#!/bin/bash
#SBATCH --job-name=snow_clean
#SBATCH --output=/gscratch/scrubbed/lhs3/snow_annotation/clean.out
#SBATCH --account=coenv
#SBATCH --partition=cpu-g2
#SBATCH --nodes=1
#SBATCH --cpus-per-task=16
#SBATCH --time=1-00:00:00
#SBATCH --mail-type=END,FAIL
#SBATCH --mail-user=laura.spencer@noaa.gov
#SBATCH --chdir=/gscratch/scrubbed/lhs3/snow_annotation

# Clean genome to remove contaminant sequences, which I identified per FCS-GX
# The public snow crab genome assembly GCA_016584305.1 was screened with FCS-GX prior to annotation.
# One 24,788-bp contig (JACEEZ010008220.1), classified as molluscan contamination, was excluded from the local annotation input FASTA.

module load coenv/samtools/1.17
module load stf/pigz/2.6

base=/gscratch/scrubbed/lhs3/snow_annotation
genome=${base}/GCA_016584305.1_ASM1658430v1_genomic.fna.gz
fasta=${base}/GCA_016584305.1_ASM1658430v1_genomic.fna
out=${base}/GCA_016584305.1_ASM1658430v1_genomic.clean.fna

# decompress genome
zcat ${genome} > ${fasta}

grep "^JAC" ${base}/gx/*.fcs_gx_report.txt | cut -f1 > ${base}/gx/contaminants.txt

samtools faidx ${fasta}
samtools faidx ${fasta} $(grep -v -f ${base}/gx/contaminants.txt ${fasta}.fai | cut -f1) > ${out}

pigz -p ${SLURM_CPUS_PER_TASK} ${out}