<tool id="crossmap" name="CrossMap" version="0.2.d">
    <description>Convert genome coordinates or annotation files between genome assemblies</description>
    
    <requirements>
        <requirement type="package" version="312">ucsc_tools</requirement>
        <requirement type="package" version="2.7">python</requirement>
        <requirement type="package" version="0.2">crossmap</requirement>
    </requirements>
    
    <stdio>
        <regex match="Usage: CrossMap.py" source="stdout" level="fatal"/>
        <regex match=".*" source="both" level="log"/>
    </stdio>
    
    <version_command>python $CROSSMAP_ROOT_DIR/bin/CrossMap.py 2&gt;&amp;1 | head -n 1 | grep -E --only-matching 'CrossMap.*'</version_command>
    
    <!-- First line of CrossMap has hardcoded "#!/usr/bin/python". This makes use of Galaxy's python: -->
    <command><![CDATA[
        python \$CROSSMAP_ROOT_DIR/bin/CrossMap.py
            ${multiple.input_format.replace("sam","bam")}
            
            #if $multiple.input_format == "vcf" and $multiple.seq_source.index_source == "cached"
                <!-- This is the 2nd dbkey, and the corresponding value has to be looked up -->
                "${filter(lambda x: str( x[1] ) == str($multiple.seq_source.input_chain ), $__app__.tool_data_tables['liftOver'].get_fields())[0][2] }"
            #else
                "$multiple.seq_source.input_chain"
            #end if
            
            #if $multiple.input_format in ["bam", "sam"]
                -m $multiple.insert_size
                -s $multiple.insert_size_stdev
                -t $multiple.insert_size_fold
            #end if
            
            "$multiple.seq_source.input"
            
            #if $multiple.input_format == "vcf"
                "$multiple.seq_source.input_fasta"
            #end if 
            
            #if str($multiple.include_fails) == "True"
            >
            #end if
            
            "$output"
            
            #if $multiple.input_format in ["bam", "sam"]
                && mv "${output}.${multiple.input_format}" "$output"
                && mv "${output}.unmap.${multiple.input_format}" "$output_unmapped"
            #else if $multiple.input_format in ["vcf"]
                && mv "${output}" "$output"
                && mv "${output}.unmap" "$output_unmapped"
            #else if $multiple.input_format in ["wig", "bigwig"]
                && mv "${output}.bw" "$output"
                && mv "${output}.sorted.bgr" "$output2"
            #end if
    ]]></command>

    <inputs>
        <conditional name="multiple">
            <param name="input_format" type="select" label="Convert a file of the following format">
                <option value="bam">BAM</option>
                <option value="sam">SAM</option>
                <option value="bed">BED or BED-like</option>
                <option value="bigwig">BigWig</option>
                <option value="gff">GFF or GTF</option>
                <option value="vcf">VCF</option>
                <option value="wig">Wiggle or bedGraph</option>
            </param>
            <when value="bam">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                        <when value="cached">
                            <param type="data" format="bam" name="input" label="BAM file">
                                <validator type="unspecified_build" />
                                <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                            </param>
                            <param name="input_chain" type="select" label="Lift Over To">
                                <options from_file="liftOver.loc">
                                    <column name="name" index="1"/>
                                    <column name="value" index="2"/>
                                    <column name="dbkey" index="0"/>
                                    <filter type="data_meta" ref="input" key="dbkey" column="0" />
                                </options>
                            </param>
                        </when>
                        <when value="history">
                            <param type="data" format="bam,sam" name="input" label="BAM/SAM file" />
                            <param type="data" format="csv" name="input_chain" label="LiftOver chain file" />
                        </when>
                </conditional>
                
                <param name="insert_size" type="float" value="200.0" label="Insert size (-m)" help="Average insert size of pair-end sequencing (bp) [default=200.0]" />
                <param name="insert_size_stdev" type="float" value="30.0" label="Insert size std. dev (-s)" help="Stanadard deviation of insert size. [default=30.0]" />
                <param name="insert_size_fold" type="float" value="3.0" label="Insert size std. dev foldchange (-t)" help="A mapped pair is considered as 'proper pair' if both ends mapped to different strand and the distance between them is less then '-t' * stdev from the mean. [default=3.0]" />
                
