Bioinformatics one-liners

ACTblastn.sh

#!/bin/sh
blastn -query 1 -subject 2 -outfmt 6 -task megablast -evalue 1e-10 | awk '{if ($4 > 4999) print $3" "$3" "$9" "$10" "$2" "$7" "$8" "$1}' > 1-2

complete_genomes.sh

#!/bin/sh
complete=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/ -l -s | grep -c "uid")
today=$(date +"%d/%m/%Y")
echo "\033[1;32m"$complete"\033[0m complete bacterial genomes on \033[1;34m"$today"\033[0m"

draft_genomes.sh

#!/bin/sh
draft=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/ -l -s | grep -c "uid")
today=$(date +"%d/%m/%Y")
echo "\033[1;32m"$draft"\033[0m draft bacterial genomes on \033[1;34m"$today"\033[0m"

getGenbank.sh

#!/bin/sh
if [ "$#" -lt 2 ]; then
    echo 'getGenbank GENUS SPECIES'
    exit 65
fi
genus=$1
species=$2
complete=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/ -l -s | grep $genus"_"$species)
draft=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/ -l -s | grep $genus"_"$species)
for genome in $complete
do
    echo "Downloading "$genome", complete"
    mkdir $genome
    files=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/$genome/ -l -s | grep '.gbk')
    for file in $files
    do
        echo $file
        curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/$genome/$file > $genome/$file
    done
    echo "-------------------------"
done

for genome in $draft
do
    echo "Downloading "$genome", draft"
    mkdir $genome
    files=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/$genome/ -l -s | grep '.gbk')
    for file in $files
    do
        echo $file
        curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/$genome/$file > $genome/$file
    done
    echo "-------------------------"
done

getNCBItaxonomy.sh

#!/bin/sh
echo "\033[1;32mDownloading NCBI taxonomy dump\033[0m"
curl ftp://ftp.ncbi.nih.gov/pub/taxonomy/taxdump.tar.gz > taxdump.tar.gz
echo "\033[1;32mUnpacking\033[0m"
gunzip -c taxdump.tar.gz | tar xf -
rm taxdump.tar.gz

specific_genomes.sh

#!/bin/sh
if [ "$#" -lt 2 ]; then
    echo 'specific_genomes GENUS SPECIES'
    exit 65
fi
genus=$1
species=$2
complete=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/ -l -s | grep -c $genus"_"$species)
draft=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/ -l -s | grep -c $genus"_"$species)
total=$(($complete + $draft))
today=$(date +"%d/%m/%Y")
echo "\033[1;32m"$total"\033[0m "$genus"_"$species" genomes on \033[1;34m"$today"\033[0m"
echo "\033[1;32m"$complete"\033[0m complete "$genus"_"$species" genomes on \033[1;34m"$today"\033[0m"
echo "\033[1;32m"$draft"\033[0m draft "$genus"_"$species" genomes on \033[1;34m"$today"\033[0m"

total_genomes.sh

#!/bin/sh
complete=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria/ -l -s | grep -c "uid")
draft=$(curl ftp://ftp.ncbi.nih.gov/genomes/Bacteria_DRAFT/ -l -s | grep -c "uid")
total=$(($complete + $draft))
today=$(date +"%d/%m/%Y")
echo "\033[1;32m"$total"\033[0m bacterial genomes on \033[1;34m"$today"\033[0m"