Mercurial > repos > dereeper > pangenome_explorer

comparison PanExplorer.xml @ 2:97e4e3e818b6 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author dereeper
date Thu, 30 May 2024 11:48:09 +0000
parents
children 4157f435fa89
comparison
equal deleted inserted replaced
1:032f6b3806a3 2:97e4e3e818b6
1 <tool id="PanExplorer2" name="PanExplorer2" version="2.0">
2 <description> Bacterial pan-genome analysis </description>
3 <requirements>
4 <!--
5 <requirement type="package" version="2.2.26">blast-legacy</requirement>
6 <requirement type="package" version="7.480">mafft</requirement>
7 <requirement type="package" version="14.137">mcl</requirement>
8 <requirement type="package" version="3.697">phylip</requirement>
9 <requirement type="package" version="1.7.2">perl-bioperl</requirement>
10 -->
11 <requirement type="package" version="2.4.2">singularity</requirement>
12
13 </requirements>
14
15 <command><![CDATA[
16 module load system/singularity/3.6.0;
17 export PANEX_PATH=${__tool_directory__};
18
19 perl ${__tool_directory__}/Perl/generateConfig.pl '$private_genomes' '$input' config.yaml '$private_genomes_fasta';
20
21 cat config.yaml >$logfile;
22
23 if [ -f \$PANEX_PATH/panexplorer.sif ]; then; else; wget -O \$PANEX_PATH/panexplorer.sif https://panexplorer.southgreen.fr/singularity/panexplorer.sif; fi;
24
25
26
27 #if $software=="pgap"
28 sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_PGAP_heatmap_upset_COG >snakefile;
29 #else if $software=="roary"
30 sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_roary_heatmap_upset_COG >snakefile;
31 #else if $software=="orthofinder"
32 sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_orthofinder_heatmap_upset >snakefile;
33 #else if $software=="cactus"
34 cp -rf \$PANEX_PATH/Snakemake_files/Snakefile_wget_cactus_heatmap_upset_COG2 snakefile;
35 #else if $software=="pggb"
36 sed "s/identity\=30/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_pggb_heatmap_upset_COG >snakefile;
37 #else
38 sed "s/identity\=80/identity\=$min_identity/g" \$PANEX_PATH/Snakemake_files/Snakefile_wget_panacota_heatmap_upset_COG >snakefile;
39 #end if
40
41 singularity exec \$PANEX_PATH/panexplorer.sif snakemake --cores 1 -s snakefile >>$logfile 2>&1;
42
43
44 cp -rf outputs/upsetr.svg $upset;
45 cp -rf outputs/pav_matrix.tsv $output;
46 cp -rf outputs/heatmap.svg.gz $heatmap;
47 cp -rf outputs/heatmap.svg.heatmap_plotly.html $heatmap_html;
48 cp -rf outputs/rarefaction_curves.txt $rarefaction_curves;
49 cp -rf outputs/rarefaction_curves.svg $rarefaction_curves_svg;
50 cp -rf outputs/heaps.tsv $heaps;
51 cp -rf outputs/heatmap.svg.complete.pdf.distance_matrix.txt $distance_matrix;
52 cp -rf outputs/heatmap.svg.complete.pdf.distance_matrix.hclust.newick $njtree;
53
54 #if $software!="orthofinder"
55 cp -rf outputs/GCskew.txt $gcfile;
56 cp -rf outputs/cog_output.txt $cogfile;
57 cp -rf outputs/cog_stats.txt $outcog_stat;
58 cp -rf outputs/cog_stats2.txt $outcog_stat2;
59 cp -rf outputs/cog_of_clusters.txt $outcog_clusters;
60 cp -rf outputs/genomes/genes.txt $genes;
61 cp -rf outputs/fastani.out.matrix.complete $fastani;
62 cp -rf outputs/fastani.out.svg $ani_svg;
63 #end if
64
65 #if $software=="pggb"
66 cp -rf outputs/all_genomes.vcf $vcf;
67 #else
68 touch $vcf;
69 #end if
70 ]]></command>
71
72
73 <inputs>
74 <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)"/>
75 <param name="private_genomes" type="data" format="zip" label="Zip of genbank or GFF files" optional="true"/>
76 <param name="private_genomes_fasta" type="data" format="zip" label="Zip of Fasta files" optional="true"/>
77 <param name="min_identity" type="text" value="80" label="Minimum percentage identity for BlastP" />
78 <param type="select" name="software" label="Choose the pan-genome software">
79 <option value="pgap">PGAP</option>
80 <option value="roary">Roary</option>
81 <option value="panacota">PanACoTA</option>
82 <option value="orthofinder">OrthoFinder</option>
83 <option value="cactus">Minigraph-Cactus</option>
84 <option value="pggb">PanGenome Graph Builder (PGGB)</option>
85 </param>
86 </inputs>
87
88 <outputs>
89 <data format="txt" name="output" label="Pangenome presence absence matrix"/>
90 <data format="newick" name="njtree" label="PanBased NJ tree"/>
91 <data format="txt" name="genes" label="Genes"/>
92 <data format="txt" name="cogfile" label="COG assignation"/>
93 <data format="txt" name="gcfile" label="GC_percent"/>
94 <data format="svg" name="upset" label="Upset Diagram"/>
95 <data format="svg" name="heatmap" label="Presence Absence Heatmap"/>
96 <data format="html" name="heatmap_html" label="Presence Absence Heatmap interactive"/>
97 <data format="tabular" name="outcog_stat" label="COG category counts"/>
98 <data format="tabular" name="outcog_stat2" label="COG category 2 counts"/>
99 <data format="tabular" name="outcog_clusters" label="COG of clusters"/>
100 <data format="tabular" name="fastani" label="ANI" />
101 <data format="svg" name="ani_svg" label="ANI heatmap" />
102 <data format="txt" name="rarefaction_curves" label="Rarefaction curves data"/>
103 <data format="svg" name="rarefaction_curves_svg" label="Rarefaction curves"/>
104 <data format="txt" name="heaps" label="Heaps law alpha"/>
105 <data format="txt" name="distance_matrix" label="Accessory based distance matrix"/>
106 <data format="vcf" name="vcf" label="VCF file"/>
107 <data format="txt" name="logfile" label="Logfile"/>
108 <data format="txt" name="roary_log" label="Roary Logfile"/>
109 </outputs>
110
111 </tool>