                <param name="include_fails" type="hidden" tvalue="False" />
            </when>
            <when value="sam"><!-- BAM and SAM are exactly the same conditions, but they need to be separate to get the proper output format -->
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                        <when value="cached">
                            <param type="data" format="sam" name="input" label="SAM file">
                                <validator type="unspecified_build" />
                                <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                            </param>
                            <param name="input_chain" type="select" label="Lift Over To">
                                <options from_file="liftOver.loc">
                                    <column name="name" index="1"/>
                                    <column name="value" index="2"/>
                                    <column name="dbkey" index="0"/>
                                    <filter type="data_meta" ref="input" key="dbkey" column="0" />
                                </options>
                            </param>
                        </when>
                        <when value="history">
                            <param type="data" format="bam,sam" name="input" label="BAM/SAM file" />
                            <param type="data" format="csv" name="input_chain" label="LiftOver chain file" />
                        </when>
                </conditional>
                
                <param name="insert_size" type="float" value="200.0" label="Insert size (-m)" help="Average insert size of pair-end sequencing (bp) [default=200.0]" />
                <param name="insert_size_stdev" type="float" value="30.0" label="Insert size std. dev (-s)" help="Stanadard deviation of insert size. [default=30.0]" />
                <param name="insert_size_fold" type="float" value="3.0" label="Insert size std. dev foldchange (-t)" help="A mapped pair is considered as 'proper pair' if both ends mapped to different strand and the distance between them is less then '-t' * stdev from the mean. [default=3.0]" />
                
                <param name="include_fails" type="hidden" tvalue="False" />
            </when>
            <when value="bed">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                        <when value="cached">
                            <param format="bed" name="input" type="data" label="BED file" help="BED format file must have at least 3 columns (chrom, start, end) and no more than 12 columns.">
                                <validator type="unspecified_build" />
                                <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                            </param>
                            <param name="input_chain" type="select" label="Lift Over To">
                                <options from_file="liftOver.loc">
                                    <column name="name" index="1"/>
                                    <column name="value" index="2"/>
                                    <column name="dbkey" index="0"/>
                                    <filter type="data_meta" ref="input" key="dbkey" column="0" />
                                </options>
                            </param>
                        </when>
                        <when value="history">
                            <param type="data" format="bed" name="input"       label="BED file" help="BED format file must have at least 3 columns (chrom, start, end) and no more than 12 columns." />
                            <param type="data" format="csv" name="input_chain" label="LiftOver chain file" />
                        </when>
                </conditional>
                
                <param name="include_fails" type="boolean" truevalue="True" checked="false" falsevalue="False" label="Include failed liftovers" help="If a coordinate can not be lift over, do you want to include it in the output (it is still being marked 'fail')" />
            </when>
            <when value="bigwig">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                        <when value="cached">
                            <param format="bigwig" name="input" type="data" label="BigWig file">
                                <validator type="unspecified_build" />
                                <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                            </param>
                            <param name="input_chain" type="select" label="Lift Over To">
                                <options from_file="liftOver.loc">
                                    <column name="name" index="1"/>
                                    <column name="value" index="2"/>
                                    <column name="dbkey" index="0"/>
                                    <filter type="data_meta" ref="input" key="dbkey" column="0" />
                                </options>
                            </param>
                        </when>
                        <when value="history">
                            <param type="data" format="bigwig" name="input"       label="BigWig file" />
                            <param type="data" format="csv"    name="input_chain" label="LiftOver chain file" />
                        </when>
                </conditional>
                
                <param name="include_fails" type="hidden" tvalue="False" />
            </when>
            <when value="gff">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                        <when value="cached">
                            <param format="gtf,gff,gff3" name="input" type="data" label="GTF/GFF file">
                                <validator type="unspecified_build" />
                                <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                            </param>
                            <param name="input_chain" type="select" label="Lift Over To">
                                <options from_file="liftOver.loc">
                                    <column name="name" index="1"/>
                                    <column name="value" index="2"/>
                                    <column name="dbkey" index="0"/>
                                    <filter type="data_meta" ref="input" key="dbkey" column="0" />
                                </options>
                            </param>
                        </when>
                        <when value="history">
                            <param type="data" format="gtf,gff,gff3" name="input"       label="GTF/GFF file" />
                            <param type="data" format="csv"          name="input_chain" label="LiftOver chain file" />
                        </when>
                </conditional>
                
                <param name="include_fails" type="boolean" truevalue="True" checked="false" falsevalue="False" label="Include failed liftovers" help="If a coordinate can not be lift over, do you want to include it in the output (it is still being marked 'fail')" />
            </when>
            <when value="vcf">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history_chain">Chain file from History</option>
                        <option value="history_all">Chain &amp; FASTA files from History</option>
                    </param>
                    
                    <when value="cached">
                        <param type="data" format="vcf" name="input" label="VCF file">
                            <validator type="unspecified_build" />
                            <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                        </param>
                        
