs>
		<data format="tabular" name="invalid"   label="$fname ANNOVAR Invalid input on ${on_string}"/>	
		<data format="tabular" name="annotated" label="$fname ANNOVAR Annotated variants on ${on_string}"/>
	</outputs>

	<help> 
**What it does**

This tool will annotate a file using ANNOVAR.

**ANNOVAR Website and Documentation**

Website: http://www.openbioinformatics.org/annovar/

Paper: http://nar.oxfordjournals.org/content/38/16/e164

**Input Formats**

Input Formats may be one of the following:
	
	VCF file
	
	Complete Genomics varfile
	
	Custom tab-delimited file (specify chromosome, start, end, reference allele, observed allele columns)		

	Custom tab-delimited CG-derived file (specify chromosome, start, end, reference allele, observed allele, varType columns)
		
	</help>

</tool>