comparison phylogenomics_analysis.xml @ 41:6b547036c627 draft

Uploaded

40:0bf214a0646a

        	#set scaffold_path = $scaffold.fields.path
        	#set scaffold_dir = $os.path.split($scaffold_path)[0]
        	#set scaffold_selection = $os.path.split($scaffold_path)[1]
            #set src_fasta_dir = $os.path.join('phylogenomicsAnalysis_dir', 'orthogroups_fasta')
            #set dest_fasta_dir = $output_fasta.extra_files_path

            #if str($options_type.options_type_selector) == 'advanced' and str($options_type.multiple_sequence_alignments_cond.multiple_sequence_alignments) == 'yes':
                #set generate_output_aln = True
                #set src_aln_dir = $os.path.join('phylogenomicsAnalysis_dir', 'orthogroups_aln')
                #set dest_aln_dir = $output_aln.extra_files_path

                    #if str($input_includes_coding_sequences) == 'yes':
                        #if str($multiple_codon_alignments) == 'yes':
                            --codon_alignments
                            #if str($use_corresponding_coding_sequences) == 'yes':
                            	#if str($input.ext) == 'pgfccs):
                                	--orthogroup_fna
                              	#else:
                              		raise Exception("Selected input does not include corresponding coding sequences")
                              	#end if
                            #end if

                        #end if
                    #end if
                #end if
            #end if

            && echo "Sequence alignments and corresponding phylogenies: `ls $src_fasta_dir | grep f | wc -l` files" > $output_fasta
            && ls -al $src_fasta_dir | grep f >> $output_fasta
            && mv $src_fasta_dir/* $dest_fasta_dir || true

            #if $generate_output_aln:
                && echo "Multiple sequence alignments using method '$multiple_sequence_alignments_option' (note that some files may be empty): `ls $src_aln_dir | grep f | wc -l` files" > $output_aln
                && ls -al $src_aln_dir | grep f >> $output_aln

                && mv $src_tree_dir/* $dest_tree_dir || true
            #end if
        ]]>
    </command>
    <inputs>
        <param name="input" format="pgfc,pgfccs" type="data" label="Select gene family clusters" help="Selection may optionally include corresponding coding sequences"/>
        <param name="scaffold" type="select" label="Orthogroups or gene families proteins scaffold">
            <options from_data_table="plant_tribes_scaffolds" />
            <validator type="no_options" message="No PlantTribes scaffolds are available.  Use the PlantTribes Scaffolds Download Data Manager tool to install and populate the PlantTribes scaffolds data table."/>
        </param>
        <param name="method" type="select" label="Protein clustering method">

                </conditional>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_fasta" format="txt" label="Sequence alignments and corresponding phylogenies on ${on_string}"/>
        <data name="output_aln" format="txt" label="Multiple alignments on ${on_string}">
            <filter>options_type['options_type_selector'] == 'advanced' and options_type['multiple_sequence_alignments_cond']['multiple_sequence_alignments'] == 'yes'</filter>
        </data>
        <data name="output_tree" format="txt" label="Phylogenetic trees on ${on_string}">
            <filter>options_type['options_type_selector'] == 'advanced' and options_type['phylogenetic_trees_cond']['phylogenetic_trees'] == 'yes'</filter>
        </data>
    </outputs>
    <tests>
        <test>

 - **Protein clustering method** - One of GFam (domain architecture based clustering), OrthoFinder (broadly defined clusters) or OrthoMCL (narrowly defined clusters).

* **Multiple sequence alignments options**

 - **Select method for multiple sequence alignments** - Method used for setting multiple sequence alignments.
 - **Input sequences include corresponding coding sequences?** - Selecting 'Yes' for this option requires that the selected input data format is 'pgfccs'.
 - **Construct orthogroup multiple codon alignments?** - Construct orthogroup multiple codon alignments.
 - **Sequence type used in the phylogenetic inference** - Sequence type (dna or amino acid) used in the phylogenetic inference.
 - **Use corresponding coding sequences?** - Selecting 'Yes' for this option requires that the selected input data format is 'pgfccs' or this tool will produce an error.

* **Phylogenetic trees options**

 - **Phylogenetic trees inference method** - Phylogenetic trees inference method.
 - **Select history item containing strings matching sequence identifiers of species for determining the most basal taxa in the orthogroups for rooting trees?** - If 'No' is selected, trees will be rooted using the most distant taxon present in the orthogroup.

