Build_kraken2_DB.xml

<pipeline name="Build_kraken2_db">
<description>
This pipeline downloads the kraken2 database with predefined libraries, filter it by length and taxonomic domains
</description>
<file id="libsfna" input="True" parameter="1" description="Output fna files under library" display_name="library fnas"/>
<dir id="outdir" input="False" parameter="2" default_output="True" description="Output directory" display_name="output dir"/>
<dir id="tax" filespec="taxonomy" in_dir="outdir"/>
<dir id="lib" filespec="library" in_dir="outdir"/>
<dir id="pfamdb" filespec="pfamlib" in_dir="outdir"/>
<file id="taxdb" in_dir="tax" filespec="taxonomy.sqlite"/>
<step name="taxdw">
  <tool name="taxdwt"
   input="outdir"
   output="tax"
   description="download_taxonomy.xml"/>
  <tool name="pfams"
   input="outdir"
   output="pfamdb"
   description="build_PFAM_db.xml"/>
  <tool name="taxondb"
   input="tax"
   output="taxdb"
   description="build_taxonomy_sqlite.xml"/>
</step>


<foreach id="libd" dir="lib">
  <file id="libfna" from_file="libsfna"/>
  <related id="sizefl" input="False" in_dir="lib" pattern="(.*.fna)" replace="\1.long"/>
  <related id="fltout" input="False" in_dir="lib" pattern="(.*.fna)" replace="\1"/>  
  <related id="domaintbl" input="False" in_dir="lib" pattern="(.*.fna)" replace="\1.out"/>
  <related id="proteins" input="False" in_dir="lib" temp="True" pattern="(.*.fna)" replace="\1.faa"/>
  <related id="domains" input="False" in_dir="lib" temp="True" pattern="(.*.fna)" replace="\1.hmms"/>
  <step name="downlib">
    <tool name="downl"
     input="outdir"
     output="libfna"
     description="download_library.xml"/>
    <tool name="sizeflt"
     input="libfna"
     output="sizefl"
     description="filter_len.xml"/>
    <tool name="domaindetect"
     input="sizefl,pfamdb"
     output="proteins,domains,domaintbl"
     description="find_taxonomy.xml"/>
    <tool name="fltdomain"
     input="sizefl,domaintbl,taxdb"
     output="fltout"
     description="filter_taxonomy.xml"/>
   </step>
</foreach>

<filelist id="fnas" in_dir="lib" foreach_id="libd" pattern="*/library.fna"/>
<step name="builddb">
  <tool name="buildit"
   input="outdir,fnas"
   description="build_kraken.xml"/>
</step>
</pipeline>