sh 下载Wormbase Genomes并为bwa，blast等设置#fasta

Posted 2021-05-07

function setup_genome
{
    mkdir -p ${1/.genomic.fa.gz}
    mv $1 ${1/.genomic.fa.gz}/${1}
    cd ${1/.genomic.fa.gz}
    # Construct blast db
    gunzip -kfc $1 | makeblastdb -in - -out $1 -title $1 -dbtype=nucl 
    mv $1 ${1/.fa.gz/.tmp}.fa.gz
    gunzip -kfc ${1/.fa.gz/.tmp}.fa.gz | bgzip > $1
    rm ${1/.fa.gz/.tmp}.fa.gz
    # Construct bwa index
    bwa index $1
    # Construct samtools fasta index
    samtools faidx $1
    cd ..
}

function setup_protein
{
    ref_dir=${1/.protein.fa.gz}
    mkdir -p ${ref_dir}
    gunzip -kfc ${ref_dir}/$1 | makeblastdb -in - -out ${ref_dir}/$1 -title ${ref_dir}/$1 -dbtype=prot
}

wget -rnd ftp://ftp.wormbase.org/pub/wormbase/species/ −−exclude−directories=protein,transcripts,annotation,assemblies,gff --accept "*WS248.genomic.fa.gz"

#
# Protein
# 
wget -rnd ftp://ftp.wormbase.org/pub/wormbase/species/ −−exclude−directories=genomic,transcripts,annotation,assemblies,gff --accept "*WS248.protein.fa.gz"


for i in `ls *.fa.gz`; do
    setup_genome ${i}
    setup_protein ${i}
done;