41:6b547036c627

        	#set scaffold_path = $scaffold.fields.path
        	#set scaffold_dir = $os.path.split($scaffold_path)[0]
        	#set scaffold_selection = $os.path.split($scaffold_path)[1]
            #set src_fasta_dir = $os.path.join('phylogenomicsAnalysis_dir', 'orthogroups_fasta')
            #if str($input.ext) == 'ptcgf':
            	#set dest_fasta_dir = $output_fasta_ptcgf.extra_files_path
            else:
            	#set dest_fasta_dir =$output_fasta_ptcgfcs.extra_files_path
           	#end if

            #if str($options_type.options_type_selector) == 'advanced' and str($options_type.multiple_sequence_alignments_cond.multiple_sequence_alignments) == 'yes':
                #set generate_output_aln = True
                #set src_aln_dir = $os.path.join('phylogenomicsAnalysis_dir', 'orthogroups_aln')
                #set dest_aln_dir = $output_aln.extra_files_path

                    #if str($input_includes_coding_sequences) == 'yes':
                        #if str($multiple_codon_alignments) == 'yes':
                            --codon_alignments
                            #if str($use_corresponding_coding_sequences) == 'yes':
                            	#if str($input.ext) == 'ptcgfcs):
                                	--orthogroup_fna
                              	#else:
                              		raise Exception("Selected input does not include corresponding coding sequences")
                              	#end if
                            #end if

                        #end if
                    #end if
                #end if
            #end if

			#if str($input.ext) == 'ptcgf':
            	&& echo "Sequence alignments and corresponding phylogenies: `ls $src_fasta_dir | grep f | wc -l` files" > $output_fasta_ptcgf
            	&& ls -al $src_fasta_dir | grep f >> $output_fasta_ptcgf
           	#else:
            	&& echo "Sequence alignments and corresponding phylogenies: `ls $src_fasta_dir | grep f | wc -l` files" > $output_fasta_ptcgfcs
            	&& ls -al $src_fasta_dir | grep f >> $output_fasta_ptcgfcs
           	#end if
            && mv $src_fasta_dir/* $dest_fasta_dir || true

            #if $generate_output_aln:
                && echo "Multiple sequence alignments using method '$multiple_sequence_alignments_option' (note that some files may be empty): `ls $src_aln_dir | grep f | wc -l` files" > $output_aln
                && ls -al $src_aln_dir | grep f >> $output_aln

                && mv $src_tree_dir/* $dest_tree_dir || true
            #end if
        ]]>
    </command>
    <inputs>
        <param name="input" format="ptcgf,ptcgfcs" type="data" label="Select gene family clusters" help="Selection may optionally include corresponding coding sequences"/>
        <param name="scaffold" type="select" label="Orthogroups or gene families proteins scaffold">
            <options from_data_table="plant_tribes_scaffolds" />
            <validator type="no_options" message="No PlantTribes scaffolds are available.  Use the PlantTribes Scaffolds Download Data Manager tool to install and populate the PlantTribes scaffolds data table."/>
        </param>
        <param name="method" type="select" label="Protein clustering method">

                </conditional>
            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="output_fasta_ptcgf" format="ptcgf" label="Sequence alignments and corresponding phylogenies on ${on_string}">
        	<filter>input.ext == 'ptcgf'</filter>
        </data>
        <data name="output_fasta_ptcgfcs" format="ptcgfcs" label="Sequence alignments and corresponding phylogenies on ${on_string}">
        	<filter>input.ext == 'ptcgfcs'</filter>
        </data>
        <data name="output_aln" format="txt" label="Multiple alignments on ${on_string}">
            <filter>options_type['options_type_selector'] == 'advanced' and options_type['multiple_sequence_alignments_cond']['multiple_sequence_alignments'] == 'yes'</filter>
        </data>
        <data name="output_tree" format="ptpt" label="Phylogenetic trees on ${on_string}">
            <filter>options_type['options_type_selector'] == 'advanced' and options_type['phylogenetic_trees_cond']['phylogenetic_trees'] == 'yes'</filter>
        </data>
    </outputs>
    <tests>
        <test>

 - **Protein clustering method** - One of GFam (domain architecture based clustering), OrthoFinder (broadly defined clusters) or OrthoMCL (narrowly defined clusters).

* **Multiple sequence alignments options**

 - **Select method for multiple sequence alignments** - Method used for setting multiple sequence alignments.
 - **Input sequences include corresponding coding sequences?** - Selecting 'Yes' for this option requires that the selected input data format is 'ptcgfcs'.
 - **Construct orthogroup multiple codon alignments?** - Construct orthogroup multiple codon alignments.
 - **Sequence type used in the phylogenetic inference** - Sequence type (dna or amino acid) used in the phylogenetic inference.
 - **Use corresponding coding sequences?** - Selecting 'Yes' for this option requires that the selected input data format is 'ptcgfcs' or this tool will produce an error.

* **Phylogenetic trees options**

 - **Phylogenetic trees inference method** - Phylogenetic trees inference method.
 - **Select history item containing strings matching sequence identifiers of species for determining the most basal taxa in the orthogroups for rooting trees?** - If 'No' is selected, trees will be rooted using the most distant taxon present in the orthogroup.