genomeannotatorncbi

<?xml version="1.0" encoding="UTF-8"?>
<!--
This Diya conf file can be used to take 454 assemblies and a *qual file as input
and output an annotated Genbank file and corresponding ASN.1 that can be submitted to
NCBI.

Example command:

nohup /usr/lib/perl5/site_perl/5.8.8/diya.pl
-conf genome-annotator-for-ncbi.xml \
-set MYUNIREFD=/Jake/data/mirror/uniprot/uniref50.fasta
-set MYSTRAIN=test0001 \
-set MYSPECIES='Test test'
-set MYRPSD=/Jake/data/mirror/ncbi/CDD/Cdd
-set MYSEQID=test0001 \
-set MYCLUSTERS=/Jake/data/mirror/ncbi/CDD/Clusters.bcp \
-set MYCDD=/Jake/data/mirror/ncbi/CDD/cddid_all.tbl
-set MYQUALFILE=454MiniContigs.qual \
-set MYTAXID=12345 454MiniContigs.fna > test0001.out &

1. copy all relevant files to a working directory
2. run glimmer3
3. run blastp of predicted peptides against UniRef50
4. run rpsblast of predicted peptides against CDD
5. run tRNAscanSE
6. run rnammer
7. run CRT
8. run phobos
9. run cmsearch
10. correct of annotation text
11. create ASN and associated files for NCBI submission

-bin /Jake/apps/bin -infile INPUTFILE -species MYSPECIES -strain MYSTRAIN -seqid MYSEQID -outdir OUTPUTDIR -newbler 200:5
assemble_pseudocontig make-single-fa MARC::glimmer3 extractCDS MARC::blastpCDS MARC::rpsblastCDS MARC::tRNAscanSE MARC::rnammer MARC::CRT MARC::phobos MARC::cmsearch gbconvert cleanup

assemble_pseudocontig make-single-fa MARC::glimmer3 extractCDS MARC::blastpCDS <<<< works
-->

<conf>
  <order>
   <names>assemble_pseudocontig make-single-fa MARC::glimmer3 extractCDS MARC::blastpCDS MARC::rpsblastCDS
   </names>
  </order>
    <script>
    <name>assemble_pseudocontig</name>
    <executable>diya-assemble_pseudocontig.pl</executable>
    <command>-infile INPUTFILE -scaff MYSCAFF -species MYSPECIES -strain MYSTRAIN -seqid MYSEQID -outdir OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputfrom></inputfrom>
    <inputformat></inputformat>
  </script>
  <script>
    <name>make-single-fa</name>
    <executable>make-single-fa.pl</executable>
    <command>MYSEQID.gbk OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputfrom></inputfrom>
    <inputformat></inputformat>
  </script>
  <parser>
    <name>MARC::glimmer3</name>
    <executable>g3-from-scratch2.csh</executable>
    <command>MYSEQID.fa MYSEQID OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputfrom></inputfrom>
    <inputformat></inputformat>
  </parser>
  <script>
    <name>extractCDS</name>
    <executable>extractCDS.pl</executable>
    <command>INPUTFILE OUTPUTFILE OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputfrom>MARC::glimmer3</inputfrom>
    <inputformat></inputformat>
  </script>
  <parser>
    <name>MARC::blastpCDS</name>
    <executable>blastall</executable>
    <command>-i INPUTFILE -p blastx -d MYUNIREFD -o OUTPUTFILE -e .001 -v 1 -b 1 -a 8</command>
    <home>/usr/bin</home>
    <inputformat></inputformat>
    <inputfrom>extractCDS</inputfrom>
  </parser>
  <parser>
    <name>MARC::rpsblastCDS</name>
    <executable>rpsblast</executable>
    <command>-d MYRPSD -i INPUTFILE -o OUTPUTFILE -e .001 -I T -p F -v 1 -b 1 -a 8</command>
    <home>/usr/bin</home>
    <inputformat></inputformat>
    <inputfrom>extractCDS</inputfrom>
  </parser>
  <parser>
    <name>MARC::tRNAscanSE</name>
    <executable>tRNAscan-SE</executable>
    <command>-B -o OUTPUTFILE MYSEQID.fa</command>
    <home>/home/jasperk/Apps/tRNAscan-SE-1.3.1</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <parser>
    <name>MARC::rnammer</name>
    <executable>rnammer</executable>
    <command>-S bac --gff OUTPUTFILE MYSEQID.fa</command>
    <home>/Jake/apps/rnammer-1.2</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <parser>
    <name>MARC::CRT</name>
    <executable>java</executable>
    <command>-cp /Jake/apps/CRT/CRT1.2-CLI.jar crt MYSEQID.fa OUTPUTFILE</command>
    <home>/usr/bin</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <script>
    <name>make-sRNAscanner-input</name>
    <executable>make-sRNAscanner-input.pl</executable>
    <command>MYSEQID OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputfrom></inputfrom>
    <inputformat></inputformat>
  </script>
  <parser>
    <name>MARC::sRNAscanner</name>
    <executable>sRNAscanner</executable>
    <command>OUTPUTDIR/Input.data</command>
    <home>/Jake/apps/sRNAscanner</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <parser>
    <name>MARC::cmsearch</name>
    <executable>cmsearch</executable>
    <command>--tabfile OUTPUTFILE -E 1.0e-20 -o MYSEQID.cmsearch /Jake/data/mirror/Rfam-10.0/sRNAs.cm MYSEQID.fa</command>
    <home>/Jake/apps/infernal-1.0.2/bin</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <parser>
    <name>MARC::phobos</name>
    <executable>phobos-linux-gcc4.1.2</executable>
    <command>-M exact -u 9 -U 1000 --outputFormat 2 MYSEQID.fa OUTPUTFILE</command>
    <home>/Jake/apps/bin</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </parser>
  <script>
    <name>gbconvert</name>
    <executable>gbconvert.pl</executable>
    <command>-agp -q MYQUALFILE -t MYTAXID INPUTFILE.gbk</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputformat></inputformat>
    <inputfrom>MARC::cmsearch</inputfrom>
  </script>
  <script>
    <name>cleanup</name>
    <executable>cleanup.sh</executable>
    <command>MYSEQID OUTPUTDIR</command>
    <home>/home/jasperk/Apps/DIYA-master/scripts</home>
    <inputformat></inputformat>
    <inputfrom></inputfrom>
  </script>
  <run>
    <mode>serial</mode>
  </run>
</conf>