#!/bin/zsh

# general settings
ZOSMA_PROPER_ORGANELLE_LIST=v2.1proper_organelle_list.txt
CONTAMINANTS=scaffolds_contaminants.txt
GENOME=/ngsprojects/zostera/data_archive/genome_assembly/Zostera_marina/final_selection2/Zostera_marina.selectedgenome2.fasta
INFO=v2.1proper.organellar.20140906
TMPDIR=/scratch/tmp/brhel/Zosma

# annotation files
echo "** RETRIEVING ANNOTATION FILES **"
perl /group/annot/sandbox/brhel/ORCAE/trunk/src/perl/programs/extract_Annot_files.pl -g Zosma -r current -c -l $ZOSMA_PROPER_ORGANELLE_LIST -type all -s "$INFO" -tempdir $TMPDIR -d y
mv -v /xtras/group/biocomp/gdb/zostera/Zosma*$INFO* .

# genome sequences
echo "** RETRIEVING GENOME SEQUENCES **"
perl $svn/brhel/trunk/perl/scripts/FileIO_ext --equal -i $ZOSMA_PROPER_ORGANELLE_LIST $GENOME | gzip - > Zostera_marina.selectedgenome.$INFO.fasta.gz

# protein and coding sequences
echo "** RETRIEVING PROT & CDS SEQUENCES **"
perl $svn/brhel/trunk/perl/scripts/ORCAE_dumpfasta.pl -d orcae_zosma -t prot -c $ZOSMA_PROPER_ORGANELLE_LIST | gzip - > Zosma_PROT.$INFO.fasta.gz
perl $svn/brhel/trunk/perl/scripts/ORCAE_dumpfasta.pl -d orcae_zosma -t cds -c $ZOSMA_PROPER_ORGANELLE_LIST | gzip - > Zosma_CDS.$INFO.fasta.gz

# annotation overview
echo "** GENERATING ANNOTATION OVERVIEW **"
perl $svn/brhel/trunk/perl/scripts/ORCAE_stats.pl -d orcae_zosma -include $ZOSMA_PROPER_ORGANELLE_LIST > annotation_info.txt