view PanExplorer.xml @ 11:ae74fc0cb39c draft

Uploaded
author dereeper
date Thu, 30 May 2024 16:41:08 +0000
parents d103c41b6931
children 38c66e401040
line wrap: on
line source

<tool id="PanExplorer" name="PanExplorer" version="1.0">
  <description> Bacterial pan-genome analysis </description>
  <requirements>
  <!--
 <requirement type="package" version="2.2.26">blast-legacy</requirement>
 <requirement type="package" version="7.480">mafft</requirement>
 <requirement type="package" version="14.137">mcl</requirement>
 <requirement type="package" version="3.697">phylip</requirement>
 <requirement type="package" version="1.7.2">perl-bioperl</requirement>
 -->
    <requirement type="package" version="1.30">perl-yaml</requirement>
    <requirement type="package" version="3.8.7">singularity</requirement>
 
 </requirements>

    <command><![CDATA[

export PANEX_PATH=${__tool_directory__};

perl ${__tool_directory__}/Perl/generateConfig.pl '$private_genomes' '$input' config.yaml '$private_genomes_fasta';

cat config.yaml >$logfile;

wget -O \$PANEX_PATH/panexplorer.sif https://panexplorer.southgreen.fr/singularity/panexplorer.sif >>$logfile 2>&1;




#if $software=="pgap"
sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_PGAP_heatmap_upset_COG >snakefile;
#else if $software=="roary"
sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_roary_heatmap_upset_COG >snakefile;
#else if $software=="orthofinder"
sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_orthofinder_heatmap_upset >snakefile;
#else if $software=="cactus"
cp -rf \$PANEX_PATH/Snakemake_files/Snakefile_wget_cactus_heatmap_upset_COG2 snakefile;
#else if $software=="pggb"
sed "s/identity\=30/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_pggb_heatmap_upset_COG >snakefile;
#else
sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_panacota_heatmap_upset_COG >snakefile;
#end if

singularity exec \$PANEX_PATH/panexplorer.sif snakemake --cores 1 -s snakefile >>$logfile 2>&1;


cp -rf outputs/upsetr.svg $upset;
cp -rf outputs/pav_matrix.tsv $output;
cp -rf outputs/heatmap.svg.gz $heatmap;
cp -rf outputs/heatmap.svg.heatmap_plotly.html $heatmap_html;
cp -rf outputs/rarefaction_curves.txt $rarefaction_curves;
cp -rf outputs/rarefaction_curves.svg $rarefaction_curves_svg;
cp -rf outputs/heaps.tsv $heaps;
cp -rf outputs/heatmap.svg.complete.pdf.distance_matrix.txt $distance_matrix;
cp -rf outputs/heatmap.svg.complete.pdf.distance_matrix.hclust.newick $njtree;

#if $software!="orthofinder"
cp -rf outputs/GCskew.txt $gcfile;
cp -rf outputs/cog_output.txt $cogfile;
cp -rf outputs/cog_stats.txt $outcog_stat;
cp -rf outputs/cog_stats2.txt $outcog_stat2;
cp -rf outputs/cog_of_clusters.txt $outcog_clusters;
cp -rf outputs/genomes/genes.txt $genes;
cp -rf outputs/fastani.out.matrix.complete $fastani;
cp -rf outputs/fastani.out.svg $ani_svg;
#end if

#if $software=="pggb"
cp -rf outputs/all_genomes.vcf $vcf;
#else
touch $vcf;
#end if
]]></command>


  <inputs>
  <param name="input" type="text" multiple="true" label="List of genbank identifiers" help="Coma separated list (ex: CP000235.1,CP001079.1,CP001759.1,CP015994.2)"/>
  <param name="private_genomes" type="data" format="zip" label="Zip of genbank or GFF files" optional="true"/>
  <param name="private_genomes_fasta" type="data" format="zip" label="Zip of Fasta files" optional="true"/>
  <param name="min_identity" type="text" value="80" label="Minimum percentage identity for BlastP" />
  <param type="select" name="software" label="Choose the pan-genome software">
    <option value="pgap">PGAP</option>
    <option value="roary">Roary</option>
    <option value="panacota">PanACoTA</option>
    <option value="orthofinder">OrthoFinder</option>
    <option value="cactus">Minigraph-Cactus</option>
    <option value="pggb">PanGenome Graph Builder (PGGB)</option>
  </param>
 </inputs>

 <outputs>
 <data format="txt" name="output" label="Pangenome presence absence matrix"/>
 <data format="newick" name="njtree" label="PanBased NJ tree"/>
 <data format="txt" name="genes" label="Genes"/>
 <data format="txt" name="cogfile" label="COG assignation"/>
 <data format="txt" name="gcfile" label="GC_percent"/>
 <data format="svg" name="upset" label="Upset Diagram"/>
 <data format="svg" name="heatmap" label="Presence Absence Heatmap"/>
 <data format="html" name="heatmap_html" label="Presence Absence Heatmap interactive"/>
 <data format="tabular" name="outcog_stat" label="COG category counts"/>
 <data format="tabular" name="outcog_stat2" label="COG category 2 counts"/>
 <data format="tabular" name="outcog_clusters" label="COG of clusters"/>
 <data format="tabular" name="fastani" label="ANI" />
 <data format="svg" name="ani_svg" label="ANI heatmap" />
 <data format="txt" name="rarefaction_curves" label="Rarefaction curves data"/>
 <data format="svg" name="rarefaction_curves_svg" label="Rarefaction curves"/>
 <data format="txt" name="heaps" label="Heaps law alpha"/>
 <data format="txt" name="distance_matrix" label="Accessory based distance matrix"/>
 <data format="vcf" name="vcf" label="VCF file"/>
 <data format="txt" name="logfile" label="Logfile"/>
 <data format="txt" name="roary_log" label="Roary Logfile"/>
</outputs>

</tool>