                        <!-- automatically fetch a FASTA file from the same DBKEY as the chain file -->
                        <param name="input_chain" type="select" label="Lift Over To (Chain file)">
                            <options from_file="liftOver.loc">
                                <column name="name" index="1" />
                                <column name="value" index="1" /><!-- It is not possible to send the *.chain file as value, and obtain the 2nd dbkey as parameter via a filter -->
                                <column name="dbkey" index="0" />
                                <filter type="data_meta" ref="input" key="dbkey" column="0" />
                            </options>
                        </param>
                        
                        <param name="input_fasta" type="select" label="Lift Over To (FASTA file)" help="The FASTA file must be on the same build (dbkey) as the LiftOver chain file">
                            <options from_file="all_fasta.loc">
                                <column name="name" index="2"/>
                                <column name="value" index="3"/>
                                <column name="dbkey" index="1"/>
                                <filter type="param_value" ref="input_chain" column="1" />
                            </options>
                        </param>
                    </when>
                    
                    <when value="history_chain">
                        <param type="data" format="vcf" name="input" label="VCF file">
                            <validator type="unspecified_build" />
                            <validator type="dataset_metadata_in_file" filename="all_fasta.loc" metadata_name="dbkey" metadata_column="1" message="LiftOver mapping (FASTA file) is not available for the specified build." />
                        </param>
                        <param type="data" format="csv" name="input_chain" multiple="false" label="LiftOver chain file" />
                        <param name="input_fasta" type="select" label="Lift Over To (FASTA file)" help="The FASTA file must be on the same build (dbkey) as the LiftOver chain file">
                            <options from_file="all_fasta.loc">
                                <column name="name" index="2"/>
                                <column name="value" index="3"/>
                                <column name="dbkey" index="1"/>
                            </options>
                        </param>
                    </when>
                    
                    <when value="history_all">
                        <param type="data" format="vcf"   name="input" label="VCF file" />
                        <param type="data" format="csv"   name="input_chain" multiple="false" label="LiftOver chain file" />
                        <param type="data" format="fasta" name="input_fasta" multiple="false" label="Full genome FASTA file" />
                    </when>
                </conditional>
                
                <param name="include_fails" type="hidden" tvalue="False" />
            </when>
            <when value="wig">
                <conditional name="seq_source">
                    <param name="index_source" type="select" label="Source for LiftOver Data (chain file)">
                        <option value="cached">Local data (in galaxy)</option>
                        <option value="history">From History</option>
                    </param>
                    <when value="cached">
                        <param format="wig" name="input" type="data" label="Wiggle file">
                            <validator type="unspecified_build" />
                            <validator type="dataset_metadata_in_file" filename="liftOver.loc" metadata_name="dbkey" metadata_column="0" message="LiftOver mapping (chain file) is not available for the specified build." />
                        </param>
                        <param name="input_chain" type="select" label="Lift Over To">
                            <options from_file="liftOver.loc">
                                <column name="name" index="1"/>
                                <column name="value" index="2"/>
                                <column name="dbkey" index="0"/>
                                <filter type="data_meta" ref="input" key="dbkey" column="0" />
                            </options>
                        </param>
                    </when>
                    <when value="history">
                        <param format="wig" name="input" type="data" label="Wiggle file" />
                        <param type="data" format="csv" name="input_chain" label="LiftOver chain file" />
                    </when>
                </conditional>
                
                <param name="include_fails" type="hidden" tvalue="False" />
            </when>
        </conditional>
    </inputs>
    
    <outputs>
        <data format="text" name="output" label="${tool.name} on ${on_string}">
            <change_format>
                <when input="multiple.input_format" value="bam" format="bam" />
                <when input="multiple.input_format" value="sam" format="sam" />
                <when input="multiple.input_format" value="bed" format="bed" />
                <when input="multiple.input_format" value="bigwig" format="bigwig" />
                <when input="multiple.input_format" value="gff" format="gff" />
                <when input="multiple.input_format" value="vcf" format="vcf" />
                <when input="multiple.input_format" value="wig" format="bigwig" />
            </change_format>
        </data>
        
        <data format="text" name="output_unmapped" label="${tool.name} (unmapped) on ${on_string}">
            <filter>multiple['input_format'] in ["bam" , "vcf"]</filter>
            <change_format>
                <when input="multiple.input_format" value="bam" format="bam" />
                <when input="multiple.input_format" value="vcf" format="vcf" />
            </change_format>
        </data>
        <data format="text" name="output2" label="${tool.name} (bedgraph) on ${on_string}">
            <filter>multiple['input_format'] in ["wig"]</filter>
            <change_format>
                <when input="multiple.input_format" value="wig" format="bedgraph" />
            </change_format>
        </data>
    </outputs>
    
