#!/bin/bash
#set -e
type="proteins"
script_path=$(dirname `realpath $0`)
database_dir="/Business/psn_company/t10/DATA/EGGNOG_DB/EGGNOG_SpDB"
declare -A software db_path
software["CDS"]="diamond blastx"
software["proteins"]="diamond blastp"
db_path["animal"]=$database_dir/Metazoa
db_path["plant"]=$database_dir/Viridiplantae
db_path["fungi"]=$database_dir/Fungi
db_path["prokaryotes"]=$database_dir/Bacteria
usage(){
    echo -e "Usage:"
    echo -e "\t$0 -i cds.fa -o outdir -t CDS -s animal"
    echo -e "OPTIONS:"
	echo -e "\t-h: help information"
    echo -e "\t-i: fasta file"
    echo -e "\t-o: output dir"
    echo -e "\t-t: proteins or CDS default:proteins"
    echo -e "\t-s: species [ animal plant fungi prokaryotes ]"
    exit 1
}

while getopts 'hs:t:i:o:' OPT;do
    case $OPT in
        h) usage;;
        s) species="$OPTARG";;
		t) type="$OPTARG";;
        i) fasta="$OPTARG";;
        o) outdir="$OPTARG";;
        ?) usage;;
    esac
done

if [[ ! $species ]] || [[ ! $fasta ]] || [[ ! $outdir ]] || [ $# == 0 ]
then
	usage
fi

if [[ ! ${db_path[${species}]} ]] || [[ ! ${software[${type}]} ]]
then
    echo "ERROR:unknown -t $type  or -s $species" 
    exit 1
fi
[ -d $outdir/eggnogmap_tmp ] && rm -rf $outdir/eggnogmap_tmp
mkdir -p $outdir/eggnogmap_tmp
source /Business/psn_company/t10/software/miniconda3/bin/activate eggnog-mapper
/Business/psn_company/t10/software/miniconda3/envs/eggnog-mapper/bin/emapper.py -i ${fasta} --output Annotation --output_dir $outdir/eggnogmap_tmp  -m diamond  --data_dir  ${db_path[${species}]} --itype ${type} --cpu 16 --usemem --sensmode more-sensitive --dbmem
less -S $outdir/eggnogmap_tmp/Annotation.emapper.annotations |cut -f1,5,7 |sed /^#/d | grep -vP "\t-\t-"|awk 'BEGIN{FS=OFS="\t"}{l=split($2,aa,",");print $1"\t"aa[l]"\t"$3}' | sed 's/@.*\t/\t/'|awk '{if(!a[$1]){print;a[$1]=1}}' > $outdir/eggnogmap_tmp/eggnog.txt
cut -f1,2 $outdir/eggnogmap_tmp/eggnog.txt > $outdir/eggNOG.txt 
cut -f1,3 $outdir/eggnogmap_tmp/eggnog.txt > $outdir/eggNOG_Category.txt
less -S $outdir/eggnogmap_tmp/Annotation.emapper.annotations |sed /^#/d | cut -f1,10 | grep GO | perl -ne 'chomp;@a=split/\t/;@b=$a[1]=~/(GO:\d{7})/g;foreach(@b){$c{$a[0]}{$_}=1};print $a[0]."\t".join(";",sort keys %{$c{$a[0]}})."\n"' |awk -F "\t" '{if(!a[$1]){print;a[$1]=1}}' > $outdir/GO.txt