#!/bin/zsh

# general settings
ZOSMA_PROPER_LIST=list_of_zostera_marina_proper_scaffolds.txt
GENOME=/ngsprojects/zostera/data_archive/genome_assembly/Zostera_marina/final_selection/Zostera_marina.selectedgenome.fasta
INFO=v2.proper.20140501

# 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 -d y -l $ZOSMA_PROPER_LIST -t all
mv /xtras/group/biocomp/gdb/zostera/Zosma*LATEST.t??.gz .
for FILE in $(ls *LATEST.t??.gz)
do
	NEW_FILE=$(echo ${FILE:t:r:r}.$INFO.${FILE:t:r:e}.gz | sed 's/_LATEST//')
	mv $FILE $NEW_FILE
done

# genome sequences
echo "** RETRIEVING GENOME SEQUENCES **"
perl $svn/brhel/trunk/perl/scripts/FileIO_ext --equal -i $ZOSMA_PROPER_LIST $GENOME | gzip - > Zostera_marina.selectedgenome.proper.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 $ZOSMA_PROPER_LIST | gzip - > Zosma_PROT.$INFO.fasta.gz
perl $svn/brhel/trunk/perl/scripts/ORCAE_dumpfasta.pl -d orcae_zosma -t cds -c $ZOSMA_PROPER_LIST | gzip - > Zosma_CDS.$INFO.fasta.gz

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