    <tests>
    <!-- BAM/SAM -->
        <test>
            <param name="input_format" value="bam"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bam_01_input_a.sam" ftype="bam"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_bam_01_output_a.sam" compare="diff" lines_diff="4"/>
            <output name="output_unmapped" file="test_bam_01_output_a.unmap.sam"/>
        </test>
    
        <test>
            <param name="input_format" value="sam"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bam_01_input_a.sam" ftype="sam"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_bam_01_output_a.sam" compare="diff" lines_diff="4"/>
            <output name="output_unmapped" file="test_bam_01_output_a.unmap.sam"/>
        </test>
    
    <!-- BED -->
        <test>
            <param name="input_format" value="bed"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bed_01_input_a.bed" ftype="bed"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_bed_01_output_a__only-matches.bed"/>
        </test>
        <test>
            <param name="input_format" value="bed"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bed_01_input_a.bed" ftype="bed"/>
            <param name="input_chain" value="aToB.over.chain" ftype="txt"/>
            <param name="include_fails" value="True"/>
            
            <output name="output" file="test_bed_01_output_a__all.bed"/>
        </test>
        <test>
            <param name="input_format" value="bed"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bed_02_input_a.bed" ftype="bed"/>
            <param name="input_chain" value="aToB.over.chain" ftype="txt"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_bed_02_output_a__only-matches.bed"/>
        </test>
        <test>
            <param name="input_format" value="bed"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bed_02_input_a.bed" ftype="bed"/>
            <param name="input_chain" value="aToB.over.chain" ftype="txt"/>
            <param name="include_fails" value="True"/>
            
            <output name="output" file="test_bed_02_output_a__all.bed"/>
        </test>
        
    <!-- BigWig --><!-- Malfuncioning in CrossMap 0.2, but patched via galaxy toolshed installer -->
        <test>
            <param name="input_format" value="bigwig"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_bigwig_01_input_a.bw" ftype="bigwig"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_bigwig_01_output_a.bw"/>
            <output name="output2" file="test_bigwig_01_output_a.sorted.bgr"/>
        </test>
    <!-- GFF -->
        <test>
            <param name="input_format" value="gff"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_gff_01_input_a.gtf" ftype="gtf"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_gff_01_output_a__only-matches.gtf"/>
        </test>
        <test>
            <param name="input_format" value="gff"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_gff_01_input_a.gtf" ftype="gtf"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="True"/>
            
            <output name="output" file="test_gff_01_output_a__all.gtf"/>
        </test>
        
    <!-- VCF -->
        <test>
            <param name="input_format" value="vcf"/>
            <param name="index_source" value="history_all"/>
            <param name="input" value="test_vcf_01_input.vcf" ftype="vcf"/>
            <param name="input_chain" value="test_vcf_01.over.chain" ftype="csv"/>
            <param name="input_fasta" value="test_vcf_01.fasta" ftype="fasta"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_vcf_01_output.vcf" compare="diff" lines_diff="4"/>
            <output name="output_unmapped" file="test_vcf_01_output.vcf.unmap"/>
        </test>
    </tests>

    <!-- WIG - Doesn't understand fixedStep -->
        <test>
            <param name="input_format" value="wig"/>
            <param name="index_source" value="history"/>
            <param name="input" value="test_wig_01_input_a.wig" ftype="wig"/>
            <param name="input_chain" value="aToB.over.chain" ftype="csv"/>
            <param name="include_fails" value="False"/>
            
            <output name="output" file="test_wig_01_output_a.bw"/>
            <output name="output2" file="test_wig_01_output_a.sorted.bgr"/>
        </test>
    <help>
CrossMap is versatile tool to convert genome coordinates or annotation files between genome
assemblies. It supports mostly commonly used file types, including BAM, BED,BigWig, GFF,
GTF, SAM, Wiggle, and VCF formats. For large plain text file types, such as BED, GFF, GTF
and VCF, reading from remote servers and file compression are supported.

CrossMap bed
------------
BED format file must have at least 3 columns (chrom, start, end) and no more than 12 columns.
BED format: http://genome.ucsc.edu/FAQ/FAQformat.html#format1
    </help>
    
    <citations>
        <citation type="doi">10.1093/bioinformatics/btt730</citation>
    </citations>
</tool>